This work presents a daily SWE product of 1980–2020 with a linear unmixing method through passive microwave data including SMMR, SSM/I and SSMIS over China after cross-calibration and bias-correction.
The pixel values in the H5 files have specific meanings: “0–240” represents the effective value of SWE, and the units of SWE is mm; “250” is for dry snow, “251” is for wet snow, “252” is for the free snow, “253” is for the water body, “254” means missing data and “255” is for outside China. To position the pixel location, the latitude and longitude matrices were also included in the H5 file.
1_Weather station measurements and snow course data
In situ snow depth measurements were from the weather stations and field survey course in China. The daily weather station measurements during the period 1980–2020 were provided by the National Meteorological Information Centre, China Meteorology Administration. The measured snow parameters are snow depth (daily) and snow pressure (every five-day), namely SWE. The dataset of daily station measurements can be accessed by scientific researchers through the submission of an application (http://data.cma.cn/en). The field snow campaign supported by the Chinese snow survey project was conducted from 2017 to 2019 winter months, and provides an important validation dataset for this study This dataset can be available from the corresponding author on request (Wang et al., Citation2018).
2_Satellite observations
Owing to the similar configurations and inter-sensor calibrations between the SSM/I and SSMIS, thus these instruments are selected to provide brightness temperature data from 1987 to 2020 (https://nsidc.org/data/NSIDC-0032/versions/2). The brightness temperature data during the period 1980–1987 were acquired from the SMMR on board the Nimbus-7 Pathfinder satellite (https://nsidc.org/data/NSIDC-0071/versions/1). The SMMR, SSM/I and SSMIS Equal-Area Scalable Earth-Grid (EASE-Grid) brightness temperature product at 25 km × 25 km resolution were used in this study.
3_Auxiliary data
In this paper, a linear unmixing algorithm was used to generate a 40-year snow depth dataset from 1980 to 2020. To develop the linear unmixing algorithm, the land cover fraction data are necessary.
In this paper, we applied a semi-empirical model with the linear unmixing method (hereafter, called LUM algorithm) designed for China’s snow cover (Jiang et al., Citation2014; Yang et al., Citation2018) to estimate snow depth. Then, a parameterized snow density (180 kg/m3) (Yang et al., Citation2019a) was used to transfer snow depth to SWE according to the ground-based measurements. A satellite passive microwave pixel usually covers several land use types due to its coarse spatial resolution.
The SWE estimates were compared to the weather station measurements during the period 2011–2019 . The overall unbiased root mean square error (unRMSE) and bias values are 5.09 cm and −0.65 cm, respectively. The correlation coefficient (corr.coe) is 0.84 (p < 0.01 at 0.05 confidence interval), representing the significant relationship between ground-based measurements and snow depth estimates.
This work is licensed under a
Creative
Commons Attribution 4.0 International License.
CSTR:11738.11.ncdc.I-SNOW.2020.6
10.12072/ncdc.I-SNOW.db0002.2020