We introduced a new evaluation method, the classification of multiple window-size based sinuosity spectrum. If the river is long enough for the analysis, the classification could be as useful, as the sinuosity spectrum, but sometimes it is more straightforward. Furthermore, for the classification, we did not need the main parameters of the river, e.g. the bankfull discharge. Each sinuosity calculation that was performed for a given window size, has been considered as one band (one channel) of a multichannel "image". Then, the sinuosity spectrums became multichannel images are of size 1 X N where N represents the length of the actual river in pixels. Using this multichannel input unsupervised ISOCLASS classification was carried out on these data, using ER Mapper software. The requested number of classes was set to 5. The results of the sinuosity calculations are scalars. Earlier, it was a subjective decision to divide the sinuosity values into the categories (low, medium-lo...
Gábor Timár hasn't uploaded this paper.
Let Gábor know you want this paper to be uploaded.
