High quality permafrost map is the basic data for the study of Permafrost Environmental Effects and engineering application in cold regions. Based on the systematic compilation of the annual average ground temperature measurement data of 237 borehole locations in the Qinghai Tibet Plateau from 2005 to 2015, the data set integrates these ground observation and remote sensing freezing index, melting index, snow days, leaf area index, soil bulk density, elevation and high-quality soil moisture reanalysis data by using support vector regression model, The annual average ground temperature distribution map of Qinghai Tibet Plateau with 1km resolution from 2005 to 2015 is simulated. 10 fold cross validation shows that the root mean square error of the simulated annual average ground temperature is about 0.75 ° C and the deviation is about 0.01 ° C. Based on the stability classification system of high altitude permafrost, the thermal stability types of permafrost are divided by using the annual average ground temperature. The data show that the permafrost area of the Qinghai Tibet Plateau is about 115.02 (105.47-129.59) × 104km ², Among them, the permafrost areas of extremely stable type (< - 5.0 ° C), stable type (- 3.0 ~ - 5.0 ° C), metastable type (- 1.5 ~ - 3.0 ° C), transition type (- 0.5 ~ - 1.5 ° C) and unstable type (> - 0.5 ° C) are 0.86 respectively × 104km ², nine point six two × 104km ², thirty-eight point four five × 104km ², forty-two point two nine × 104km ² And 23.80 × 104km ²。 The data set can be used for the planning, design, ecological planning and management of projects in cold regions, and can be used as a data benchmark for the current situation of permafrost to evaluate the changes of permafrost in the Qinghai Tibet Plateau in the future. For more detailed methods of this data, please refer to the paper of Chinese Science: geoscience (ran et al., 2021).
The annual average ground temperature measurement data of 237 boreholes in the Qinghai Tibet Plateau from 2005 to 2015 compiled in this data set are mainly from the existing literature and a small amount of unpublished data. The freezing index, melting index, snow days, leaf area index, soil bulk density, elevation and soil moisture data as predictors are from remote sensing observation and reanalysis data, For details, see the paper (ran et al., 2021).
The annual average ground temperature prediction variables such as freezing index, melting index, leaf area index, snow days, soil humidity, precipitation, soil properties (bulk density, organic carbon content, sand content, silt content and clay content) and terrain related factors (altitude and potential incident solar radiation) based on remote sensing surface temperature are optimized, Select a few important variables to construct statistical learning models (generalized linear model, generalized additive model, support vector machine regression, random forest, geographically weighted regression and set average of five models). secondly; According to the cross validation results, the optimal model is selected to simulate the annual average ground temperature. Finally, the stable types and ranges of permafrost are divided according to the simulation results of annual average ground temperature.
Both direct and indirect show that the uncertainty of annual average ground temperature is less. Based on the existing permafrost distribution range and ground observation data, it is found that the permafrost range obtained in this study has higher accuracy.
| # | number | name | type |
| 1 | XDA19070204 | Strategy Priority Research Program (Category A) of Chinese Academy of Sciences |
This work is licensed under a
Creative
Commons Attribution 4.0 International License.
CSTR:11738.11.ncdc.NIEER.db2513.2022
10.11888/Geogra.tpdc.270672