A combination of remote sensing data and topographic attributes for the spatial and temporal monitoring of soil wetness

This work presents a methodology for soil wetness monitoring over the Peace Athabasca Delta, PAD, a northern watershed located in the Mackenzie River Basin in North-Western Canada. The approach is based on the use of a combination of passive microwave data from AMSR-E instrument, data in the visible wavelengths from MODIS sensor and topographic attributes derived from the SRTM Digital Elevation Model. Classic topography-based wetness indices are static and unable to capture the temporal variability of soil moisture. Also, they do not account for vegetation effect on the spatial organisation of soil wetness. We therefore suggest overcoming these limitations by assessing the temporal variability of soil wetness using passive microwave data and introducing vegetation parameters into a new topography-based wetness index (TWI). The proposed approach is an improvement in a dynamic sense over classic topographic indices. It also distinguishes between wetness and open waterbodies contributions to the passive microwave signal. AMSR-E 37 GHz data have been used to regularly assess total soil wetness in the PAD. MODIS images have been used to develop a rating curve relationship between discharge observations and the extent of flooded areas. A basin wetness index (BWI) has been calculated as the difference between total wetness fraction from AMSR-E and inundated surface fraction retrieved from MODIS images. The new topographic wetness index disaggregates BWI values and distinguishes between inundated areas and wet soils responses in the passive microwave signal. Results were compared to those derived from the classic and static index. The proposed index has provided a better estimation of the wetness over the basin with respect to the classic wetness index. The correlation coefficients between the disaggregated wetness index and observed precipitations at Fort Chipewayn A station were 0.7 and 0.69 for 2003 and 2002 respectively. The obtained map of waterbodies in the delta which is the result of the downscaling process agrees well with a Landsat image as the determined probability of detection was equal to 0.7. This implies that a combination of passive microwave data and topographic attributes has an interesting potential to improve the water balance closure over northern basins.