Inversion of a Snow Emission Model Calibrated WithIn SituData for Snow Water Equivalent Monitoring

F Vachon,D De Seve,A Royer,K Goita

doi:10.1109/tgrs.2009.2026892

F Vachon, D De Seve + Show 2 more

https://doi.org/10.1109/tgrs.2009.2026892

Copy DOI

Export

Save

Cite

Abstract
Full-Text
Similar Papers

Abstract

Listen

Snow water equivalent (SWE) is critical information for hydrological studies over northern latitudes. This information is needed to adequately simulate spring runoff in order to optimize hydropower generation. In this paper, a new geophysical inversion methodology for monitoring SWE evolution on a daily basis is presented. The methodology is based on a modified version of the Helsinki University of Technology (HUT) snow emission model and assimilates in situ data as ground truth. Previous studies used the mean snow grain size (?) as an optimization parameter to retrieve SWE with the HUT model. Therefore, a priori information on ? was needed to estimate SWE, but this information is rarely measured in the field. In the HUT model, the mean snow grain size is used to compute the total extinction coefficient (? e ) of the snowpack. We modified the model to eliminate the required a priori knowledge on mean snow grain size and used the total extinction coefficient as an optimization parameter instead. The methodology was applied in a subarctic territory in northern QuE?bec (Canada) using snow measurements acquired during the 2001 to 2006 winters (January 1 to March 31). A metric based on the brightness temperature difference between the special sensor microwave/imager 19- and 37-GHz (vertical polarization) was used for SWE retrieval. In the monitoring process, the result obtained from the previous day inversion was used to initialize the current day SWE retrieval. When in situ measurements are available, they are assimilated and used to calibrate the model. The results obtained with this methodology show an improvement in the experimental context considered (high SWE values). For all winters considered in this paper, the results described in this paper show that SWE was estimated with a root-mean-square error (rmse) of less than 19% after the first process of in situ data assimilation and with an rmse of 14% for nondense forested areas and for SWE values below 300 mm.

Full Text