Evaluation and Adjustment of the AMSR2 Snow Depth Algorithm for the Northern Xinjiang Region, China

Abstract

It is still a challenge to accurately map snow depth using passive microwave remote sensing. This study first validates the standard AMSR2 snow depth product in Northern Xinjiang, China, and then proposes an improved snow depth retrieving algorithm using the AMSR2 brightness temperature data in combination with in-situ measurements in the same region. The results show that: 1) in the past 15 years the mean snow depth based on the metrological stations ranges from 1.1 to 20.4 cm (mean 8.9 cm); 2) the standard AMSR2 snow depth product overestimates (underestimates) snow depth when snow depth thinner (thicker) than 30 cm, with an overall increased estimation error (root mean squared error) as snow depth increase; and 3) for the ascending mode (AMSR2_A) and descending mode (AMSR2_D), our improved algorithm shows smaller bias (2.5 and 3.9 cm) and smaller error (6.9 and 8.2 cm) as compared with the standard AMSR2 products (5.7 and 6.7 cm) and (11.2 and 12.1 cm), respectively. This suggests that the improved algorithm based on brightness temperature data of AMSR2_A has better accuracy and smaller error and can be used to retrieve snow depth of nonforest areas in cold period in the Northern Xinjiang.