Development of a Dry-snow Satellite Algorithm and Validation at the CEOP Reference Site in Yakutsk

Abstract

This paper presents a snow algorithm for the Advanced Microwave Scanning Radiometer (AMSR)and the AMSR for the Earth Observation System (AMSR-E). We validate the algorithm using snow-depth data recorded at the Coordinated Enhanced Observing Period (CEOP) Reference Site in Yakutsk, Russia. A new radiative transfer model for layered snow is developed by combining the 4-stream fast model and the dense media radiative model (DMRM); this model is then introduced into the new algorithm. The algorithm considers the effects of land-surface hydrological conditions under the snow layer and snow-panicle grain size on brightness temperatures in the microwave region by using the multi-frequency channels of AMSR and AMSR-E. The algorithm was validated at seven snow-measurement points within the CEOP Reference Site in Yakutsk from October 2002 to March 2003, corresponding to the first half of the third Enhanced Observing Period (EOP3).We calculated the root mean square error (RMSE) based on the error between observed and estimated values and calculated the residual standard deviation (RSD) for all verification periods. We also calculated the proportion and RMSE of overestimated and underestimated values. The average RMSE is low (4.0 cm) and the average RSD is 2.8, indicating only minor variation. In addition, 56% of values were overestimated, and the average RMSE of the overestimated values was 3.9 cm; the average RMSE for the 44% of values that were underestimated was 2.4 cm. Accordingly, the estimated snow depth is in relatively good agreement with in situ data.For the analyses undertaken at each site, we used the proposed algorithm to assess the influence of forest cover, frozen soil, and cloud cover on the process of estimating snow depth. We propose that such analyses are necessary when estimating snow depth.