On predicting channel initiation from gridded elevation data

Three threshold conditions for channel initiation are evaluated by using gridded elevation data derived from a detailed lidar survey, a reliable algorithm for the determination of surface flow paths, and accurate field observations of channel heads for a study area located in the eastern Italian Alps. These three threshold conditions are determined by considering part of the observed channel heads and computing for them the related values of (1) the drainage area A, (2) the monomial area-slope function AS2 of the drainage area A and the local slope S, and (3) the Strahler order ω* of surface flow paths extracted from gridded elevation data. Attention is focused on the dependence of the obtained threshold values on the size of grid cells involved, and on the ability of threshold conditions to predict channel heads. The results indicate that (i) the uncertainty in the threshold values of the Strahler order ω* is significantly smaller than that affecting the threshold values of the drainage area A and the area-slope function AS2, (ii) threshold values of A, AS2, and ω* are all significantly dependent on grid cell size, and (iii) the threshold values of the Strahler order ω* follow quite well a power function relationship of grid cell size. The comparison between predicted and observed channel heads, performed on three different drainage basins, indicates that Strahler classification of surface flow paths and pruning of exterior links provide a sound basis for the determination of channel heads formed essentially by surface erosion. In addition, it reveals that more comprehensive methods are needed to predict channel heads affected by groundwater seeping upwards in areas characterized by strong geologic controls.