Spaceborne observation and modeling of erosion and non-point source pollution in agricultural catchments draining into lakes

摘要

Soil losses by erosion and nutrients transport (phosphorus) have a negative impact on agricultural soils and water quality of rivers and lakes. We studied four (4) aspects of the problem: (a) the need to quantify and map theses losses at the watershed level, (b) to find a roughness indicator that give us information on the sensitivity of the soil to erosion and on agricultural practices, (c) to estimate the organic matter content of a soil as an indicator at the same time of the cohesion between soil particles and of the quality of the soil and finally (d) to quantify and map soil conservation practices at the watershed level. We used the SWAT model (Soil and Water Assessment Tool, USDA) for modeling the losses. We also used three (3) systems to estimate the ground surface roughness (pin-profiler, 3D stereo model and Laser scanning system) and extract a roughness indicator. Soil sampling, spectral signatures of soil and crop residues, vegetation indices and satellite imagery were then used to estimate organic matter content and soil conservation practices. Results show that the SWAT model has to be adapted to Quebec conditions but it was able to map areas sensible to losses. For the surface roughness, the pin-profiler proves to be the cheapest, fastest and easiest way to get data, however the standard deviation (sd) of heights is not the only or best indicator. An autocorrelation test combined with sd should lead to a more representative indicator of roughness. Organic matter content is estimated best with the second derivative at 561 nm of normalized reflectance spectra. Tests need to be done on an a different area and with satellite imagery. Finally, results using a Landsat TM image with the CRIM index prove to be useful to map crop residues as an indicator of soil conservation. Validation needs to be done on the ground.