WEBVTT

1
00:00:00.001 --> 00:00:10.820
robertmthorson: Well it's at 803 on a sunny morning after daylight savings time, so I think i'll just get this started it's 803 going and NATO, for the first talk begins at 805.

2
00:00:11.750 --> 00:00:19.580
robertmthorson: i've lost the welcome screen that GSA put up that I had up there about two minutes ago I don't dare to find it for fear that I might lose it.

3
00:00:20.510 --> 00:00:33.500
robertmthorson: i'm Robert thorson and i'm a professor geosciences at the University of Connecticut and I am coordinating this session i've been involved in stonewall thinking mapping and working since I moved here in 1984.

4
00:00:34.040 --> 00:00:44.210
robertmthorson: Will we met will coordinate with me and he is temporarily offline because he is trying to you know get someone else online, who is actually a presenter.

5
00:00:45.530 --> 00:00:49.580
robertmthorson: So i'm looking forward to a nice session today and.

6
00:00:50.600 --> 00:01:01.400
robertmthorson: it's it's nice to see the fairly large audience that we have out there i'm not as tech savvy as most so i'll be lucky to get through my first two talks.

7
00:01:02.030 --> 00:01:15.260
robertmthorson: And and and I have something to share with you, so the basic list of docs we're going to have beginning at 805 is a first talk on integrating aerial mapping and field investigations that i'm involved in.

8
00:01:16.400 --> 00:01:22.910
robertmthorson: Men and Dr Richard is the was the lead author and participant in the field but i'm giving the talk on his behalf.

9
00:01:23.780 --> 00:01:42.590
robertmthorson: And then onto my endless saga of classifying stone walls at 825 I have a talk about how my process and procedures are going there and then then we're going to have she one is the one will is trying to help get on and that's an automated stonewall mapping and then the.

10
00:01:44.030 --> 00:01:54.620
robertmthorson: Last two talks will be one by will we met on lidar mapping and then finally one that's maybe a bit of a technical challenge we're going to go to a poster session for john Leonard at the end.

11
00:01:55.340 --> 00:02:03.200
robertmthorson: So those are the the talks that we have lined up for this morning I want to thank you for for joining, let us if you have questions put them in the chat.

12
00:02:04.010 --> 00:02:19.160
robertmthorson: And i'll be able to get them, especially after i'm not presenting my own talks Okay, so now what i'm going to do is head to the first talk, which is one that i'm going to be giving and there it is right there share that screen.

13
00:02:21.170 --> 00:02:21.710
robertmthorson: and

14
00:02:23.510 --> 00:02:31.370
robertmthorson: i'm going to play from start and see if it works Can somebody confirm that they see that talk nicely okay good.

15
00:02:32.000 --> 00:02:43.460
robertmthorson: Okay we're often running now, the first of many talks this morning titled integrating aerial mapping and field investigations, the one thing I haven't done is set my timer so will, will you give me a.

16
00:02:44.630 --> 00:02:46.940
robertmthorson: Three minute warning, please okay.

17
00:02:48.020 --> 00:03:03.710
robertmthorson: Integrating field investigations, a case study in three forested locales in northeastern Connecticut this project, supported by the Department geosciences and on partly part of wills nsf grant on on the.

18
00:03:05.300 --> 00:03:10.760
robertmthorson: The whole phenomenon here of anthropocene in New England OK, the next slide.

19
00:03:12.710 --> 00:03:23.630
robertmthorson: Okay line, our main thing is an exercise in remote sensing I think we all know that, even if they are perfectly mapped and feel jack that is every single stone wall out there is perfect.

20
00:03:25.070 --> 00:03:35.840
robertmthorson: walls are defined principally as one thing they're an image map line and those that are field check turn out to be stone walls, others are various lines that are seeing.

21
00:03:36.260 --> 00:03:41.180
robertmthorson: But the idea is that quantification and interpretation based on all these marvelous lidar studies.

22
00:03:41.480 --> 00:03:56.270
robertmthorson: is really about counts and links and presence absence here, there and whether they're dense or not, so the spatial patterning of stone walls is really, really important, perhaps the most important but it's still basically one thing walls on a map.

23
00:03:58.760 --> 00:04:13.550
robertmthorson: But there are many kinds of walls, you know, as with many kinds of trees, one would not walk out with a forester and say that all trees are the same, but that's pretty much the way we're treating stone walls, right now, at least in a.

24
00:04:14.570 --> 00:04:15.620
robertmthorson: systematic way.

25
00:04:17.000 --> 00:04:29.990
robertmthorson: For example, there's objective categories, based on mythology there's a dolomite wall on the left and our coast in the middle and a nice wall on the right now the spatial patterning of these mythologies over the landscape should matter.

26
00:04:30.560 --> 00:04:38.840
robertmthorson: All right, but right now, they wouldn't be currently captured in lidar investigations or, for example, subjective categories, based on aesthetics.

27
00:04:39.230 --> 00:04:51.500
robertmthorson: And I don't know about you, but this is my candidate for the ugliest stonewall i've seen in New England, yet, and yet that would be mapped as a line you know in lidar along with some other the most fantastic and beautiful walls.

28
00:04:52.670 --> 00:05:02.390
robertmthorson: Or, for example, indirect categories, you know, based on chronology we would map these walls these along the famous battle road in in concord and lexington.

29
00:05:03.080 --> 00:05:14.090
robertmthorson: Massachusetts and, and these are two 20th century walls built to to mark those along the battle road in the 18th century on the famous shot that heard round the world day.

30
00:05:14.930 --> 00:05:26.120
robertmthorson: So so basically it's kind of hard to map chronology and mythology and aesthetics from the air, so what i've got is an in progress taxonomy that i've been working on for.

31
00:05:27.050 --> 00:05:44.090
robertmthorson: Since seriously since about 2005 and you'll hear more about that in the next talk, but basically in order to truly define a wall here in green on the second column in from the right, in order to define one as a wall, you have to carve it out of something it has to be an out group.

32
00:05:45.770 --> 00:05:53.810
robertmthorson: Out of something else, and so, so I had to invent something called the stone domain, which is the entire residue of stone all over the landscape.

33
00:05:54.590 --> 00:06:07.100
robertmthorson: That is one of four domains alright so there's a class called wall it's a it's one of one of four basic classes in the stone domain, the others being lines concentrations and notable stones.

34
00:06:08.180 --> 00:06:22.520
robertmthorson: And then within the class wall, there are five basic types freestanding which are the you know 90% of what we're mapping and lidar, but there are others flanking or supporting blocking and, in closing, and even though these are, are you know.

35
00:06:23.960 --> 00:06:34.340
robertmthorson: functional titles, you know the objective criteria are used to define them and then subtypes we have banned single double and so forth it's like any taxonomy it's a nested.

36
00:06:34.970 --> 00:06:39.950
robertmthorson: A nested set of things so here's just some examples and in closing wall is a town pound.

37
00:06:40.250 --> 00:06:48.350
robertmthorson: And it has criteria i'm not going to have the time to explain all of the criteria used to define this a blocking wall is something that's usually used to control water.

38
00:06:49.130 --> 00:06:59.150
robertmthorson: And it's usually one sided and perpendicular a supporting wall hold something up its defined by the level top criteria and other things, and at least a late degree of construction.

39
00:06:59.480 --> 00:07:07.700
robertmthorson: A flanking wall is any retaining wall that one's in Poland springs that one's a one one tiered three course retaining wall.

40
00:07:08.150 --> 00:07:20.570
robertmthorson: Which is a category flanking and then a freestanding wall right down here, and then the various kinds of freestanding wall we're familiar with, or many of us are bands double broad single in a budding walls.

41
00:07:20.960 --> 00:07:28.370
robertmthorson: OK so moving rapidly along this pilot study started to try and integrate the lidar mapping and the field classification.

42
00:07:28.940 --> 00:07:31.850
robertmthorson: And it's essentially a classification based on structure.

43
00:07:32.240 --> 00:07:47.750
robertmthorson: And it can be interpreted to indicate historic land use scientifically and quantitatively that's the goal here not to give things label, but to be rigorously scientific about it, and it does seem to be working, but it does have its issues.

44
00:07:48.410 --> 00:07:58.700
robertmthorson: We chose three forest areas and we mapped 273 segments that were mapped in their entire length, you know from the field using data entry techniques that i'll get to later.

45
00:08:00.170 --> 00:08:07.190
robertmthorson: Okay, the three project areas here's stores for reference and the nearest city to us, as will a manic down there in northeastern Connecticut.

46
00:08:07.850 --> 00:08:16.790
robertmthorson: area, one is a hardscrabble mid slope in yukon forest very near here area to is wormwood Hill in stores it's an upland.

47
00:08:17.450 --> 00:08:29.240
robertmthorson: If you like sick till landscape, you know which is better for agriculture area three is a mill village completely abandoned in ashford and our goal was to try and create a taxonomy that could incorporate.

48
00:08:29.780 --> 00:08:39.230
robertmthorson: You know this great variety of the phenomenon in different settings and our methods were pretty straightforward we start with the lidar mapping then to on the right here.

49
00:08:39.620 --> 00:08:47.330
robertmthorson: We have a data form that had about nine iterations before we were comfortable with it, you know how to describe and measure stone walls in the field.

50
00:08:48.170 --> 00:08:54.770
robertmthorson: And then the third thing we did was build a database, out of all that data we're trying to automate it but that's a little bit down the road.

51
00:08:55.430 --> 00:09:02.840
robertmthorson: And then, for we would take a photo log that attaches to these descriptions and then five you know do our best that interpretation.

52
00:09:03.830 --> 00:09:10.490
robertmthorson: So here's a quick look at the blow up of the field data form the segment in the upper left is the key item it's not the wall.

53
00:09:11.180 --> 00:09:26.750
robertmthorson: The tax on is the name of it that's a check the box thing contacts, you know when you're inside of a wall, you have different segments that contact each other, a budding salvage gradation Ben gap, you know there's just all kinds of things terminations dimensions and presentations.

54
00:09:28.100 --> 00:09:34.490
robertmthorson: could give a whole talk about this form but clearly we're trying to capture stone order and size and shape.

55
00:09:35.420 --> 00:09:43.610
robertmthorson: and so forth Okay, and then that goes into a database it's pretty familiar basically the the date the person the parcel the wall, the ID the segment.

56
00:09:43.820 --> 00:09:48.530
robertmthorson: You know the work direction why I presented at the tax on the hybrid and then down here.

57
00:09:48.770 --> 00:09:57.230
robertmthorson: All these other things about links and terminations and contracts it's all pretty straightforward at this point we've gone through it a few times, and then the photo log then would be.

58
00:09:57.830 --> 00:10:05.150
robertmthorson: The next thing we do and here's just a few you know there's a single wall on the Left there's a double wall on the right and boy was it shaded.

59
00:10:05.630 --> 00:10:11.960
robertmthorson: We have no choice but to work right in the middle of sort of June, July, and it was really, really shaded but still at work pretty well.

60
00:10:12.680 --> 00:10:20.900
robertmthorson: And there is a broad wall that is basically becoming a hybrid wall because somebody is building a new wall off to the left of it.

61
00:10:21.680 --> 00:10:30.650
robertmthorson: And then on the right, we just have a standard broad wall and abroad wall is anything wider than it than it needs to be to have that double structure.

62
00:10:31.490 --> 00:10:41.870
robertmthorson: Okay, and then the final thing we did that Richard pretty much did is we, and this is a first draft for us is develops a map symbolism, so that we're not just mapping lines on a.

63
00:10:42.200 --> 00:10:56.840
robertmthorson: On a on a lidar we're actually mapping classification taxa on a lidar because then it's the pattern of those taxa that we can use to interpret landscape change and we have some some examples of that to come.

64
00:10:57.710 --> 00:11:07.190
robertmthorson: So basically you don't need to worry about which is which, but we have about eight or nine different tax, so that we can map, most of which are going to be single double broad walls and dance.

65
00:11:09.200 --> 00:11:18.650
robertmthorson: Okay, so area one yukon forest, this is pretty standard, this is what it looks like this is a very nearby road mostly back and for us in our areas down here in the middle.

66
00:11:19.520 --> 00:11:31.670
robertmthorson: And there it is on the lidar 2016 lighter will we met provided all the light our images here and the lidar mapping, so this is the basic hillside map we've seen plenty like this.

67
00:11:33.470 --> 00:11:42.860
robertmthorson: And then, this is we add the walls and it looks like this there's a whole bunch of them, of course, and you notice that there are some liniers there that are not walls and some.

68
00:11:43.700 --> 00:11:51.680
robertmthorson: And so there is some subjective classification that goes in here, and these are manually added, I know there were trying to work out an automated way of doing this.

69
00:11:52.610 --> 00:11:56.510
robertmthorson: And then, this is the classification mapped on top of.

70
00:11:57.290 --> 00:12:04.760
robertmthorson: You know the the lidar map So these are two basic it's mapped with lidar you know but it's classified with these various pattern.

71
00:12:05.090 --> 00:12:14.600
robertmthorson: lines that you see there and I know you can't really see much there so we're going to basically peel away the lidar or peel away the red lines to see only the classification, right here.

72
00:12:15.350 --> 00:12:26.120
robertmthorson: Okay, and what you see is the pattern for nearby yukon forest and what you see is the yellow jumps right out at you, and what that means is that single walls are more common.

73
00:12:26.660 --> 00:12:35.630
robertmthorson: Another thing is, if you look at where the broad walls which are the the pinkish broad lines they tend to be on the downhill sides on the downhill sides.

74
00:12:36.530 --> 00:12:44.630
robertmthorson: Of parcels, and that makes sense, because these are the disposal walls, you know the ones with with way more stone and then, then you need to be to build a double wall.

75
00:12:45.290 --> 00:12:52.100
robertmthorson: We can see that the double walls which are in red are localized in just a few places where there might be explanations for that.

76
00:12:52.670 --> 00:13:08.150
robertmthorson: And finally, we only had one hybrid wall, and it was right up here on the left a short segment a hybrid basically what's happening here is a boulder wall is being built next to an old historic wall because it's right next to a past year, so we're able to capture that that that.

77
00:13:09.320 --> 00:13:11.930
robertmthorson: sort of 19th century wall with the 20th century one.

78
00:13:13.370 --> 00:13:18.050
robertmthorson: And then wormwood Hill, the same thing for the next two areas are similar here's wormwood hill.

79
00:13:19.250 --> 00:13:23.600
robertmthorson: This is a same similar pattern, but much more regular which you might expect.

80
00:13:23.930 --> 00:13:36.350
robertmthorson: Now this is where lidar is better, you know light, are you can you can do a better job on the geometry, you know from the area, then you can on the field so it's the combination of the lidar mapping and the field classification and description.

81
00:13:36.620 --> 00:13:42.140
robertmthorson: That are really going to reveal you know changes over the landscape, as we move this into a science.

82
00:13:42.830 --> 00:13:50.840
robertmthorson: So here in wormwood Hill, is it a script is a mapping of the simple walls much more Rector linear and here is our mapping, on top of those.

83
00:13:51.530 --> 00:13:57.890
robertmthorson: And so i'll take it one more step and we have only the field mapping left and you see a dramatic pattern here.

84
00:13:58.790 --> 00:14:02.990
robertmthorson: And this makes sense, this is still this was still being farmed in the 1930s.

85
00:14:03.470 --> 00:14:11.960
robertmthorson: You know, so what we have is on two separate drum lines, one of them has got all kinds of double and and broad walls and the other one has none of them.

86
00:14:12.320 --> 00:14:18.560
robertmthorson: In the House in the road or over here on the left, where you can see those double walls reading right up to the road.

87
00:14:19.010 --> 00:14:32.000
robertmthorson: So the access point is on the left with the beautiful double wall is coming in, and then a variety of walls near that and then very few in the more remote or pioneering phase aspect of the walls to the right.

88
00:14:32.360 --> 00:14:41.450
robertmthorson: This is on wormwood Hill and so it's really clear that the pattern of walls can be mapped after field verification and that can start telling us landscape stories.

89
00:14:42.170 --> 00:14:50.930
robertmthorson: This also has a lot of bands and band, is the most primitive degree of wall it's just basically a pile of rubble or stone, with no with no.

90
00:14:52.670 --> 00:15:01.970
robertmthorson: You no structure whatsoever so it's a an unbuilt wall, but it does classify as a wall Okay, and then area three an asteroid similar.

91
00:15:02.780 --> 00:15:15.560
robertmthorson: In tone in forest completely for us it but it's very different in pattern, we have a sort of a farm scene up in the upper left here, but in the lower right is the village, we were trying to capture, so the walls, we mapped are here.

92
00:15:16.580 --> 00:15:19.400
robertmthorson: Based from lidar, but when we went in the field, we found.

93
00:15:19.940 --> 00:15:27.410
robertmthorson: four walls that were missed by lidar in this concentrated setting and we began to classify them differently i'm going to blow this one up so you can see it.

94
00:15:27.800 --> 00:15:30.350
robertmthorson: Okay, so what we have here are two foundations.

95
00:15:30.860 --> 00:15:43.850
robertmthorson: In green and a cemetery in yellow and a dike along the Channel way, this is an industrial village with several mills and that's the that's the the dam itself, so we have you know I mean it's actually a beautiful.

96
00:15:44.270 --> 00:15:54.170
robertmthorson: You know millpond damn up there and that's a wall, so we want to make sure to capture it within the chronology and other thing you'll see here is that is that we have broad walls showing up.

97
00:15:54.890 --> 00:16:06.800
robertmthorson: You know in this in this scene here where they were missed by the lidar because they were too low, but they're definitely there but they're just not built up, so we could argue there's a whole group of broad wall, I mean of.

98
00:16:07.790 --> 00:16:18.320
robertmthorson: Bands out there that are low, to the ground that are being missed by lidar but would qualify as stone walls okay so that's our third area no i'm nearly done will have, am I doing on time.

99
00:16:23.810 --> 00:16:24.050
William Ouimet: When.

100
00:16:24.350 --> 00:16:25.520
William Ouimet: you're in good shape or.

101
00:16:25.580 --> 00:16:30.500
robertmthorson: I mean you haven't done i'm shooting for questions so so i'm hoping to get through this.

102
00:16:31.100 --> 00:16:36.590
robertmthorson: So basically what happens is that in the lidar the kinds of things you let's look at the bottom of the screen here.

103
00:16:36.860 --> 00:16:49.130
robertmthorson: In the lidar you're really can map parcels so our units are are basically a parcel is the field, or the the the Polygon, if you like, you know that has wall surrounding it.

104
00:16:49.490 --> 00:17:00.170
robertmthorson: And within a parcel we have walls and there may be, you know, up to a multiple number of walls that define an outline of parcel, then there are segments, you know that, basically, make up a wall.

105
00:17:01.010 --> 00:17:09.290
robertmthorson: And, and then there are stones within segments so there's like a spatial classification that goes parcel wall.

106
00:17:10.520 --> 00:17:21.680
robertmthorson: segment stone or feature in the G on the lidar does a wonderful job mapping parcels areas sides of parcels and the patterns of parcels on the landscape.

107
00:17:22.910 --> 00:17:25.640
robertmthorson: And it can also measure links with a walls.

108
00:17:27.050 --> 00:17:40.700
robertmthorson: But the field methods allow you to do so much more we can do terminations that is the nature of how a wall ends, whether it's built or not, and it allows us to define segments all of the dimensions, you know top bottom with.

109
00:17:42.050 --> 00:17:48.140
robertmthorson: You know intermediate with media and with whatever tax on contacts, you know what kind of wall is it.

110
00:17:48.890 --> 00:18:01.100
robertmthorson: You know how does, how does, how does the walls blend to the junction do they bend they merge gaps in walls that are missed online darn symmetry preservation and by the stones we're talking about mythology size, shape order and surfaces.

111
00:18:01.520 --> 00:18:07.490
robertmthorson: And so there's all kinds of things you everybody knows this in the audience that you see on the ground that you can't see.

112
00:18:07.910 --> 00:18:13.070
robertmthorson: And i'm just trying to point out that if we can systematically collect that information.

113
00:18:14.060 --> 00:18:26.690
robertmthorson: At the stone or segment or wall level, you know by being in the field with some kind of a rapid fire protocol, you know working to be anywhere near as good as the lidar is we're really going to advance Dome all science.

114
00:18:27.650 --> 00:18:39.410
robertmthorson: And so, their conclusions Now this is simply Council for sense that we collected from our 273 segments in these three areas, and this is where i'm going to end my talk, because this is a pilot.

115
00:18:39.410 --> 00:18:45.920
robertmthorson: study we did not go back in and study the LIFE ology we didn't go in and study the aesthetics, of the pattern or the.

116
00:18:46.550 --> 00:18:53.810
robertmthorson: You know, we just let all that complicated stuff go for now, because we were just interested to see if we could find some patterns in counts and frequencies.

117
00:18:54.440 --> 00:19:01.100
robertmthorson: And indeed, we can, and let me remind you, the key unit here is not the wall it's the segment and the segment is invisible on lidar.

118
00:19:01.790 --> 00:19:11.840
robertmthorson: And so, of the 273 segments, what we have our that in yukon forest and wormwood Hill, these were areas that were farmed we do not see any of the of the.

119
00:19:12.740 --> 00:19:19.190
robertmthorson: You know the enclosing supporting and blocking walls, we just not seeing them we're seeing the standard New England pattern.

120
00:19:19.700 --> 00:19:23.090
robertmthorson: that's going to be responsible for most who we have zero percentage of those.

121
00:19:23.570 --> 00:19:31.340
robertmthorson: And, most of the stone walls are you know bands single double and brought this is a continuum of band is simply rubble on the ground.

122
00:19:31.550 --> 00:19:41.420
robertmthorson: A single wall is it stacked up with some degree of order a double wall is it's more carefully built from both sides and abroad wall is is basically like a.

123
00:19:41.990 --> 00:19:51.680
robertmthorson: They call it, they call them consumption walls in Britain, this is the classic extra wide wall, where you have to walls build side by side and you feel the middle in with rubble.

124
00:19:52.130 --> 00:20:06.590
robertmthorson: Okay, but if you look at the pattern there isn't a huge difference between between the wormwood hill site, which is the really good farming site that continued into the 20th century in the yukon for us that was abandoned much earlier, there are differences.

125
00:20:07.820 --> 00:20:20.090
robertmthorson: For example, the wormwood hill surprisingly has more bands, you know, then, does the more hardscrabble farm and the reason is, is that it's a more productive farm that is continuing to expand out into the woods.

126
00:20:21.260 --> 00:20:29.240
robertmthorson: The single walls are roughly the same that's the standard notice how single walls are more common than anything else, but at least we can say this.

127
00:20:29.810 --> 00:20:37.070
robertmthorson: You know we've known it, but now we can really say it with numbers, at least in these three parcels the double walls are architectural.

128
00:20:37.760 --> 00:20:45.200
robertmthorson: You know it's like they're a step up above single walls and I don't see any significant differences there and broad walls of roughly the same.

129
00:20:45.650 --> 00:20:53.780
robertmthorson: If anything there's an inversion broad walls might be slightly more common and yukon forest because of the hardscrabble more rapidly nature of it.

130
00:20:54.260 --> 00:20:58.490
robertmthorson: I wish we had better data and look at this the bottom one in ashford the village.

131
00:20:59.210 --> 00:21:05.030
robertmthorson: very, very different percentages, and of course that's a function of really how far out we expanded our scale.

132
00:21:05.390 --> 00:21:13.910
robertmthorson: If we expanded our scale all the way the villages would be absorbed within the larger data set and that's in fact what we're doing by picking three pilot areas.

133
00:21:14.360 --> 00:21:21.380
robertmthorson: With that I think i'm going to call it quits and end up with my Thank you slide i'd like to acknowledge certainly will and richards.

134
00:21:21.920 --> 00:21:31.160
robertmthorson: able assistance throughout this project, my apologies to them for falling behind on on moving in as forward as fast as I wanted, but the GSA.

135
00:21:31.580 --> 00:21:44.570
robertmthorson: talk today pushed me forward a little bit and, of course, every everybody at uconn geosciences yukon geosciences are very involved in the stonewall effort, right now, so with that I think i'll quit and take questions and.

136
00:21:45.080 --> 00:21:48.620
robertmthorson: I didn't get a chance to set my timer so I don't know how much of my time is left.

137
00:21:48.800 --> 00:21:51.470
William Ouimet: yeah Thor so you know your next again so.

138
00:21:51.560 --> 00:21:53.090
robertmthorson: yeah I do know that I do know that.

139
00:21:53.240 --> 00:21:57.980
William Ouimet: So it's a 24 So if you want to entertain a couple questions you can.

140
00:21:58.040 --> 00:21:58.820
robertmthorson: One is fine.

141
00:21:59.270 --> 00:22:02.120
robertmthorson: i'm going to stop share and then pick up the other one yeah.

142
00:22:02.480 --> 00:22:05.960
William Ouimet: And the other one you do have a question from Bob noon.

143
00:22:06.020 --> 00:22:07.100
William Ouimet: In the chat i'll just read it.

144
00:22:07.400 --> 00:22:08.600
robertmthorson: Okay, thank you that's.

145
00:22:08.750 --> 00:22:17.300
William Ouimet: 54 walls crossing a slope, do you measure wall high on both the uphill and downhill sides to capture sediment accumulation on the hillside.

146
00:22:17.510 --> 00:22:27.050
robertmthorson: Yes, yes we're definitely doing that the database up by the way, i'll share the database, and all this data with anybody who wants it provided you're okay with that well.

147
00:22:27.350 --> 00:22:34.310
robertmthorson: And, and in our in our methods and our protocols were very i'm very close it's embarrassing i've had a wall.

148
00:22:34.520 --> 00:22:41.900
robertmthorson: i've had a taxonomy ready to publish for a number of years and I just didn't finish the damn thing you know whoops sorry about that I just didn't finish it.

149
00:22:42.530 --> 00:22:54.950
robertmthorson: But it's going to be going in very soon and it's shareable it's already been out for review by a number of people as collegial reviews and i'm definitely interested in sharing and contact me if you want it and that's a subject to the next talk.

150
00:22:55.340 --> 00:23:05.750
William Ouimet: And Thor another quick question one I think women here, much of the areas are well preserved about how much wall has been potentially removed or material that some Ben fellow fellows.

151
00:23:06.080 --> 00:23:19.610
robertmthorson: We in these areas we don't know and we really can't tell, but we are making field annotations for that effect when we're there that's a very important point so let's move on to i've got to share my screen now and.

152
00:23:19.850 --> 00:23:20.420
robertmthorson: My other.

153
00:23:20.750 --> 00:23:23.840
William Ouimet: yeah you pull up there's one more question I think you'll address, and the next one.

154
00:23:23.960 --> 00:23:26.090
robertmthorson: Is this is the taxonomy talk up.

155
00:23:26.360 --> 00:23:28.190
robertmthorson: yep because.

156
00:23:30.620 --> 00:23:34.400
robertmthorson: Otherwise, because i'm already behind on that so i'm going to share the screen.

157
00:23:36.890 --> 00:23:38.030
robertmthorson: There it is good.

158
00:23:42.200 --> 00:23:42.620
William Ouimet: And the last.

159
00:23:42.980 --> 00:23:44.690
robertmthorson: Will you please confirm that everything looks.

160
00:23:44.720 --> 00:23:45.770
robertmthorson: looks hunky dory.

161
00:23:45.920 --> 00:23:47.540
William Ouimet: Your I just go to presenter mode.

162
00:23:47.900 --> 00:23:49.550
robertmthorson: yeah i'm trying to find it yeah.

163
00:23:53.120 --> 00:23:54.920
William Ouimet: Okay okay sounds good you're good.

164
00:23:55.250 --> 00:23:55.910
robertmthorson: Okay, good.

165
00:23:57.410 --> 00:24:06.140
robertmthorson: Now i'm going to check out my clock yeah i'm already two minutes behind this is the longer talk again support from the nsf i'd also like to say support from.

166
00:24:06.500 --> 00:24:19.250
robertmthorson: My publisher with a travel advanced to travel 20,000 miles throughout everywhere in New England and come up with this book exploring stone walls, they bought me a computer and gave me a travel advanced and said go class if I was how do you like that and publisher doing it.

167
00:24:20.870 --> 00:24:22.310
robertmthorson: And of course geosciences.

168
00:24:22.370 --> 00:24:23.840
robertmthorson: i've just lost my cursor.

169
00:24:27.770 --> 00:24:28.460
robertmthorson: not good.

170
00:24:29.930 --> 00:24:35.510
robertmthorson: No, it works I click forward, let me see if I can oh there we go i'll just use my toggle bar.

171
00:24:36.140 --> 00:24:48.860
robertmthorson: Okay let's just ask some questions about this guy this mock revolutionary guy standing in front of the walls at minimum National Park, you know, is this a stone wall well some people would say it isn't they would call it a hybrid wall.

172
00:24:50.450 --> 00:25:00.290
robertmthorson: Or, they would call it a wood fence with a stone bottom but that's a pretty basic question we know if everything rotted you know we'd have nothing left except the stone and we would call it a stone wall well.

173
00:25:00.680 --> 00:25:08.300
robertmthorson: What is this stone wall, how do you define a stone wall, I mean we can map them, but without defining them it isn't really that scientific.

174
00:25:09.140 --> 00:25:16.760
robertmthorson: And so the lidar definitions are clear, you know but they're basically what you can see, on the map do walls have parts.

175
00:25:17.330 --> 00:25:23.840
robertmthorson: Well, of course, they do, and those are fundamentally called segments, and then of course the stones and are there, different kinds of walls.

176
00:25:24.770 --> 00:25:29.570
robertmthorson: Of course, and how do you describe a wall, of course, this is what we're working through.

177
00:25:29.870 --> 00:25:33.350
robertmthorson: On this it's been taken forever so look at all those, this is my outline.

178
00:25:33.560 --> 00:25:45.140
robertmthorson: we're never going to get through all of these but, basically, I want to make sure you know, the rationale the previous work, the principles of classification no different from you know, Carl and as and clinics in biology and so forth.

179
00:25:45.410 --> 00:25:56.810
robertmthorson: i'm trying to follow a rigorous method here i'm defining a stone wall based on five criteria there's a fundamental unit is the segment a little protocols for description.

180
00:25:57.530 --> 00:26:03.530
robertmthorson: A little bit about taxonomy and a segment called consider this where we just look at one wall different ways.

181
00:26:03.920 --> 00:26:09.770
robertmthorson: i'm going to skip the pilot study but i'm going to show you that I actually have one i'd forgotten about this, believe it or not.

182
00:26:10.520 --> 00:26:18.560
robertmthorson: It was actually published in an extended abstract in the Massachusetts historical society and I completely forgot about it inventory and conclusions.

183
00:26:19.250 --> 00:26:26.720
robertmthorson: So the rationale, the first lines of the abstract, you can read them in four or five segments, you can do this later that's the rationale for doing it.

184
00:26:27.050 --> 00:26:38.660
robertmthorson: The basic idea is, we had a sprawling agro ecosystem all over interior and island New England, this is block island and that that that was just the the LIFE way.

185
00:26:39.860 --> 00:26:50.960
robertmthorson: And then of course stone is the leftover part of that material culture, the tangible result of that lifeway or that former civilization is stone everywhere and, of course.

186
00:26:51.440 --> 00:27:02.840
robertmthorson: The most common of those stone items, but not the only ones, the most common are the stone walls, and so I dubbed them the signature land form of rural New England.

187
00:27:03.110 --> 00:27:17.090
robertmthorson: Because they are land forms and they are just about everywhere in our region, the issue is explaining their absence elsewhere, the issue is explaining their presence and that's a world of difference and that's a whole nother talk.

188
00:27:18.200 --> 00:27:28.130
robertmthorson: Okay light our mapping, you know of walls, that is, the thing from the Johnson, and we met paper in on the yellow you know they're basically all one thing and I don't have a problem with that.

189
00:27:28.700 --> 00:27:35.000
robertmthorson: that's wonderful we actually need that we need to start somewhere and move on with the mapping lidar was a revolution to me.

190
00:27:35.390 --> 00:27:54.140
robertmthorson: This is the guy that in 1985 at uconn taught a course on aerial and photo interpretation you know back in the Old World War Two days of technology lidar has revolutionized everything but there's great variability in the landscape, even in the wall we're looking at there's variability.

191
00:27:55.340 --> 00:28:05.690
robertmthorson: And look at this, I mean we range from these rounded court sites, you know, on montauk point you know, to the quarry stain granted right here on the yukon campus.

192
00:28:05.990 --> 00:28:13.610
robertmthorson: And there's a drastic wall in the upper left you know, in a field stonewall in plainville in needs obviously not high enough to be a fence.

193
00:28:14.000 --> 00:28:30.050
robertmthorson: There are remnant of a walls everywhere different shifts on the left and Nice on the right, you know, surely this matters, if you found this somebody built a wall out of Nice, you know, on a hill underlined by chest, it would tell you they imported the stone.

194
00:28:31.250 --> 00:28:41.420
robertmthorson: Three project goals very distinct but they all add to each other let's define a wall let's figure out description protocols for a wall.

195
00:28:41.930 --> 00:28:51.410
robertmthorson: And let's find a classification for them it's it's been agonizing and delightful at the same time, because when you invent a classification.

196
00:28:51.620 --> 00:29:02.930
robertmthorson: You have to choose the sequence of binaries that lead you to the outcome and swapping those sequence of binaries around leads to better or worse classifications very much trial and error.

197
00:29:04.460 --> 00:29:13.160
robertmthorson: Okay previous work the vernacular historic terminologies are are everywhere, you know it's a pastor wall it's a throne wall it's a toss wall it's a dumped wall.

198
00:29:13.400 --> 00:29:21.260
robertmthorson: it's a it's a mill damn wall it's a whatever I have researched these to my extent, and I can find any consistencies to them.

199
00:29:22.160 --> 00:29:38.390
robertmthorson: And if you go to England and look for prototypes into the British islands, for example, or or the Mediterranean, you know what you find are principally architectural and archaeological or cultural overlay styles and types, not the fundamental attributes of the wall.

200
00:29:39.470 --> 00:29:47.240
robertmthorson: Through three books on stone walls in the late 90s early 2000s, you know I really, really started working on this.

201
00:29:47.570 --> 00:30:00.380
robertmthorson: And by 2005 I wanted to publish a in a you know, in a book I published a stone wall classification taxonomy in this book exploring stone walls, but the goal and that was partially, supported by the nsf.

202
00:30:01.250 --> 00:30:10.850
robertmthorson: On a project that I had at the time, but the goal, there was pretty pretty much to get one out there in the hands of as many readers as I could, so we could iterate and work our way towards a better one.

203
00:30:11.090 --> 00:30:18.860
robertmthorson: That was never intended to be the end I wanted to get feedback and I got tons of feedback and it helped improve the classification a lot.

204
00:30:19.790 --> 00:30:24.890
robertmthorson: By 2008 I had an online taxonomy in a pilot project in 2009.

205
00:30:25.220 --> 00:30:37.670
robertmthorson: And and i've had a manuscript circulating for about two years now, and you know what every time someone reads it they improve it so it's one of these things where I think i'm holding back until I get it right because i'd hate to have it, to begin with.

206
00:30:38.150 --> 00:30:45.680
robertmthorson: But that's an excuse we've all played over the years here's what it looks like on the stonewall initiative you can go there and see an older version of this taxonomy.

207
00:30:46.130 --> 00:30:55.820
robertmthorson: haven't dared change it, because otherwise you know why would the Why would a journal want to publish a new one, especially such a big thing is this, I thought of doing this in book form at a.

208
00:30:56.510 --> 00:31:05.990
robertmthorson: At a you know, an academic press and do that kind of review, but because it's too big for a single journal article Okay, the principles of classification.

209
00:31:06.890 --> 00:31:12.740
robertmthorson: Really really important this is an objective classification anyone in this audience should be able to walk up to it.

210
00:31:13.010 --> 00:31:22.730
robertmthorson: And, if properly trained, they should basically reach the same conclusion, it should be objective that is you're describing what you see, not what you interpret what you see, not what you think.

211
00:31:23.180 --> 00:31:27.860
robertmthorson: Okay, as opposed to a genetic classification, in other words, you have to know what.

212
00:31:28.280 --> 00:31:36.050
robertmthorson: You have to know what happened in order to classify it like a criminal you can't see a criminal by looking at them, you know, but a criminal is somebody who committed a crime.

213
00:31:36.500 --> 00:31:38.720
robertmthorson: Alright, so that's a genetic classification.

214
00:31:39.500 --> 00:31:43.370
robertmthorson: A functional classification of purpose stone fence, for example.

215
00:31:43.610 --> 00:31:50.690
robertmthorson: Is a functional classification and man that's created lots of problems because walls are not all fences and fences are not all walls.

216
00:31:50.900 --> 00:32:03.020
robertmthorson: But you're throwing that functional classification in their chronological stages early middle late, you know, or whether we're looking at a sadistic sequence, you know this wall leads to that wall leads to that wall or an aesthetic.

217
00:32:03.500 --> 00:32:10.220
robertmthorson: Know there's so many basis of classification, we as scientists want the Objective one, the end Members, as you know.

218
00:32:11.060 --> 00:32:16.310
robertmthorson: we're all eukaryotes you know, but that really doesn't differentiate us very much from homonyms does it.

219
00:32:16.730 --> 00:32:26.570
robertmthorson: You know, or for that matter, even mammals, you know So yes, the most inclusive definition is a good one, because it will you know will not artificially include the wrong things.

220
00:32:26.900 --> 00:32:34.910
robertmthorson: But the narrowest categories are also bad because you know we're all homo sapiens were one thing yet look at the variety within one thing.

221
00:32:36.080 --> 00:32:48.800
robertmthorson: So what we're really trying to do is get a mid range in the between those end members in stone walls and it's it's tough because there's trade offs every time you adopt one protocol for classifying you know you short another one.

222
00:32:50.390 --> 00:32:58.760
robertmthorson: The categorical rigor there's a typology these are pigeonhole sorting bins no rules genres that's really what we have today single wall or double wall.

223
00:32:59.390 --> 00:33:08.150
robertmthorson: Those are probably the best almost taxonomic units that i've chosen to adopt and, by the way, in my language, I try to take as much of the common language as possible.

224
00:33:08.510 --> 00:33:15.560
robertmthorson: And not invent very complicated ones sorting vince we're not doing that the taxonomy is a rule driven decision tree.

225
00:33:16.100 --> 00:33:22.520
robertmthorson: In other words, you if you know how to use it, you shouldn't make any mistakes, because its presence or absence I took a.

226
00:33:22.970 --> 00:33:31.670
robertmthorson: plan taxonomy in college and i'll never forget how how cool that was to learn that it's one little tiny rule at a time that will lead you into the right place.

227
00:33:32.360 --> 00:33:39.020
robertmthorson: There nested binaries there's a trial and error sequencing of choices that i've talked about and I finally ended up with the ones that I like.

228
00:33:39.440 --> 00:33:43.940
robertmthorson: And, and apparently my reviewers like as well, because they haven't told me to use something different.

229
00:33:44.480 --> 00:33:55.790
robertmthorson: there's an in group out group consequence for the whole thing every single time you make a definition, you are creating an in group that's separate from every other group and we do this at every level in the taxonomy.

230
00:33:56.960 --> 00:34:00.320
robertmthorson: And this requires an explicit definition for every tax on.

231
00:34:01.340 --> 00:34:13.160
robertmthorson: it's composed of shared and unique characteristics, now the definition, then, if you call something a double wall, it means it's also a freestanding wall which means it's also a wall.

232
00:34:13.460 --> 00:34:26.840
robertmthorson: Which means it's part of the stone domain, in other words by just calling it a double wall you're carrying all of the more primitive attributes in with the definition it's a way of short handing it and allowing faster and better communication.

233
00:34:27.620 --> 00:34:33.140
robertmthorson: So some sample binaries here's a here's a here's a wall, the oldest wall, I know of in our continent here.

234
00:34:33.620 --> 00:34:45.170
robertmthorson: That looks like this is the European agro equal cultural wall on the Viking settlement a Newfoundland Okay, but think about the chronologies we could invent here or the binaries there's Viking or posts liking.

235
00:34:46.520 --> 00:34:49.670
robertmthorson: there's a field stone component and accord component.

236
00:34:50.750 --> 00:35:02.120
robertmthorson: there's a lead component where you're setting stones in place and a dumped component in a size component of things too big to carry or half and those that are not.

237
00:35:03.110 --> 00:35:11.120
robertmthorson: So what's the correct sequence do we start with sighs and go to chronology or do we start with source and go to order well.

238
00:35:12.170 --> 00:35:15.560
robertmthorson: we'll find out, so my purpose in.

239
00:35:16.820 --> 00:35:34.160
robertmthorson: In defining a wall is to isolate it as an as an in group and the method is a status, you know it's basically follows the rules of logistics, except we're not looking at primitive to advanced we're looking at an organization hierarchical organizations in groups.

240
00:35:35.480 --> 00:35:47.690
robertmthorson: The main in group is material culture and I, I think a bulk of material culture not tea pots and dinner plates and clay pipes and the sort of buttons that and musket balls you know that that dominate historic archaeology.

241
00:35:47.870 --> 00:35:59.240
robertmthorson: i'm thinking about the bulk outdoor stuff you know what's there on a clipper ship loaded with stone ballast well basically it's would you know it's metal it's fiber and it's stone.

242
00:35:59.840 --> 00:36:13.010
robertmthorson: And, and so we're working at the outdoor scale on big things so so the the we're looking at at just the stone domain and material culture and we're looking at just the class of stone walls within the stone domain.

243
00:36:14.810 --> 00:36:17.420
robertmthorson: And here's the five criteria that I think are necessary.

244
00:36:18.470 --> 00:36:31.940
robertmthorson: In the upper left it has to be made of stone and i'm not saying this can't be a synthetic stone, because a brick wall, I mean it just gets complicated but, but you will, for now, just stone, it has to be made of stone otherwise it's not a stone wall.

245
00:36:33.140 --> 00:36:42.140
robertmthorson: Okay, in the upper right a stone wall is a granular thing it's made of multiple things just like our body is made of cells, a stone wall is made of stones.

246
00:36:43.550 --> 00:36:57.200
robertmthorson: A stone wall has to be continuous go to the lower left that's not a stone wall it's a stone line it's not a stone wall because it's not continuous I just define a wall as being continuous you can walk right through that so i'm not going to call it a wall.

247
00:36:58.520 --> 00:37:12.410
robertmthorson: The one in the lower middle elongation that is not elongated everything else would work, though, you know it's just not elongated so it's not a stone wall it's a Karen or an art object, or whatever else you want to call it it's a concentration.

248
00:37:14.300 --> 00:37:22.490
robertmthorson: On the right stone walls have to be a certain height, this is hard, if you have stone, on top of stone that works for me in every case.

249
00:37:22.940 --> 00:37:32.150
robertmthorson: But that doesn't do it all, because sometimes you have a budding walls, where the large stones are put together and they make a wall, so there has to be a a you know, a height.

250
00:37:32.630 --> 00:37:41.990
robertmthorson: And it's somewhere between you know knee height or below and anke somewhere between ankle and he is the right height and we can quibble about that, but it's going to be arbitrary.

251
00:37:42.770 --> 00:37:52.250
robertmthorson: And, and maybe that can be set locally So these are the five criteria, and this is one of the funniest slides sent to me by calling from the museum on the left.

252
00:37:52.700 --> 00:38:05.480
robertmthorson: That was a clearing operation of an old auto junkyard in Vermont but everything about that photo works as a stone wall, except that the material it's made of old tires not of stones so.

253
00:38:05.510 --> 00:38:06.290
William Ouimet: or five minutes.

254
00:38:06.650 --> 00:38:08.120
robertmthorson: yeah okay So here we go.

255
00:38:09.320 --> 00:38:17.600
robertmthorson: The cement that's a stone that's a bag so man somebody left out that's an outcrop on the right okay so i'm going to speed real Thank you well for that morning.

256
00:38:17.990 --> 00:38:25.280
robertmthorson: The stonewall the fundamental unit is the segment, the thing in blue there they some two walls, they some two parcels.

257
00:38:25.760 --> 00:38:31.190
robertmthorson: There you see it on the on the on the on the left there, there are two segments, and here it's even better.

258
00:38:32.000 --> 00:38:46.130
robertmthorson: that's a shift wall corey just and corey granted that's one wall, but two segments, and on the right, we have two segments, we have a laid nice wall on the right and we have a blocky stacked granted wall on the left at Nathan hale forest.

259
00:38:46.580 --> 00:38:51.590
robertmthorson: description rock stone boulder erratic soil till grade all of these things.

260
00:38:52.640 --> 00:39:01.340
robertmthorson: shape I think i'll leave it at this i'll speed through my talk, there are different shapes that you can describe that are based on on math, if you like.

261
00:39:01.940 --> 00:39:18.320
robertmthorson: There are a geologists are really used to that their size source degree of order mythology you know wall unit and setting look at all that hierarchy view segment context wall terminations These are the kinds of things we're recording in our in our field data.

262
00:39:19.430 --> 00:39:30.980
robertmthorson: The line of the wall, the top of the wall, the face of the wall, the end of the wall in different directions mountains triangles trap, besides panels asymmetric a tier wall is how many tears tears of the bulk.

263
00:39:31.010 --> 00:39:31.700
robertmthorson: groupings.

264
00:39:31.850 --> 00:39:50.300
robertmthorson: In a vertical direction and courses are the individual layers drywall wet wall filled rubble harding CAP Cape all of these terms are used in the classification and here's some examples, the wall on the left is the stones are balls that were headed made of Granite Nice and they're stacked.

265
00:39:51.530 --> 00:40:07.910
robertmthorson: The walls on the right on the on the bottom part there you were looking at the face of a wall all one segment no contacts between segments it's basically one tiered not course there's no matrix and it's sort of an upland setting.

266
00:40:09.140 --> 00:40:19.850
robertmthorson: So the taxonomy is ridiculous, but there it is the wall is the biggest category that's the class below that's a family blow that's a type below that's a subtype below that's a variant.

267
00:40:20.240 --> 00:40:30.320
robertmthorson: And there are informal vernacular terms in the middle column for all of these things that we've we've been talking about and what i'm trying to do is give each one a definable name that is objective.

268
00:40:31.760 --> 00:40:42.620
robertmthorson: And the challenges naming conventions, what do you do with buildings stone circles or another challenge, there are lots of arbitrary thresholds that are necessary, even though their scientific.

269
00:40:43.430 --> 00:40:53.810
robertmthorson: flanking walls are complicated because of Bob newton's question, in other words, you can build a wall on a slope, and it looks like a retaining wall but it's not a retaining wall, so I try to avoid that.

270
00:40:54.380 --> 00:41:02.270
robertmthorson: functional classification and then hybrids in tears and courses, there are a lot of problems but here's the basic levels and then i'll quit okay.

271
00:41:03.110 --> 00:41:17.270
robertmthorson: A wall meets all five criteria material granularity continuity elongation and height align fails the continuity criteria, a concentration fails elongation criteria and a notable stone fails the granular criteria.

272
00:41:19.130 --> 00:41:20.210
robertmthorson: And here's the basic.

273
00:41:21.950 --> 00:41:29.690
robertmthorson: You know types of walls, there are bands and they're all defined different ways, forget the names, you know they're defined explicitly.

274
00:41:30.740 --> 00:41:35.750
robertmthorson: Even though a budding you know is is a suggestion functional term there, we can define them.

275
00:41:36.410 --> 00:41:36.860
William Ouimet: One minute.

276
00:41:38.030 --> 00:41:44.870
robertmthorson: All right, and then and then within that within the within the freestanding walls, we have broadband double a budding and single.

277
00:41:47.330 --> 00:41:57.980
robertmthorson: and consider this i'm going to speed right over at here's a wall that is called a stone fences stonewall a rock wall a toss wall a throne wallet pastors wall a farmer's wall a single wall, etc.

278
00:41:58.310 --> 00:42:09.890
robertmthorson: You know and we're going to call it something different it's a three tiered paneled single wall it belongs in the class wall family freestanding type single subtype panel and variant fitted and i'm just going to blast right through that.

279
00:42:10.700 --> 00:42:14.690
robertmthorson: And then, our pilot test was in seller on square in mansfield it looks something like this.

280
00:42:15.230 --> 00:42:26.030
robertmthorson: actually came up with numbers, you know for the number of segments we map 4468 feet and the average Wall was only for segment was only 44 feet long.

281
00:42:26.750 --> 00:42:33.590
robertmthorson: And blah blah blah blah blah the inventory looks like this you've seen that in the previous talk if you joined it.

282
00:42:34.280 --> 00:42:40.550
robertmthorson: And then the conclusions are the taxonomy works works and all it's a tool.

283
00:42:40.940 --> 00:42:51.410
robertmthorson: A new methodology, in the same way that that lidar is a tool that allows you to do great science at the spatial scale a good taxonomy would be a tool.

284
00:42:51.860 --> 00:43:00.290
robertmthorson: You know, a new methodology that would allow us to move forward into better stone wall science and I have worked with countless people.

285
00:43:00.470 --> 00:43:07.670
robertmthorson: I mean I don't even have an acknowledgement section, because the terms were actually invented from talks and interviews and guidance and discussions.

286
00:43:07.970 --> 00:43:20.750
robertmthorson: And it's really been a collaborative project and i'm going to i'm very close to synthesizing the whole thing so with that I think i'll be done, thank you very much for your attention and i'll take a few questions until will kicks me off because.

287
00:43:22.520 --> 00:43:24.320
robertmthorson: he's got the timer so i'm going to.

288
00:43:24.560 --> 00:43:25.070
William Ouimet: The Board.

289
00:43:25.370 --> 00:43:38.480
William Ouimet: Why don't you share so that you can start sharing and if there's any questions I would either I don't know Thor addresses but but do go ahead and type questions into the chat box or put a flag raise your hand and hopefully will.

290
00:43:38.810 --> 00:43:40.760
robertmthorson: I can, I can try and answer them.

291
00:43:41.690 --> 00:43:52.670
William Ouimet: One quick one quick question that was was from last talk, but I might apply here low bandwidth back with asked just to confirm a segment is defined as one side of a Polygon that's that's.

292
00:43:53.000 --> 00:44:02.600
robertmthorson: segments know a segment is is a linear it's a stretch along the wall, where things are more or less a uniform, you know as you define them.

293
00:44:03.110 --> 00:44:13.490
robertmthorson: And, and you know you can see two segments can about each other, they can salvage each other, they can grade into each other, you know, one might be all granted boulders and one might be all just rubble.

294
00:44:13.850 --> 00:44:20.180
robertmthorson: You know, and those would be in a straight line on one side of a field that would be one wall, you see, with two segments.

295
00:44:21.860 --> 00:44:23.660
robertmthorson: I hope that answered the question yeah.

296
00:44:24.080 --> 00:44:30.560
William Ouimet: That did it last question well gee loads for talk i'm taking a note from biology sorry my chat window.

297
00:44:32.930 --> 00:44:37.940
William Ouimet: Are stone walls going going extinct and do we need to have a stone wall preservation society.

298
00:44:38.480 --> 00:44:51.200
robertmthorson: Yes, well, yes, they are they Yes, they are, I think we're actually probably building them fast and they're disappearing, but I don't know that yet that's where you're going to have to combine the lidar you know with the classification in order to figure that out.

299
00:44:51.440 --> 00:44:52.820
robertmthorson: I think we originally had.

300
00:44:52.850 --> 00:45:01.700
robertmthorson: Probably 250,000 plus or minus you know great big standard air how many walls, we had at the outset, and I think an awful lot of them disappeared.

301
00:45:02.090 --> 00:45:14.090
robertmthorson: And one of the things that motivated me was the conservation effort, I gave 1000 talks stump conservation and I really believe we need a well the stonewall initiative one of its goals is to conserve stone walls.

302
00:45:14.450 --> 00:45:22.280
robertmthorson: And that's been that's been a big part of my effort, I think, just turning archaeology into something else is not a good idea.

303
00:45:22.490 --> 00:45:25.730
William Ouimet: that's good, well, we would hope that this session is part of that preservation.

304
00:45:25.730 --> 00:45:38.480
William Ouimet: Yes, so the next we're going to move on to the next speaker continuing on the yukon yukon stonewall focus today we have you on stuff from the Department geography, you can take it away gee.

305
00:45:39.200 --> 00:45:49.400
Ji Won Suh: Thank you good morning everyone, my name is do you want to talk and our groups work is about automated mapping of stores in northeastern United States using these learn.

306
00:45:50.450 --> 00:46:02.600
Ji Won Suh: And here is our brief outline so start with the introduction and i'm going to explain about data method, and then I will explain results and discussion and conclusion.

307
00:46:03.620 --> 00:46:08.690
Ji Won Suh: So in the introduction, Section i'm going to briefly touch on background and objectives.

308
00:46:09.860 --> 00:46:20.360
Ji Won Suh: So in New England history land yes legacy is presented through anthropogenic landscape features such as stone worse, so if you are living in.

309
00:46:20.810 --> 00:46:37.190
Ji Won Suh: New England, you might see the walls made up stones and these stone words are ubiquitous features in the forest landscape and they were built to mark property boundaries and agriculture activities during a teens to early 20th century.

310
00:46:38.480 --> 00:46:44.870
Ji Won Suh: and identifying and mappings donors throughout this region is really important, because it allows.

311
00:46:45.860 --> 00:46:54.590
Ji Won Suh: For detailed understanding of historic land use and land cover and also human impacts prior to the mid to late 20th century.

312
00:46:55.460 --> 00:47:03.350
Ji Won Suh: And the most widely used methods for solar mapping is manually the citation using one meter lidar data sets.

313
00:47:03.980 --> 00:47:15.890
Ji Won Suh: Because stoneware is really small scale features, so that we cannot see those stone worse in lesser data Center to data which has chorus resolution.

314
00:47:16.340 --> 00:47:32.510
Ji Won Suh: But lighter works for represent the stone worse, so when you look at the middle pictures, you can see the stone worse on the lighter and also in some cases, we can see stonework through air photo and, obviously, we can see in the field.

315
00:47:33.950 --> 00:47:39.680
Ji Won Suh: However, manual digitization is very time consuming particularly remapping best area.

316
00:47:40.490 --> 00:47:46.850
Ji Won Suh: So, in response to this limitation deep learning has great potential to automate the mapping process.

317
00:47:47.240 --> 00:48:05.180
Ji Won Suh: Therefore, our study aims to firstly to detect stone worse in forested area using deep learning algorithm particular unit model and then secondly, conduct a Christian assessment with field verified stone worse to evaluate the model prediction.

318
00:48:06.590 --> 00:48:27.110
Ji Won Suh: So before jumping to the data method i'm going to briefly introduce the area and for deep learning, we need to different study area first one is for training and the other one is for testing so training data, says the collected five patches from three different house in Connecticut.

319
00:48:27.350 --> 00:48:30.500
Ji Won Suh: So rootstock count ashford account and random pounds.

320
00:48:30.830 --> 00:48:39.590
Ji Won Suh: And he tried to include stone words in with different settings such as smooth terrain area rugged terrain and develop area.

321
00:48:40.730 --> 00:48:49.130
Ji Won Suh: And for this testing sites, we tested based on the yukon forest and wormwood feel mansfield and kinetic.

322
00:48:51.470 --> 00:48:59.780
Ji Won Suh: And for the data set on, there are two purpose for collecting data set first one is for training the model so training data set.

323
00:49:00.170 --> 00:49:02.450
Ji Won Suh: And the other one is drawn to this data set.

324
00:49:03.350 --> 00:49:18.950
Ji Won Suh: So for the training data set we basically use the 2016 lighter Dr bm from here, we created help shape map and slope map and we compensate those layers as a training input data sets.

325
00:49:19.760 --> 00:49:40.190
Ji Won Suh: And for the ground for status that we basically do the onscreen many manual visitation and fields working so for a manual digitization we use a map and leave off also photo and Google Street View and also we collected field verified stone words in the field.

326
00:49:42.140 --> 00:49:57.050
Ji Won Suh: And this bigger shows that brief workflow that we went through so there are four steps so firstly we collect ground truth data fed, as I mentioned before we use for field marketing, as well as 10 digitizing.

327
00:49:57.680 --> 00:50:05.000
Ji Won Suh: And then we prepare the images for model we created whoa shade and float map and we do the composite images.

328
00:50:05.570 --> 00:50:16.130
Ji Won Suh: Based on three scenarios and i'm going to explain it later and chopping trending images into certain size of images and then on do this.

329
00:50:16.730 --> 00:50:30.680
Ji Won Suh: The data augmentation to increase the total number of training data set, and then we train the model we used unit based model and we use the Python for training.

330
00:50:31.700 --> 00:50:43.130
Ji Won Suh: And lastly, we apply the trained model to the to test sites and they're required to lay the three evaluation matrix which are recall procedures and F1 score.

331
00:50:43.880 --> 00:50:56.630
Ji Won Suh: And you can see, to T n FN n SP until something like that in the recall and precision, this is very important concept to understand accuracy assessment results so i'm going to.

332
00:50:57.140 --> 00:51:06.770
Ji Won Suh: address it a little bit so team is true positive that modal predicate as somewhere, and that is actual stoneware in the field.

333
00:51:07.580 --> 00:51:22.190
Ji Won Suh: and false positives and false negatives miss modal was wrong but little bit different so for esteem is false positive moral credited as stonework, but that is not an actual story the field.

334
00:51:23.570 --> 00:51:36.230
Ji Won Suh: On the other hand, as an means false negative model predicted as known for, but that is actual still worried, the field, so this is very important concept to understand.

335
00:51:37.220 --> 00:51:46.850
Ji Won Suh: And i'm gonna explain more detail in his stuff so firstly collecting ground truth data we use basically one meter lidar.

336
00:51:47.330 --> 00:52:00.830
Ji Won Suh: killer shape and then, in some cases we use the leap off air photo or a Google Street View to double check and then we went out to the field to get the vet field verified stone worse and the.

337
00:52:02.240 --> 00:52:06.230
Ji Won Suh: End based on the steps we got the grant recipients set.

338
00:52:07.250 --> 00:52:13.700
Ji Won Suh: And then we prepare the images based on three different scenario first scenario we use only Northwest.

339
00:52:14.180 --> 00:52:34.700
Ji Won Suh: directional hill shade and second scenario we completed three different bands North East Northwest kill chain and slope man and the last scenario will be five different directional to shape and then look map, so in this case we assume that the more layers are completed, the better results.

340
00:52:36.500 --> 00:52:52.940
Ji Won Suh: And then, to avoid the computational issue we sliced the input images as 2565256 pixels and then the total number of training images that is about 13,000 set images that.

341
00:52:54.110 --> 00:53:02.450
Ji Won Suh: And this figure shows that structure of unit model and, as you can see, you can see the U shape.

342
00:53:03.860 --> 00:53:18.080
Ji Won Suh: Here, so there are two parts one is convolution of love, which is sky blue boxes so then images are going to this combination of block the modal extract the characteristics of features.

343
00:53:18.680 --> 00:53:37.850
Ji Won Suh: And then, when it goes into the up composition as law, which is maybe color box it starts to recommend the images the features so as an output, we can get black or white image black is for background or non stone or images and right for the stone words.

344
00:53:39.710 --> 00:53:54.200
Ji Won Suh: And then, as I mentioned before the output is an image which is rather data, but the the ground truth data we collected is vector poly lines, so that we need to convert it to the vector.

345
00:53:55.070 --> 00:54:05.450
Ji Won Suh: Before titling the recall procedure and effort score, so we do some line adjustment and buffering to calculate the aerial tp and FN n SP area.

346
00:54:06.830 --> 00:54:17.510
Ji Won Suh: So, here are some reasons so firstly i'm going to show you a visual comparison between build verify stone worse and you're not results so let's figure shows that ground truth data sets.

347
00:54:17.990 --> 00:54:38.960
Ji Won Suh: And the rights to your show the model prediction and different colors mean from different scenario of the model so as one refers to scenario one things like that, so when you look compare those two, we can see a modal predicted quite well and also this is for a warm itself sites.

348
00:54:40.160 --> 00:54:51.020
Ji Won Suh: And you can see the in the middle, the smooth terrain area, you can see model predicates quite good, but when we look at this impervious cover, which is rose.

349
00:54:52.040 --> 00:54:57.740
Ji Won Suh: We can see a lot of false negatives reach me as model or read the stone words.

350
00:55:00.080 --> 00:55:10.100
Ji Won Suh: and also in terms of accuracy assessment results those three images are from the scenario One scenario two and number three.

351
00:55:10.550 --> 00:55:21.050
Ji Won Suh: And we can see the distribution of true positive area and false negatives and false positives and also when you look at the right bottom table, you can see the.

352
00:55:21.800 --> 00:55:40.850
Ji Won Suh: score for the accuracy matrix depending on the scenario so as a result snare or three, which is three perform the best in terms of the procedure and recall and F1 score so F1 score is 88% accuracy.

353
00:55:42.440 --> 00:55:52.310
Ji Won Suh: And for the false positives which is green color It means that model predictive as stone worse, but that is not an actual store.

354
00:55:52.880 --> 00:56:07.940
Ji Won Suh: And the modal detects trails or streams as a stone words because they have similar qualities and facilitating niche and overall model shows low Commission era and relatively high on missionary.

355
00:56:11.120 --> 00:56:28.910
Ji Won Suh: And this one is for the warm and phil site and as a result, also scenario three performed the best about 77% accuracy and then we can see a lot of false negatives, the Red color are distributed along the road.

356
00:56:30.410 --> 00:56:38.420
Ji Won Suh: And overall in the similar trend model shows the low Commission error and relatively high omission air.

357
00:56:40.040 --> 00:56:49.280
Ji Won Suh: and based on our results, we have on other questions can we apply the trained model too broad area with reliable chrissy.

358
00:56:49.760 --> 00:57:13.430
Ji Won Suh: So we select coin word town in Connecticut for a town level mapping, because it is for us dominant town where a modal my performed well, so this corn or town is located in the north west side of Connecticut and you can see the associated map and also let me use maps in 2016.

359
00:57:15.920 --> 00:57:23.720
Ji Won Suh: So here is our results, the middle on his hands these ties stone the words and rights to your shows dad model prediction results.

360
00:57:24.500 --> 00:57:32.870
Ji Won Suh: The compared to hand this ties map i'm going to overlay those two lines, the modal pass you over esteem, the words.

361
00:57:33.380 --> 00:57:51.770
Ji Won Suh: But there is uncertainty that some of them are actual because the hand is typing map it's not field verified, so that there is some uncertainty and also, we can see no easy segments in the conifer area that is part of the due to the fact that lighter.

362
00:57:53.210 --> 00:57:56.240
Ji Won Suh: The quality is not good in the conifer area.

363
00:57:57.560 --> 00:58:02.540
Ji Won Suh: So discussion, the discussion session i'm going to briefly explain the challenges.

364
00:58:03.590 --> 00:58:25.340
Ji Won Suh: As an example of Commission errors and omissions so for the Commission error if there is similar policy could characteristics in the hill shade, on the model really confused those lines such as edges of trails will roll those cards things like that so in the seniors josie example.

365
00:58:27.290 --> 00:58:42.830
Ji Won Suh: And so the omission arrow if there is consistent direction between sunlight angle and stone worse model did not predict that well because when you look at the last vezzosi the first figures.

366
00:58:44.510 --> 00:58:58.820
Ji Won Suh: The sunlight angle and stone were direction is the same in this area, you cannot see the old words but in the in the middle, when you look at the middle picture, you can see stone words quite clearly.

367
00:58:59.840 --> 00:59:02.540
Ji Won Suh: so that you can see a lot of false negatives.

368
00:59:03.560 --> 00:59:08.180
Ji Won Suh: are distributed alone, the North West Northwest directional.

369
00:59:09.950 --> 00:59:26.630
Ji Won Suh: And also model with a very sensitive when stone words are located on shaded slope or its own light facing slope shown us nearly black or white in the flow sheets so, but in this case even human cannot differentiate.

370
00:59:27.770 --> 00:59:28.820
Ji Won Suh: The stone worse pipe.

371
00:59:31.100 --> 00:59:41.510
Ji Won Suh: So conclusion so in this study, the proposed a fully automated stone word mapping method using unit model and lighter product.

372
00:59:42.110 --> 00:59:54.320
Ji Won Suh: So the results shows that different directional phil shades for the training input data sets can improve the model actress such as nearly three performs the best among the three scenarios.

373
00:59:55.130 --> 01:00:06.200
Ji Won Suh: And then, and then model works better on their smooth terrain, with, for us it conditions and the highest F1 score me God was 80% in your Conference.

374
01:00:07.580 --> 01:00:22.100
Ji Won Suh: And the trained model in this study, have a potential to be applied to broad areas of just larger than town level, which is the rich will be helpful to reconstruct extend of history land use in the future.

375
01:00:24.110 --> 01:00:26.450
Ji Won Suh: And this is my presentation.

376
01:00:27.530 --> 01:00:29.780
Ji Won Suh: If you have any questions or.

377
01:00:31.040 --> 01:00:32.810
Ji Won Suh: comments, please let me know.

378
01:00:36.620 --> 01:00:37.490
robertmthorson: Thank you very much.

379
01:00:41.690 --> 01:00:44.120
robertmthorson: I did have a question it's a methodological question.

380
01:00:44.870 --> 01:00:53.780
robertmthorson: I don't know how you identify it's I just don't understand it, how do you identify a false negative line on a surface area I mean there should be millions of lines on any given area.

381
01:00:55.400 --> 01:00:59.060
robertmthorson: don't you have to make an assumption about the trend of that line.

382
01:00:59.870 --> 01:01:02.510
Ji Won Suh: of human that false negative one, yes.

383
01:01:05.180 --> 01:01:06.500
Ji Won Suh: My question is general.

384
01:01:07.730 --> 01:01:13.910
robertmthorson: In other words, if you if you're working on a surface area, how do you define a false negative line.

385
01:01:16.220 --> 01:01:25.250
Ji Won Suh: um when you look at the look at this right figures, that is, the output from the model and the and then.

386
01:01:26.720 --> 01:01:32.060
Ji Won Suh: I compare those with ground truth data set to extracted false negatives.

387
01:01:33.230 --> 01:01:34.820
Ji Won Suh: Why yeah yeah.

388
01:01:35.090 --> 01:01:37.970
robertmthorson: My question is more general than any methodology here.

389
01:01:39.350 --> 01:01:42.620
robertmthorson: In other words, a surface area has an infinite number of lines.

390
01:01:43.220 --> 01:01:48.740
robertmthorson: And so, how do you find out whether a line is a negative false negative or a or a positive.

391
01:01:49.040 --> 01:01:59.180
William Ouimet: genius go back up to your two by two matrix of fall of prediction it's more obvious than that Thor because yeah it's yeah.

392
01:02:01.670 --> 01:02:07.250
robertmthorson: Well that's okay i'm sure it's just something i'm not understanding about the about the the geometry so i'll just let it go.

393
01:02:10.610 --> 01:02:22.790
William Ouimet: But it's not infinite it basically it is either in the ground truth data or it's not is the point, so you start with a set of lines that someone has has delineated over that infinite surface.

394
01:02:22.820 --> 01:02:24.320
robertmthorson: What actually okay.

395
01:02:25.460 --> 01:02:31.010
robertmthorson: So you have to start with a map set of lines yes okay yep then that explains my answer, thank you.

396
01:02:31.970 --> 01:02:39.860
William Ouimet: And then bob's got one to Bob Newton G, did you consider a more complex hill shades like multi directional oblique waited.

397
01:02:41.210 --> 01:02:44.150
Ji Won Suh: um yeah I may can try.

398
01:02:47.270 --> 01:02:50.000
William Ouimet: i'll just stop sharing too, and you get it oh yeah sorry okay.

399
01:02:59.660 --> 01:03:01.760
William Ouimet: You can answer that while I get ready.

400
01:03:03.050 --> 01:03:06.500
robertmthorson: Okay yeah we'll Europe what minus one minute.

401
01:03:06.920 --> 01:03:07.220
yeah.

402
01:03:08.420 --> 01:03:16.130
Ji Won Suh: You know, of course, we can try with complex associate like multi directional like waited yeah yeah.

403
01:03:18.530 --> 01:03:30.620
Ji Won Suh: But I didn't try this with I just use the basic for the directional and then also murky directional kill shade in the scenario three.

404
01:03:32.240 --> 01:03:34.190
robertmthorson: Okay, so thank you very much, you want.

405
01:03:34.550 --> 01:03:36.140
robertmthorson: Now we're moving on to will we met.

406
01:03:36.170 --> 01:03:39.470
robertmthorson: he's going to give us a talk about lidar and stone walls are you ready to go well.

407
01:03:39.950 --> 01:03:42.050
William Ouimet: Just about you can everyone good didn't see me.

408
01:03:42.890 --> 01:03:43.370
robertmthorson: I got it.

409
01:03:44.000 --> 01:03:44.330
Okay.

410
01:03:45.680 --> 01:03:46.370
William Ouimet: We ready to go.

411
01:03:47.300 --> 01:03:48.350
robertmthorson: yeah works for me.

412
01:03:49.280 --> 01:03:49.760
robertmthorson: One minute.

413
01:03:50.420 --> 01:03:56.600
William Ouimet: Well, thank you so basically my my goal here is to really to really speak about the broader aspects of the ARV our yukon project.

414
01:03:57.140 --> 01:04:03.920
William Ouimet: To map historic when these features in to reconstruct historical and use we've obviously been talking about about stone walls.

415
01:04:04.400 --> 01:04:10.550
William Ouimet: So far in the talk of the sessions about stone walls, but i'm gonna i'm going to broaden it a little bit to kind of implications for this stonewall mapping.

416
01:04:10.910 --> 01:04:20.480
William Ouimet: and other features that help us reconstruct historic when you select activities okay so um you know and again this is, this is a very large project that we have undergoing.

417
01:04:21.020 --> 01:04:32.900
William Ouimet: In relation to uconn and a number of collaborators and students who are involved and i'm just going to focus in on kind of two different types of land use activity, but both speak to a central.

418
01:04:34.100 --> 01:04:38.510
William Ouimet: land clearing and land use practices that involve the clearing of land and might.

419
01:04:38.570 --> 01:04:44.630
William Ouimet: initiate changes on the land surface and whether that's erosion or sediment transport possible generation of legacy sediment.

420
01:04:45.380 --> 01:04:52.910
William Ouimet: Basically i'm really interested in teasing apart the ways in which people use the land and potentially the impacts of those processes.

421
01:04:53.510 --> 01:05:02.960
William Ouimet: On on the on the land forms but also the original incident transport properties are active in the landscape okay so we're going to focus on two different aspects of.

422
01:05:03.380 --> 01:05:11.450
William Ouimet: of land use, ladies activity and i'll talk about these in a second one of which we've been talking about, of course, which is stone walls and.

423
01:05:12.590 --> 01:05:25.700
William Ouimet: arranging the land for agricultural and pastoral activities Okay, but I just do want a quick quick note, this is a large project lots of students and collaborators, to knowledge, many parts this project or kind of kind of being kind of highlighted throughout the course of.

424
01:05:26.870 --> 01:05:29.810
William Ouimet: My talk today and you've also been hearing aspects of it.

425
01:05:30.350 --> 01:05:40.850
William Ouimet: With with thor's talking and she's talking and you'll also hear this with john's talk coming coming up, but too much to cover here really I don't really have time to talk talking in length about the the geo morphic.

426
01:05:41.300 --> 01:05:50.990
William Ouimet: kind of more process based questions that were asking with these data sets such as soil erosion, or soil soil modification some of that is presented in the session this afternoon.

427
01:05:51.470 --> 01:06:03.110
William Ouimet: Sam Dallas giving a talk on on some of the direct soil impacts of stone wall activity i'll relate to some of these these these areas, the same focus areas that you're seeing in the slides so far today.

428
01:06:03.590 --> 01:06:16.790
William Ouimet: So lots lots of work here so again, we have been talking about stone walls for quite some time now, and this this revelation really that lidar reveals under the forest canopy this this network of stone wall terrain, but also stonewall.

429
01:06:17.420 --> 01:06:20.390
William Ouimet: configuration that really highlights historic farmstead.

430
01:06:20.780 --> 01:06:29.060
William Ouimet: In the use of the land for various agricultural and pastoral activities okay now, the one that I also want to highlight here that we haven't talked have time to talk about.

431
01:06:29.450 --> 01:06:38.990
William Ouimet: That we're also using lidar to illuminate to illuminate is this notion of charcoal hearts, and this is the idea of harvesting forest hardwoods.

432
01:06:39.650 --> 01:06:50.390
William Ouimet: To end to convert them to charcoal for for a fuel source of various industries, chief among them the the iron making industry that was that was prevalent throughout much of the region.

433
01:06:51.320 --> 01:06:58.370
William Ouimet: You know, in the in the in the late 18th and 19th century, so this is a This is another example of a landscape feature.

434
01:06:58.880 --> 01:07:11.030
William Ouimet: finite in size, in this case the 99 to 15 meters that we're using a variety of topia graphic derivatives from lidar data to map out and to reconstruct this particular historical activity.

435
01:07:11.450 --> 01:07:18.950
William Ouimet: In the region, and it was pretty broad it was kind of illuminating for me, particularly for Connecticut to dis appreciate.

436
01:07:19.400 --> 01:07:24.620
William Ouimet: That, in addition to a significant component of length during that we know to be involved.

437
01:07:25.250 --> 01:07:34.880
William Ouimet: With stonewall activity agricultural and pastoral activities throughout the region that in some parts of the region charcoal was it was a also a very significant.

438
01:07:35.360 --> 01:07:40.910
William Ouimet: aspect of deforestation Okay, and so it's really the combine the combination of the two.

439
01:07:41.600 --> 01:07:47.960
William Ouimet: That that combined to give us this classic curve, that the date the Harvard forest groups have produced for many of the states in the in the region.

440
01:07:48.350 --> 01:07:59.510
William Ouimet: But all I find very interested in Connecticut we had we see like that peak charcoal been associated with peak iron industry is also associated with peak land clearing and Connecticut so it's the combination of the two.

441
01:07:59.960 --> 01:08:08.300
William Ouimet: Ladies activities that really gives us what you know this curve and so it's, so I think the appreciation for charcoal Horace in addition to stone walls.

442
01:08:08.540 --> 01:08:18.320
William Ouimet: are really illuminates the full the full distribution of landers activities that we're characterizing the MID 19th century and I, and I want to show some nice examples of that later on and.

443
01:08:19.160 --> 01:08:25.130
William Ouimet: So, of course, this all starts with lidar and we've been in Thor has introduced it in Jesus talked about using it for automated mapping.

444
01:08:25.550 --> 01:08:35.570
William Ouimet: But I do want to talk about some some summary aspects that we are also exploring to on to appreciate the limitations of lidar data, because it is not perfect, as we, as we know, we like to think it's perfect and it's a.

445
01:08:36.050 --> 01:08:44.300
William Ouimet: Great visualization of the land surface, but it does have its flaws, and so we are dealing with point clouds of data here that illuminate find scale features.

446
01:08:45.140 --> 01:08:53.690
William Ouimet: Of the historical archaeology period in the region, and this data is widespread, and so we should have really appreciate that that we're in an age where.

447
01:08:53.930 --> 01:09:02.930
William Ouimet: The state is rapidly coming online and we have the ability to to download and use this data for a variety of June morphic applications, chief among them as we're talking about today.

448
01:09:03.230 --> 01:09:18.440
William Ouimet: The application of land use mapping and understanding that historic period, not all lidar data sets are created equally and they have different differences in terms of their their points spacing DM DM resolution and this has of course implications for cultural resource mapping.

449
01:09:19.520 --> 01:09:27.470
William Ouimet: So here, you see a scene in Massachusetts which has two different minor qualities and you can see in the configuration of stone walls foundation.

450
01:09:27.770 --> 01:09:34.520
William Ouimet: And then you can convince yourself that at some scale both resolutions to meter in one meter are pretty good at picking up.

451
01:09:35.030 --> 01:09:39.380
William Ouimet: These very long gated stone wall features, but the smaller and less continuous the lidar.

452
01:09:39.890 --> 01:09:47.480
William Ouimet: The stonewall gets the harder it is to pick it up so even the find scale aspect of whether this is two or one meter lidar is enough to kind of.

453
01:09:47.690 --> 01:09:54.950
William Ouimet: Introduce error and into the mapping and we need, we do need to appreciate that, as we span out across the region that we have different quality align our data sets.

454
01:09:55.280 --> 01:10:03.950
William Ouimet: Of course, a lot of our DM are derived from point clouds point clouds are not perfect and ground classified points have a very significant variation.

455
01:10:04.550 --> 01:10:13.220
William Ouimet: And that we do want to keep this in mind, so here's an example of forest type I given rise to differences in point in ground point spacing which gives rise to.

456
01:10:13.580 --> 01:10:20.150
William Ouimet: A kind of a weirdly interpolated DM such that, when we derive our hill shades and when we derive our slope maps.

457
01:10:20.690 --> 01:10:27.170
William Ouimet: We have we have lower lower or higher quality part portions of our of our DM rosters okay.

458
01:10:27.470 --> 01:10:38.090
William Ouimet: And that's important so that, as we, as we tried to map individual features that are less than 10 meters and in size that we have points space and variation that will give rise to significant uncertainty.

459
01:10:38.690 --> 01:10:44.450
William Ouimet: In our ability to map out these features and there are other landscape features that we also want to keep keep our eye on.

460
01:10:44.870 --> 01:10:54.920
William Ouimet: Roads trails buildings ditches riverbanks These are all things that may be linear or or or they have small scale morphologic variations that might give rise to similar looking.

461
01:10:55.760 --> 01:11:00.290
William Ouimet: Land forms, as we, as we would like to map okay so there's lots of different ways to do this.

462
01:11:01.010 --> 01:11:09.590
William Ouimet: And i'll just going to kind of give a plug for the different ways that you might work with lidar so a lot of us tend to work with process DM tiles again when you get to this level.

463
01:11:10.100 --> 01:11:18.710
William Ouimet: Again you're not making your own DM so you do have to worry a little bit about ground point spacing, but this is probably the most common ways that we would go and manipulate our own our own our own.

464
01:11:19.130 --> 01:11:29.390
William Ouimet: lidar data sets, and the reason you might go download the data yourself, rather than streaming it which i'll discuss next is simply because you have more flexibility in terms of your topographic derivatives.

465
01:11:29.720 --> 01:11:33.110
William Ouimet: So I think I think Newton brought up last in this question like what about.

466
01:11:33.470 --> 01:11:42.980
William Ouimet: These more complicated multi directional hill shape that that's fine you can't do that with stream data, you have to download the data and there's a whole host of visualizations that you might choose to do.

467
01:11:43.340 --> 01:11:51.080
William Ouimet: What you need to download the data to be able to do that, perhaps more commonly what we're seeing throughout the region is this idea of streaming the data from variety of.

468
01:11:51.410 --> 01:12:01.460
William Ouimet: image and map services from state agencies around around the around the region it's pretty common because we can we have the ability to download and manipulate very large areas of lidar data.

469
01:12:02.630 --> 01:12:07.400
William Ouimet: into to be able to bring into your GIs into map quite significantly.

470
01:12:07.760 --> 01:12:17.510
William Ouimet: And there's lots of different places to download this data and then it's not too hard to arrive at the picture here where we're pretty quickly in your GIs you can have a full stream of all.

471
01:12:17.810 --> 01:12:25.490
William Ouimet: Entire regional lidar data set of both the hill shades and the slope maps, so I really this is this is this is at your fingertips, if you so choose to.

472
01:12:25.910 --> 01:12:34.250
William Ouimet: want to be able to bring it in into map map it okay so that's what we've been that's where the data is coming from some of the important implications for what it is.

473
01:12:34.670 --> 01:12:43.370
William Ouimet: We have been mapping these features quite extensively, along with others in the region to really understand the distribution of these features and their implied.

474
01:12:43.940 --> 01:12:52.250
William Ouimet: Land Use activity and therefore forestry historic for us reconstructions, and so this has been discussed, but we use a variety of oh perfect derivatives.

475
01:12:53.060 --> 01:12:59.090
William Ouimet: predominantly we hand digitize but we're getting into this world of of doing automated mapping and other other.

476
01:12:59.510 --> 01:13:11.570
William Ouimet: Other remote sensing techniques okay So where are we, where are we in terms of our regional mapping so here's just an example of some of the yukon groups, a regional mapping mapping.

477
01:13:13.430 --> 01:13:22.520
William Ouimet: The extent of our ladder mapping, so my group tends to focus at the town level and i'll talk about that why why that is in a second because we like to kind of complete one town.

478
01:13:22.880 --> 01:13:30.170
William Ouimet: have some sense of error and knowledge of completeness of one one particular area and so that way we can compare that data sets to other.

479
01:13:30.680 --> 01:13:42.110
William Ouimet: to other aspects of historical historical census data and so another great example this rick, I believe, is in the audience, so that the new Hampshire group has been has been mapping quite extensively.

480
01:13:43.190 --> 01:13:50.090
William Ouimet: lidar sorry stone walls and lidar data sets it for the whole the whole state of new Hampshire and they have a large majority of the state.

481
01:13:50.690 --> 01:13:55.280
William Ouimet: mapped at least this kind of initial level which, which admittedly has some issues to do with.

482
01:13:55.580 --> 01:14:05.630
William Ouimet: lidar quality and parts in parts of the state, but we are, we are we're really building this regional data set and it quite extensive way and it's going to be very powerful to kind of think about spatial spatial patterns.

483
01:14:06.170 --> 01:14:15.110
William Ouimet: So one thing that we can do, certainly is when you look at the work of the town love, will you can look at these town distributions of stone walls, the total area, the total wall length.

484
01:14:15.590 --> 01:14:22.610
William Ouimet: Think about this notion of stone wall density and there are significant variations, as you go north of the examples in Vermont which I don't provide here.

485
01:14:22.910 --> 01:14:33.380
William Ouimet: Have a much lower density, so you can see, this kind of history of of occupation, but also, you know, population growth, at a time and the 19th century that corresponded with this type of.

486
01:14:33.830 --> 01:14:44.990
William Ouimet: Forest agriculture in clearing of tilt tilt domain into into stone walls OK, and then the other examples in Massachusetts showing some of these results.

487
01:14:46.670 --> 01:14:55.880
William Ouimet: If you switch gears and just look about charcoal hearts were doing the same thing it's it's a regional question really for charcoal horse and we and it's a large area that we're trying to consider because we get up we find.

488
01:14:56.510 --> 01:15:01.400
William Ouimet: Information about these things all over the region that we go looking for them in the lidar, and this is very.

489
01:15:01.880 --> 01:15:06.110
William Ouimet: It is not extensive in terms of our our completeness of the data, but you can see certain.

490
01:15:06.440 --> 01:15:15.980
William Ouimet: patterns emerging were certainly in western Massachusetts Eastern New York and Northwest Connecticut we have the highest density of this of this particular type of activity.

491
01:15:16.340 --> 01:15:24.260
William Ouimet: And we're beginning to tease that out with our regional mapping we do some really nice kind of close up studies, this is Northwest Connecticut where we can really tie.

492
01:15:24.950 --> 01:15:30.290
William Ouimet: All of this activity of charcoal into a very prolific and productive period of iron.

493
01:15:30.860 --> 01:15:38.870
William Ouimet: iron production in the Northwest part of the state, and this is where we we've mapped the majority of our charcoal hearts in over 20,000 easily.

494
01:15:39.590 --> 01:15:47.090
William Ouimet: But we can we have done some mapping more regionally to this is, this is the the best example that I have a complete charcoal mapping.

495
01:15:47.540 --> 01:15:53.780
William Ouimet: Where we have some sense of the space of distribution around the region in relation to this particular type of land use activity.

496
01:15:54.650 --> 01:16:03.170
William Ouimet: In the 90s in 19th century, and you can see certain certain patterns here emerging that had to do with that iron cord or really the core of the appalachians.

497
01:16:03.680 --> 01:16:11.240
William Ouimet: Where you're finding some of this is this kind of dirty iron ore and processing it throughout the 19th century okay.

498
01:16:11.990 --> 01:16:15.800
William Ouimet: One thing, as we, as we work to expand this work into other areas.

499
01:16:16.100 --> 01:16:25.490
William Ouimet: Of the region it's important to keep in mind that that what we have been talking about is what I would call this more temporary he'll slope charcoal mount or charcoal hearth process.

500
01:16:25.910 --> 01:16:31.940
William Ouimet: Where that was not the same, and other parts of the region, so an example is in maine and Vermont and New York.

501
01:16:32.630 --> 01:16:39.410
William Ouimet: The idea was to produce charcoal and more permanent stone and brick kills and so here, the idea is that these these.

502
01:16:40.220 --> 01:16:46.580
William Ouimet: These callers are still clearing lots of land to produce charcoal and so there's a lot of deforestation implied.

503
01:16:47.150 --> 01:16:52.730
William Ouimet: But they're concentrating the formation of the charcoal so it's it's harder to reconstruct from the individual.

504
01:16:53.300 --> 01:16:57.830
William Ouimet: And the individual charcoal horse where exactly the force were cleared for these these particular activities.

505
01:16:58.610 --> 01:17:08.690
William Ouimet: This is also there's always some important limitations that aren't darkness are that are just to do with with hand digitization some areas walls are removed through time through various.

506
01:17:09.290 --> 01:17:17.030
William Ouimet: agricultural practices, continuing to today, but also through other urban and suburban development, so I we do see a lot of our maps, as always, a minimum.

507
01:17:17.450 --> 01:17:21.350
William Ouimet: And I noticed Thor talked about possible expansion of stone walls.

508
01:17:22.190 --> 01:17:30.500
William Ouimet: as defined in the stone domain expanding in the late 20th century that's probably true, but we are still seeing the removal of a lot of these older historic historic walls.

509
01:17:31.070 --> 01:17:38.840
William Ouimet: here's just a nice example of when you look close to 1934 imagery you see walls in place and 34 is that are no longer there.

510
01:17:39.320 --> 01:17:45.800
William Ouimet: In the modern ortho photos or or in the lidar so clearly large scale loss of stone walls in parts of the landscape.

511
01:17:46.280 --> 01:17:53.360
William Ouimet: So we are doing work to kind of reconstruct historic land use activity, so this is the idea of how do you use a wall that's mapped.

512
01:17:53.960 --> 01:18:02.210
William Ouimet: With the lidar to actually extrapolate to area, do you use like a political boundaries kind of contained within individual stone walls, do you buffer.

513
01:18:02.450 --> 01:18:12.050
William Ouimet: Or do you some kind of Dudes to kind of density analysis and that has implications for just how you actually go from individual walls to actual area amount amount of area cleared.

514
01:18:12.410 --> 01:18:18.350
William Ouimet: The same thing with charcoal horses that you know so a certain amount of acreage of tree a tree.

515
01:18:18.800 --> 01:18:28.310
William Ouimet: A tree forest for forest trees and therefore court would cottage is implied with each individual charcoal so you can play these games about how to reconstruct exactly how much lengthening.

516
01:18:29.090 --> 01:18:36.560
William Ouimet: occurred for each site and then what we've been doing to really bring this all together is think about the spatial distribution of the both.

517
01:18:36.830 --> 01:18:44.060
William Ouimet: Both types and how we can use both land use types to really reconstruct the total amount of land clearing and forest.

518
01:18:44.660 --> 01:18:54.890
William Ouimet: Land Use plan cover change in the region so here's a here's a three town focus area in North Northwest Connecticut where we've mapped stone walls and you can see the kind of density depicted here.

519
01:18:56.780 --> 01:19:02.060
William Ouimet: depicted here in the black and green and the same thing for the charcoal hearts, and when you combine the two.

520
01:19:03.320 --> 01:19:05.270
robertmthorson: Three minutes left yes awesome i'm.

521
01:19:05.390 --> 01:19:09.620
William Ouimet: just getting there, and you can combine the two, and you can you can see the full mosaic.

522
01:19:09.950 --> 01:19:17.120
William Ouimet: And the combination of the total amount of land clearing in the 19th century associated with both both the practices and then, of course.

523
01:19:17.480 --> 01:19:25.730
William Ouimet: We know this is areas for us today, and so we know this area is somewhere along this curve, and so what I would argue in the lower the lower left portion of this.

524
01:19:27.020 --> 01:19:39.440
William Ouimet: map here is that is a combined combined historic land use plan cover map really from both the land use practices and really this is, this is an important update on an old theme, the old theme is that we know that the upper left curve exists.

525
01:19:39.800 --> 01:19:52.430
William Ouimet: But having the ability to map the individual landings practices is really you know, is really important, because, because the combine they give us the cumulative total land impacted rather than just one census level.

526
01:19:53.390 --> 01:20:02.180
William Ouimet: time period of what's happening in terms of the total percent but we get up to 90 95% land clearing total affected which which exceeds the historical estimates.

527
01:20:02.570 --> 01:20:12.110
William Ouimet: Again that's a cumulative total and then for the first time with mapping in lidar we're seeing the detailed special arrangement of these multiple entities practices and they're in their intensity.

528
01:20:12.950 --> 01:20:22.460
William Ouimet: we've done a little bit of work to kind of think about comparisons historical data and I don't quite have time to go into this, but you can we can go into the archival records and really compare this to.

529
01:20:23.630 --> 01:20:33.260
William Ouimet: unimproved and improved land these these these classic notions of of manufacturing census of how we really reconstruct land clearing from from historical documents.

530
01:20:33.950 --> 01:20:38.030
William Ouimet: Not another focus area and williamstown doing the same thing.

531
01:20:38.480 --> 01:20:47.900
William Ouimet: The point here, they just emphasize is just how prolific the stonewalling was within the valleys, where people were living and the charcoal was really up in the more rugged more rugged.

532
01:20:48.590 --> 01:20:53.750
William Ouimet: terrain and we can do that in Connecticut and then we can we can actually tease apart the relationship between relief and slow.

533
01:20:54.230 --> 01:21:02.330
William Ouimet: And really kind of kind of you know, understanding, where this this activity and the feedbacks that ultimately are occurring, so we have glacial materials land use type.

534
01:21:02.690 --> 01:21:09.380
William Ouimet: Soil erosion potential are all kind of in a feedback for ultimately determine where these two different land use policies are occurring.

535
01:21:09.890 --> 01:21:14.240
William Ouimet: And so there's lots of implications and ongoing work don't have time to cover it all now, this is, this is my.

536
01:21:14.750 --> 01:21:21.950
William Ouimet: Good in slide are really you know, we have the reconstruction of these outlandish activities, provide the foundation for all future studies.

537
01:21:22.340 --> 01:21:33.800
William Ouimet: aiming to understand all kinds of impacts, whether it's soils for structures, ecology emotion and seven transport legacy sediment that's all comes from this potential of knowing where this these activities were.

538
01:21:34.250 --> 01:21:48.650
William Ouimet: And I will go ahead and there was just a plug for our our stone wall in our charcoal heart public matters kind of similar to the new Hampshire group anyone can go online and contribute to these to these resources of mapping these features around the region, thank you very much.

539
01:21:49.250 --> 01:21:49.880
robertmthorson: Thank you well.

540
01:21:51.110 --> 01:21:51.830
robertmthorson: Perfect timing.

541
01:21:53.000 --> 01:21:54.200
robertmthorson: We have time for one question.

542
01:21:55.970 --> 01:21:57.080
William Ouimet: anything in the chat or.

543
01:21:58.400 --> 01:21:58.880
robertmthorson: Looking.

544
01:22:00.140 --> 01:22:01.310
robertmthorson: I haven't seen anything in the chat.

545
01:22:02.240 --> 01:22:03.950
robertmthorson: Right at least not a public one.

546
01:22:06.170 --> 01:22:08.090
William Ouimet: i'm just i'm looking in right now to.

547
01:22:11.180 --> 01:22:13.130
robertmthorson: What we're right on time to move for the next one.

548
01:22:14.360 --> 01:22:15.950
William Ouimet: Next, one is a poster so we just want to.

549
01:22:16.220 --> 01:22:19.070
robertmthorson: Get I know it's scheduled for a 10 minute session.

550
01:22:20.090 --> 01:22:21.230
William Ouimet: As well yep.

551
01:22:22.250 --> 01:22:27.800
robertmthorson: And i'm not quite sure what the methodology was on that I actually went through the training and they were not clear.

552
01:22:28.280 --> 01:22:29.720
William Ouimet: Yes, it's a Johnny you here.

553
01:22:32.900 --> 01:22:34.490
Jon: Yes, I am good.

554
01:22:34.610 --> 01:22:40.730
William Ouimet: So in the in the training, will you told the player video or was the video only for viewing after before the fact.

555
01:22:43.910 --> 01:22:45.530
Jon: i'm sorry you say only time is it to me.

556
01:22:45.590 --> 01:22:46.490
robertmthorson: During your training.

557
01:22:46.880 --> 01:22:53.900
William Ouimet: You recorded a video do you want to do we want to play that video right now, or do we want just want to do a question answer based on that, on the poster.

558
01:22:54.920 --> 01:23:00.500
Jon: Oh either way to me, I thought it was just I thought it would play on it all the time I was exactly what was going and going.

559
01:23:00.680 --> 01:23:11.330
robertmthorson: So well it's scheduled for 10 minutes, so if you could give the presentation we're fine with that if you don't have a have a problem with it because we perhaps can watch the video later if we need to so.

560
01:23:11.390 --> 01:23:15.260
William Ouimet: it's or can we we go ahead and click on a poster in the in the session link.

561
01:23:16.130 --> 01:23:17.150
robertmthorson: i'm gonna have to find it.

562
01:23:18.020 --> 01:23:19.070
William Ouimet: I haven't here to.

563
01:23:19.220 --> 01:23:26.930
William Ouimet: Okay i'm pretty sure we are supposed to play it yeah again, this is a kind of a combo oral poster so we're kind of.

564
01:23:28.430 --> 01:23:35.780
robertmthorson: done, they said they'd have somebody detailed to this site at this time to make that transfer transfer that's what they said during my.

565
01:23:36.260 --> 01:23:39.680
William Ouimet: Well, let me just go hold on I can I can broadcast this.

566
01:23:41.330 --> 01:23:51.740
William Ouimet: From my computer so i'm going to go ahead and play the poster from my computer Okay, and then that way i'll share my screen and then we'll have a question answer period for john after after his yes okay.

567
01:23:52.190 --> 01:24:00.200
robertmthorson: I would remind that if if technology goes awry you know, this is the last talk of the session, and we hope to have a little discussion after the end of it, so thank you.

568
01:24:00.980 --> 01:24:01.700
William Ouimet: Okay, so.

569
01:24:04.010 --> 01:24:05.090
William Ouimet: Or what do you guys seen right now.

570
01:24:05.810 --> 01:24:07.310
robertmthorson: i'm seeing a poster.

571
01:24:09.980 --> 01:24:10.400
William Ouimet: OK.

572
01:24:11.840 --> 01:24:14.120
robertmthorson: And now i'm seeing mostly john and the bottom there.

573
01:24:14.810 --> 01:24:17.810
William Ouimet: But I think i'm supposed to play it like this, or.

574
01:24:24.380 --> 01:24:25.850
William Ouimet: i'm going to give it a shot all right.

575
01:24:30.770 --> 01:24:36.680
William Ouimet: hi everyone, my name is Jonathan leather and geography graduate student at the University of Connecticut studying with Dr William.

576
01:24:37.490 --> 01:24:46.640
William Ouimet: Now, as you may be familiar for previous talks stone was our focus Fridays region that into document, the late 19th century agricultural practices.

577
01:24:47.480 --> 01:24:59.270
William Ouimet: quantifying past human impacts is important for studying with someone call the advocacy and the human environment interaction associated in order to distinguish markers of human induced change though stone loss can be you.

578
01:25:00.590 --> 01:25:08.120
William Ouimet: know, one of the biggest obstacles from backup stone walls is landscape variation which seven things convinced stone walls and the lighter.

579
01:25:08.720 --> 01:25:15.560
William Ouimet: natural features that just buildings and bridges and interesting features such as fences roads and even swimming pools.

580
01:25:16.460 --> 01:25:25.970
William Ouimet: Now well why is this important well stonewall is preventing perpetual margins property boundaries, past and present day boundaries between various crops and so i'm.

581
01:25:26.840 --> 01:25:40.670
William Ouimet: Standing there and variation with regard to mathematics is valuable for those studying religion and doing geospatial in the search and feel ecology social science anthropology history and so on.

582
01:25:42.110 --> 01:25:49.940
William Ouimet: Now, but this project we had for GIs users have a mapping walls and each of them was getting the exact same data set for five regions of interest.

583
01:25:51.290 --> 01:25:57.530
William Ouimet: There were given one minute lighter based roadmap to one meter lighter explosive maps with different solar estimates.

584
01:25:58.430 --> 01:26:05.510
William Ouimet: Digital innovation model and aerial photograph retrieving now research question one, given the same input data.

585
01:26:06.020 --> 01:26:18.410
William Ouimet: hub get a sense of manually device was a very distinct individual matters and having results very functional landscape question to I will give us mentally filters balls compare with field verified stonewall datasets.

586
01:26:20.300 --> 01:26:27.980
William Ouimet: permit is manual installation was done using editing tools in our region of interest three was field mapping stonewall verified.

587
01:26:28.880 --> 01:26:40.700
William Ouimet: Analysis analysis was done using our toolbox and i'll explain in this three window panel our window on the left, shows the power lines that were digitized as an example.

588
01:26:41.270 --> 01:26:51.020
William Ouimet: Now the status as a secondary, Sir, and we wanted to restaurants it's my calculations these here, so we did a three minute buffer and we restaurants in line and be classified it.

589
01:26:52.250 --> 01:26:55.400
William Ouimet: From there to the window on the right multiple number one.

590
01:26:55.820 --> 01:27:08.240
William Ouimet: We actually added each line on top of each other, as you can see how many users agree on what the walls and what we're about so, for example, if you see other the other line that is one user, so, while in digitized it.

591
01:27:09.050 --> 01:27:22.160
William Ouimet: gets warmer let's say the purple companies all full users every get that with a stone wall now study area we had five regions of interests that are regions of interest that are generally in eastern tag.

592
01:27:24.080 --> 01:27:25.460
William Ouimet: results for Question one.

593
01:27:27.050 --> 01:27:32.450
William Ouimet: As an example, in the napa auto region of interest one user one you'll see in the table.

594
01:27:33.950 --> 01:27:48.860
William Ouimet: 7.3% in agreement with the group now this percent agreement with the group names for these didn't times wall that user, one day, the other three users digitize was 7070 70.3% in agreement with that.

595
01:27:51.050 --> 01:27:58.700
William Ouimet: So if you go through each region of interest to see that the more yellow lines, there are, that means the more ambiguous that terrain is.

596
01:27:59.390 --> 01:28:13.520
William Ouimet: The more purple lines, the more not ambiguous so to smooth the Rings have more purple lines they're easier to see, and the more vigorous he was terminated by suburban environment clear fields and filtering have more yellow lines.

597
01:28:15.020 --> 01:28:21.680
William Ouimet: Now results for question to this, the field comparison and just to take an excerpt here again, if you look at the table.

598
01:28:22.340 --> 01:28:31.010
William Ouimet: User three represents an individual who consistently matt stone walls segments that exist in the field but generally under back and see by the long leg.

599
01:28:31.700 --> 01:28:38.510
William Ouimet: And didn't digitize many walls that do exist in the field, meanwhile, user to represent individual generally all over the map.

600
01:28:39.230 --> 01:28:50.240
William Ouimet: The inability of any individuals that have greater than 85% of what we're seeing the field may indicate that there are subtle walls and stone lines and the deal that are not well expressed in later some key takeaways that.

601
01:28:50.600 --> 01:28:57.110
William Ouimet: overall agreement between all our allies 8% which means to be mostly me on what was in the field that are actually stone walls.

602
01:28:58.580 --> 01:29:03.020
William Ouimet: Smooth for the ground truth some statistics were deal that.

603
01:29:04.460 --> 01:29:14.810
William Ouimet: We agreed with the field 94% time but occasionally miss balls because of the only remember those 78% and for the field work and analysis can be done with different landscape types to.

604
01:29:15.470 --> 01:29:23.450
William Ouimet: Better the data sets and analysis Thank you so much for listening to myself and I welcome any questions comments or suggestions, thank you.

605
01:29:27.920 --> 01:29:30.380
William Ouimet: Okay, so I think that so john is here.

606
01:29:30.470 --> 01:29:30.890
William Ouimet: hey john.

607
01:29:33.470 --> 01:29:35.870
William Ouimet: And so I think that there's a period of time.

608
01:29:39.350 --> 01:29:51.380
William Ouimet: That we can ask any questions, but I think let's get let's give at least john a couple minutes here of questions and then, and then I think for said that we cannot open it up to more broader questions for the whole group, if there are any at this time.

609
01:29:51.590 --> 01:30:01.040
robertmthorson: And i'll just say I don't know when GSA kicks us off, but we're supposed to stay on schedule and to get to 935, but I hope we have time for questions and.

610
01:30:03.080 --> 01:30:08.420
William Ouimet: Alright, so any questions for john and they the user kind of interpretation, the cognitive experiment that he did.

611
01:30:11.540 --> 01:30:19.490
robertmthorson: Back in the old days they used to test people for aerial photo interpretation and they found great variability and in terms of.

612
01:30:20.060 --> 01:30:36.470
robertmthorson: people's ability to spot known things on the ground, it had to do with military Defense during World War Two, and the same sort of internal psychology or neurology must be involved with individuals and I wondered if you've looked into that research, some people truly better than others.

613
01:30:39.200 --> 01:30:47.990
Jon: I haven't looked into that research but i'll be googling googling and certainly for a little bit of background and overview I did see an another couple of papers that we've done some decades back about.

614
01:30:49.370 --> 01:30:55.670
Jon: About not necessarily image image interpretation, with regard to manual digitization, but it was more for geologic features yeah.

615
01:30:56.720 --> 01:31:06.830
William Ouimet: I think that the example that I know best, is that ios for the agu journal, I know, within the last 15 years did a study of like 100 people looking at a seismic line.

616
01:31:07.310 --> 01:31:19.430
William Ouimet: and doing a seismic interpretation and just how varied yeah the interpretations were, and that was something that you would argue that it's a significant Max expertise is is is needed just to even begin.

617
01:31:19.700 --> 01:31:20.390
robertmthorson: yeah right.

618
01:31:20.450 --> 01:31:27.530
William Ouimet: What I love about this one is that is that you would hope that anybody can quickly look at a lidar image and have some sense of what a stone wall is.

619
01:31:28.880 --> 01:31:34.100
William Ouimet: If you trust them to map what kind of what kind of error error is involved, I I.

620
01:31:34.250 --> 01:31:46.940
robertmthorson: I taught air photo interpretation three or four times you know the standard geology interpretation land forms and I found sort of anecdotal wide variability with the many, many students that I had yeah it's fascinating subject.

621
01:31:47.300 --> 01:31:52.970
William Ouimet: john so, can you just update the group on where other field mapping you're doing to do more comparisons.

622
01:31:53.600 --> 01:32:00.740
Jon: yeah well, we did region of interest three, which is a smooth terrain and mask over also did some interest.

623
01:32:01.580 --> 01:32:08.660
Jon: To which was another smooth terrain at the conference that was previously brought up and will be also doing region of interest one.

624
01:32:09.020 --> 01:32:21.230
Jon: which was the healing environment in the east village and aspirate section in the Highlands hope to be able to get out to the suburban areas like no stone unturned and whatnot, but a lot of private property, so we have to figure something out.

625
01:32:27.200 --> 01:32:32.480
robertmthorson: Well, if a final question, perhaps, where we are now at 935.

626
01:32:41.030 --> 01:32:41.210
well.

627
01:32:42.830 --> 01:32:49.640
robertmthorson: You ready to go, will you want to have a have a final say here i'm just thinking that we are right on schedule.

628
01:32:50.240 --> 01:33:04.940
robertmthorson: I would just like to say that that I very much appreciate it everybody in the audience for showing up and showing their interest and, of course, all of the presenters at this session, and my you know co coordinator will we met here you want the final word well.

629
01:33:05.600 --> 01:33:09.110
William Ouimet: Oh no but no thanks to see everyone's interest we do have a question here.

630
01:33:09.680 --> 01:33:20.690
William Ouimet: Okay, from Ellen Chamberlain general question for stonewall people do you see these degrade falling and predictable the fusion model, or are they too young to see any shape degradation.

631
01:33:22.520 --> 01:33:25.250
robertmthorson: Are you are you referring to the form of the wall itself.

632
01:33:27.470 --> 01:33:28.400
William Ouimet: Yes, yeah.

633
01:33:28.490 --> 01:33:35.600
robertmthorson: Although they are granular granular I haven't seen it, I think a diffusion model could be applied to the but not the traditional GEO morphic one.

634
01:33:36.230 --> 01:33:44.840
robertmthorson: Because it has more to do with the substrate below the wall and the way the Wall was built than the actual particle size, I mean you know go ahead, well.

635
01:33:45.170 --> 01:33:55.610
William Ouimet: Oh i'll just going to add that I mean when you're when you spend a lot of time in the field, away from roads, because, of course, there can be modifications do to do to people slightly building that building the backup but.

636
01:33:56.090 --> 01:34:08.930
William Ouimet: let's assume you're away from any any any modifications, the biggest thing or trees falling on top of them and knocking down, which is added in an aggregate sense of the whole segment could maybe model be modeled like diffusion.

637
01:34:10.040 --> 01:34:15.920
William Ouimet: yeah the not necessarily like the toppling down from a vertical like a linear vertical notch.

638
01:34:16.940 --> 01:34:24.650
robertmthorson: They don't go trap avoid you know too low mount a uniform, you know, distribution and there's a lot of slope asymmetry in it as well.

639
01:34:24.950 --> 01:34:33.170
robertmthorson: And and absolutely trees are the worst and if a tree roots in a wall or near a wall it doesn't degrade the form it just blows the whole thing out.

640
01:34:33.590 --> 01:34:49.130
robertmthorson: And if a tree falls on a wall with a trunk it takes a bite out of it, you know, so this just I think will is right that, if you look at the length of the wall, we could probably come up with good statistics about relative preservation and that might fall fall on a model but.

641
01:34:49.160 --> 01:34:58.820
William Ouimet: But I would add, though, is that I think from Thor from the classifications studies that we're doing you need you can't assume they were all perfectly stack so when you walk up to a.

642
01:34:59.900 --> 01:35:06.470
William Ouimet: stone wall and you just see kind of some crumbling behavior you can't assume that it was well stack To begin with, yes.

643
01:35:07.040 --> 01:35:21.230
William Ouimet: that's part of the classification and, of course, the chronology is such that it can it can crumble and fall apart through time, but it also doesn't necessarily have to so so that, so I think you do want to be careful about just interpreting any wall that is that looks crumbled.

644
01:35:21.620 --> 01:35:23.360
William Ouimet: up to them to have been stacked.

645
01:35:23.360 --> 01:35:29.810
William Ouimet: Because otherwise if they're all fallen over then we we have earthquakes right or or or something that actually it might be triggering all that.

646
01:35:30.050 --> 01:35:39.230
robertmthorson: And that's the very definition of a band, you know, a band is structuralist and you know that's the end product of collapse if you want, but it could also be that that's the way it was at the beginning.

647
01:35:39.710 --> 01:35:46.700
robertmthorson: I mean when I first started getting into this business, they were talking about all these toppled walls, you know as a nice literary term, and I wanted to know what that meant.

648
01:35:47.210 --> 01:35:57.500
robertmthorson: In our field designation we're differentiating a collapse from a top you know and that's basically the structure of how it's built to begin with, and they can be seen right right.

649
01:36:02.180 --> 01:36:02.480
robertmthorson: All right.

650
01:36:03.170 --> 01:36:08.180
William Ouimet: Two more questions two more questions here has wall density versus official mapping it's been studied.

651
01:36:08.810 --> 01:36:19.430
William Ouimet: And until but in places where the old timers brought the stones great absolutely so so in our one of our publications woody we actually the early the Johnson, and we met 2016.

652
01:36:19.910 --> 01:36:29.930
William Ouimet: kind of compares the stone wall density to superficial mapping units, but and also the orientation of the of the walls, a relative to the English orientation.

653
01:36:30.380 --> 01:36:35.300
William Ouimet: And there's absolutely good agreement, but there's also interesting variation and the variation could be because.

654
01:36:35.900 --> 01:36:44.630
William Ouimet: The mapping is not as good as you think it is right, so, so we have been looking specifically at that at that relationship, but we have what we haven't done yet is like.

655
01:36:45.410 --> 01:36:49.340
William Ouimet: invested the time to map, a town in the middle of the Connecticut river Valley, which we know they're.

656
01:36:49.730 --> 01:36:55.130
William Ouimet: There won't there shouldn't be that many stone walls, but there probably will be linear features when we do that.

657
01:36:55.550 --> 01:37:01.250
William Ouimet: A stone wall map from the same, I mean it's gonna be hard to unbiased yourself to the idea that there shouldn't be stone walls.

658
01:37:01.880 --> 01:37:16.910
William Ouimet: In the true kind of glacier like hitchcock terrain, so that that's something that we will be pursuing but we I I can send you some of the some of the data and the graphs that we have from that 2016 paper we published in the anthropocene.

659
01:37:17.780 --> 01:37:18.860
Woody Thompson: Great thanks a lot well.

660
01:37:19.490 --> 01:37:25.820
robertmthorson: there's, also the case of the of the farmer who who moved his wife down in the 19th century, from the Highlands to the lovin.

661
01:37:26.120 --> 01:37:29.420
robertmthorson: And she was unhappy so he built her a stone wall down there on.

662
01:37:29.570 --> 01:37:37.340
robertmthorson: This floor of glacial lake hitchcock you know, to make her happy, you know Those are the kinds of anecdotal things that are are going to constantly confuse what we're doing.

663
01:37:38.450 --> 01:37:44.420
William Ouimet: We have a point by Linda that stone walls ever moved by people absolutely and then we have a question from Susan.

664
01:37:44.810 --> 01:37:58.730
William Ouimet: Have you noticed the orientation of a stone was laid out historically by using magnetic north to vary, with the wandering or magnetic North over decades and centuries so there's actually a paper that's been published recently I do physical paper you know you know the author's name.

665
01:37:58.880 --> 01:38:00.890
robertmthorson: I can't recall it right now but yeah so very good.

666
01:38:00.890 --> 01:38:13.850
robertmthorson: paper and it searchable yep and it was very, very definitive that they're the declination has changed since the walls were in place, you know that's the kind of thing that we can do fairly readily down the road so with all this excellent mapping we're doing by lidar.

667
01:38:13.940 --> 01:38:26.330
William Ouimet: Right and he used lidar maps to to do the modern orientation yeah also, I believe he he needed really high precise historical maps which actually had the orientation.

668
01:38:26.840 --> 01:38:37.550
William Ouimet: Of the wall written on the map, when it was laid so there's a lot of uncertainty about exact age of the wall, I think, when I reviewed informally some of that work that was one of the questions like what.

669
01:38:38.150 --> 01:38:45.440
William Ouimet: What are you assuming about the age of the wall, when you when you compare it against that map and the light, because the light or, of course, is recent recently flown and so.

670
01:38:45.680 --> 01:38:54.470
robertmthorson: complicated and, and so I haven't gotten that, in other words a good survey, or would do their own astronomical you know calibrations of North.

671
01:38:55.100 --> 01:39:03.350
robertmthorson: But others would just simply adopt the prevailing standard and so there's a lot of a lot of details in that study, but I think it's potentially useful yep.

672
01:39:05.390 --> 01:39:06.260
William Ouimet: Okay, so.

673
01:39:08.090 --> 01:39:09.860
William Ouimet: We think we could call it that.

674
01:39:10.160 --> 01:39:15.470
robertmthorson: i'm fine, I think, because people do have other talks to go to and GSA likes us to stay on time.

675
01:39:15.860 --> 01:39:17.300
William Ouimet: awesome well I get.

676
01:39:17.330 --> 01:39:24.710
robertmthorson: That thanks everybody thanks will thanks to all of us who are really, really, contributing to this yeah.

677
01:39:24.890 --> 01:39:30.860
William Ouimet: I really, really appreciate Thor and thanks for all your pioneering work on this and it's it's really been fun to see how this is all evolved.

678
01:39:30.920 --> 01:39:31.910
William Ouimet: yeah I just.

679
01:39:32.810 --> 01:39:37.100
robertmthorson: I just want that classification published it's going to happen awesome all right bye bye.

680
01:39:37.580 --> 01:39:39.920
William Ouimet: Have a you have a great GSA northeast yes everyone.

681
01:39:40.100 --> 01:39:43.670
robertmthorson: To everyone and the yukon types will see you tonight yeah.

682
01:39:51.110 --> 01:39:52.070
robertmthorson: And will, are you there.

683
01:39:55.730 --> 01:39:56.330
William Ouimet: Can you hear me.

684
01:39:56.660 --> 01:40:02.690
robertmthorson: yeah can you join my room when you're done, are you going to stay and go off on to some other things, I just wanted to be brief, at your leisure.

685
01:40:03.050 --> 01:40:08.180
William Ouimet: sure if it's being recorded now, so we should probably log off.

686
01:40:08.270 --> 01:40:15.200
William Ouimet: or i'll Come on, I have to leave i'm in my office, I have to leave in about 10 minutes or so so let's do a quick deep do a quick debrief and.

687
01:40:15.200 --> 01:40:15.350
I.

688
01:40:16.400 --> 01:40:16.550
William Ouimet: Do.

689
01:40:18.710 --> 01:40:21.000
William Ouimet: yeah i'll see you i'll see in your room bye.

William Ouimet: yeah i'll see you i'll see in your room bye.