PROGRESS IN GEOGRAPHY ›› 2021, Vol. 40 ›› Issue (8): 1386-1396.

• Articles •

### Near surface air temperature estimation based on MODIS atmospheric profile product over Qinghai Province

TIAN Shengrong1,2(), ZHU Wenbin1,*(), ZHOU Shijian3

1. 1. Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
2. Faculty of Geomatics, East China University of Technology, Nanchang 330013, China
3. Nanchang Hangkong University, Nanchang 330063, China
• Received:2020-11-06 Revised:2021-04-20 Online:2021-08-28 Published:2021-10-28
• Contact: ZHU Wenbin E-mail:shengrongt@163.com;zhuwb@igsnrr.ac.cn
• Supported by:
Key Science and Technology Project of Qinghai Province(2019-SF-A4);Basic Research Program of Qinghai Province(2020-ZJ-715);Scientific Research and Promotion Projects of the Second Phase Project of Ecological Protection and Construction of the Three Rivers Source in Qinghai Province(2018-S-3);National Natural Science Foundation of China(42071032);Youth Innovation Promotion Association of Chinese Academy of Sciences(2020056)

Abstract:

Near surface air temperature is one of the most essential variables in a wide range of disciplines such as hydrology, meteorology, and ecology. Accurate estimation of air temperature for continuous spatial and temporal coverage is also crucial to climate change studies. Considering the sparse distribution of meteorological stations in Qinghai Province of China, this study estimated its near surface air temperature (Ta) based on meteorological observations, MODIS products, and SRTM DEM data from 2011 to 2019. First, daily instantaneous Ta under clear and cloudy sky conditions was retrieved from MOD07_L2 and MOD06_L2 products without the support of meteorological observations. Three parameterization schemes were used in this process and their results were compared. Second, these daily Ta estimates with highest accuracy were combined with meteorological observations and SRTM DEM data to generate high-precision monthly air temperature through the multiple regression method. Finally, the spatial and temporal distribution of Ta in Qinghai Province was analyzed. The results show that the accuracy of daily instantaneous Ta estimates can be improved significantly through averaging the near surface air temperature retrieved from MOD07_L2 and land surface temperature retrieved from MOD06_L2 without the calibration of meteorological observations. Specifically, the correlation coefficient (r) and root mean square error (RMSE) under clear sky conditions was 0.93 and 4.71 ℃, respectively; and the r and RMSE under cloudy sky conditions was 0.89 and 5.16 ℃, respectively. The multiple regression model with the calibration of meteorological observations provided a good estimation of monthly air temperature. The cross-validation results indicate that the determination coefficient (R2) and RMSE were generally above 0.8 and below 2.5 ℃, respectively. The high accuracy of monthly Ta estimation made it possible for us to investigate the spatial and temporal distribution of Ta in Qinghai Province. To be specific, the maximum and minimum monthly air temperature occurred in July and January, which was 13.59 ℃ and -9.44 ℃, respectively. Its spatial distribution was dominated by elevation. The average adiabatic lapse rate over the whole province was 4 ℃/km, which agreed well with the ground-based meteorological observations. The analysis indicates that MODIS atmospheric profile product holds great potential for all-weather air temperature estimation. The introduction of ground-based observations can reduce significantly the systematic error of air temperature estimation. The combination of MODIS products and ground-based observations proves to be useful for accurate air temperature estimation over large areas with complex topography.