TREE RINGS, DISCHARGE, AND FRACTAL FILTERS

Two piñon (Pinus edulus) tree-ring chronologies developed on each of three substrates (sandstone, shale, and alluvial fan deposits) in southern Utah for the period 1702 to 1997 demonstrate that geologic substrate affects dendrochronologic streamflow reconstructions. Chronologies from alluvial fan deposits explain the most variance of cool-season (October 1 to May 31) flow with an adjusted coefficient of determination (Ra2) equal to 0.59. Chronologies from sandstone deposits account for 52 percent of the variance, while those on shale deposits account for 45 percent. Correlation coefficients among the three substrates are significantly different at the 95% confidence level. The different substrate response is attributed to varying amounts of clay in each substrate affecting infiltration and available water for tree growth. Low summer reconstruction values are attributed to spatially and temporally variable thunderstorms.

The fractal parameters (fractal dimension and Hurst exponent), calculated using the roughness-length method, describe the long-term persistence of each tree-ring series and of the hydrologic record. The fractal dimensions range from 1.739 to 1.939 for the tree-ring series for the calibration period, and from 1.884 to 1.946 for the entire chronology periods. The fractal dimension for the annual hydrologic record is 1.802, and 1.819 from October 1 through May 31. Modification of each tree-ring series based on the ratios of the Hurst exponent of each series forced the fractal dimensions of the tree-ring series to be closer to that of the hydrologic series. Fractal modification of the tree-ring series failed to improve streamflow reconstructions, but the modification of the tree-ring and hydrologic series suggests a more realistic reconstruction. Fractal analysis is useful to examine the stationarity of tree-ring series.