Main content area

Ground-based evaluation of MODIS snow cover product V6 across China: Implications for the selection of NDSI threshold

Zhang, Hongbo, Zhang, Fan, Zhang, Guoqing, Che, Tao, Yan, Wei, Ye, Ming, Ma, Ning
The Science of the total environment 2019 v.651 pp. 2712-2726
algorithms, altitude, moderate resolution imaging spectroradiometer, snow, snowpack, temperature, China
The new MODIS daily NDSI snow cover product version 6 (V6) is released to replace V5 with significant revisions. This study evaluates, for the first time, the accuracy of product V6 across China based on daily snow-depth measurements during 2003–2013 from 279 and 252 stations for Terra and Aqua, respectively. Three schemes of selecting NDSI thresholds for Terra and Aqua were tested and compared including: (1) the locally optimal NDSI threshold, (2) the minimum valid NDSI of 0.1, and (3) the global reference NDSI threshold of 0.4. The mean Cohen's Kappa (CK) of the optimal, minimum and global reference thresholds for Terra (Aqua) are 0.80 (0.60), 0.77 (0.58), 0.72 (0.51), respectively, while snow depth ≥ 1 cm. The NDSI threshold of 0.1 is demonstrated to be more reasonable than the threshold of 0.4 for use in China. This is also supported by the accuracy comparison conducted for the clear-day snow-cover day calculation. Terra V6 and Terra V5 have comparable accuracies whereas Aqua V6 shows better accuracy than Aqua V5 does. The revised temperature screen algorithm employed in V6 is found to be problematic with large snow commission errors in high altitude stations. Regionally, product V6 presents low CKs of 0.61 and 0.35 for the optimal thresholds of Terra and Aqua in the Tibetan Plateau, which are attributed to its high elevation and relatively small snow depth. This study provides practical implications for use of MODIS snow cover production V6 in China.