440 x 445 pixel data to generate the radiance image. The sliced signatures also served as values for R_f, the "pure" end-member spectra.

In the unmixing proper, applying the appropriate crop and soil type signatures is determined through the use of agriculture land use and soil series maps with validation for soil sampled from the field. Other field data such as leaf dimensions and number, and plant height, spacing and density for each full grown crop served as an input to compute their respective LAI_0s while values for C were obtained from Gilabert et al.(2000) but may be obtained if leaf samples are brought to the laboratory for an LAI experiment. Built-up areas and bare soil types were masked out and automatically assigned LAI_0s of zero. Selective LAI₀ and C usage likewise makes use of the agriculture land use map.

The DEM and channel routes were obtained by the digitizing the topographic map while data from the four rain gauges were used to generate a continuous distributed rainfall map for four rainfall events by inverse distance interpolation. The hydrologic model was then implemented in computational grid size of 10 meters at 10-second time intervals inside the watershed boundary. Aside from the Courant-Friedrichs-Lewy (CFL) routing scheme stability criterion, the selection of grid size and time interval depended on the consideration for an ideal grid size that minimizes detail loss due to spatial aggregation while maintaining a manageable number of cells for optimal computation time. It is assumed that for a single grid cell, only the fractions of soil and one specific vegetation type are present. Finally, runoff, water and sediment discharge simulation results are compared with observed values by means of regression techniques.

3. Results and Discussion

3.1 Land Use, LAI Estimates and Soil Detachment Rates


Fig. 1 In rainfall event 1,(a) cumulative soil detached per unit area; (b) soil detachment rate to increasing LAI for the same land use at maximum rainfall intensity.