TY  - Data
T1  - Grid data of soil organic carbon density with 2m resolution in Dayekou watershed of Qilian Mountains
A1  - Zhu Meng
A1  - Zhang Chengqi
A1  - Zhang Jutao
A1  - Feng qi
DO  - 10.12072/ncdc.qlsst.db0012.2021
PY  - 2021
DA  - 2021-03-01
PB  - National Cryosphere Desert Data Center
AB  - The measured soil organic carbon density data of this dataset are mainly from 263 sampling points on slope and watershed scale obtained by general projects of NSFC (41771252, 31270482, 91025002).According to the framework of "scorpion" (soils, climate, organizations, relief, parent material, age, geographic position), based on Digital Soil mapping, The spatial distribution of soil organic carbon density in 0-100 cm soil layer of Dayekou watershed in Qilian Mountains was simulated by integrating 2 m DEM data (Zhang Yanli, 2020) and QuickBird 2.5 m resolution multispectral images (Guo Jianwen, 2019). The prediction method is mainly based on xgboost (extreme gradient boosting) algorithm in machine learning, which uses grid data such as climate, precipitation, radiation, terrain, vegetation index, spectral information and spatial position as input variables for spatial mapping. Through repeated modeling of bootstrap, each bootstrap sample is modeled in space, and the frequency distribution of modeling results is obtained. The uncertainty of modeling is expressed by standard deviation (SD). The RMSE and R2 of the model were 5.34 kg / m2 and 0.84 kg / m2, respectively. In the final data product, mean and SD represent the mean and standard deviation of 30 repeated modeling, respectively, in kg / m2, representing the mass of soil organic carbon in 0 ~ 100cm soil layer per unit area.
DB  - NCDC
UR  - http://www.ncdc.ac.cn/portal/metadata/2d975743-d6aa-4465-98cb-337286eea328
ER  -