Snow is the most widely distributed and significantly changing member of the cryosphere. Its unique high albedo and low thermal conductivity characteristics make it play a key role in regulating surface energy balance, ground atmosphere interactions, and hydrological processes. Due to the limited spatial resolution, temporal resolution, and available time series of early sensors, there is still a lack of long-term series Snow Cover Extend (SCE) products with a spatial resolution of 500m before 2000. To fill this data gap, the study extended the existing SCBOT (Similar Conditional Probability&OTSU) algorithm to the entire China region and made improvements to address some of the shortcomings in the original algorithm. Firstly, the similarity conditional probability (SCP) between images at different scales was calculated using MODIS SCE and AVHRR SCE covering the same time period. Furthermore, based on the 5km AVHRR SCE during the period of 1981-2000, this set of downscaled products was generated through a series of post-processing. The spatial resolution of the product is 500m, the temporal resolution is 1 day, and the data naming format is "YYYYMMDD. tif", such as "19810626. tif", which contains six pixel values (0=land, 1=SCP identified snow, 2=spatiotemporal cube (STCPI) interpolated snow, 3=snow depth interpolated snow, 4=water, 255=fill value). This product effectively fills the data gap during this period and is of great significance for long-term climate change research, water resource management, and ecological protection.
Based on MODIS surface reflectance data (MOD09GA/MYD09GA) and AVHRR surface reflectance version 4 (AVHRR SR V4), combined with multi-level decision tree classification, hidden Markov model, multi-source data fusion and other methods, the snow cover research team of the Northwest Institute of Ecological Environment and Resources (NIEER) of the Chinese Academy of Sciences systematically completed snow pixel recognition and cloud pixel filling, respectively constructed the cloud free MODIS SCE and AVHRR SCE datasets day by day, providing important basic data support for this study. In addition, auxiliary data includes passive microwave snow depth data jointly generated by Che et al. and Dai et al. based on multi-source satellite passive microwave observations, ERA5 land surface reanalysis surface temperature products, and a 90 μ m resolution DEM provided by the Space Shuttle Radar Topography Mission (SRTM). The validation data includes snow depth observation data provided by the China Meteorological Administration (CMA) and multiple Landsat-5 images.
Firstly, the similarity conditional probability (SCP) between images at different scales is calculated using MODIS SCE and AVHRR SCE covering the same time period, and the optimal segmentation threshold for SCP images is obtained using Otsu's method. Based on the above SCP images and Otsu threshold, the 5km AVHRR SCE from 1981 to 2000 was subsequently downscaled. For the preliminary downscaling results obtained, various post-processing methods such as STCPI interpolation, passive microwave snow depth interpolation, temperature altitude snow cover shielding, etc. were successively applied to finally obtain the downscaled product.
The product is based on high-precision remote sensing data and strictly controls the quality of each step in the production process. The validation results based on ground snow depth observation data in China indicate that the overall accuracy (OA) of the downscaled product is 0.84, the recall rate (RC) is 0.77, the precision rate (PC) is 0.93, and the Coampa coefficient (CK) is 0.69. Further validation was conducted using Landsat-5 SCE data from the same period, and the results showed that OA, RC, PC, and CK were 0.82, 0.83, 0.81, and 0.63, respectively. In addition, accuracy evaluation also shows that the product can maintain high interannual stability in a long time series, with OA consistently above 0.8 and CK stable at around 0.7. The above results indicate that the downscaled products have good reliability and stability, and can effectively fill the current data gap.
| # | number | name | type |
| 1 | 42125604 | The National Science Fund for Distinguished Young Scholars | |
| 2 | 42271373 | National Natural Science Foundation of China |
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
CSTR:11738.11.NCDC.LZU.DB6929.2025
10.12072/ncdc.lzu.db6929.2025
