North-Central Section - 38th Annual Meeting (April 1–2, 2004)

Paper No. 3
Presentation Time: 8:40 AM

A RASTER-BASED GIS MODEL OF BASE-LEVEL INDUCED CHANNEL ENTRENCHMENT


HEINE, Reuben A., Environmental Resources and Policy, Southern Ilinois Univ, 1259 Lincoln Drive, Mail Code 4324, Carbondale, IL 62901, rheine@siu.edu

A new GIS model is presented for estimating channel entrenchment of tributary streams in response to base level change on the trunk river and is designed to be applicable over anthropogenic time scales. The entrenchment model is based on an iterative process of successive steps of channel degradation starting with the highest-order channels and continuing to the headwater streams with each step setting the base level for lower-order streams. In the simplest case, the methodology uses only a digital terrain model (DTM) and knowledge of the drop in base level to estimate and map the spatial extent and depth of channel entrenchment. The methodology also allows for the input of natural or created hard-points which can inhibit a migrating knick point, such as bedrock controls or grade control structures or dams.

The raster-based entrenchment methodology is demonstrated on the Upper Cache River in Southern Illinois. The Cache River experienced a dramatic drop in base level of 12.2 meters in 1915 following the construction of the Post Creek Cutoff which shortened the river by more than 64 km. The Cache River and many of its tributaries have incised into their beds and subsequently eroded their banks since the cutoff. Using a time series of existing longitudinal profiles, it was determined that bedrock hard-points have dominated this rejuvenating fluvial system, and thus the model required the input of geologic controls to incision. Three input data scenarios were crafted, and their results were compared to field data. A significant improvement in the modeling results was obtained by including the locations of field-measured hard points but this improvement came at a considerable cost of time and effort. The second data-input scenario, which utilized hydrologic data from a long-term stream gage, produced the best R^2 to time ratio, predicting 95% of the variation with only 2.7 hours of data creation time.