Journal of Geographical Sciences ›› 2020, Vol. 30 ›› Issue (1): 18-36.doi: 10.1007/s11442-020-1712-0

• Special Issue: Global and Regional Land Surface Characteristics and Socio-economic Scenarios • Previous Articles     Next Articles

Implications of diurnal variations in land surface temperature to data assimilation using MODIS LST data

FU Shiwen1,2, NIE Suping3, LUO Yong1,*(), CHEN Xin1   

  1. 1. 1. Department of Earth System Science, Tsinghua University, Beijing 100084, China
    2. Numerical Weather Prediction Center, China Meteorological Administration, Beijing 100081, China
    3. National Climate Center, China Meteorological Administration, Beijing 100081, China
  • Received:2018-12-20 Accepted:2019-03-26 Online:2020-01-25 Published:2020-03-25
  • Contact: LUO Yong E-mail:yongluo@tsinghua.edu.cn
  • About author:Fu Shiwen (1994–), Master, specialized in land surface data assimilation. E-mail: fswflora@gmail.com
  • Supported by:
    National Key Research and Development Program of China, No.(2017YFA0603703);National Key Research and Development Program of China, No(2016YFA0602102)

Abstract:

Based on the Beijing Climate Center’s land surface model BCC_AVIM (Beijing Climate Center Atmosphere-Vegetation Interaction Model), the ensemble Kalman filter (EnKF) algorithm has been used to perform an assimilation experiment on the Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST) product to study the influence of satellite LST data frequencies on surface temperature data assimilations. The assimilation results have been independently tested and evaluated by Global Land Data Assimilation System (GLDAS) LST products. The results show that the assimilation scheme can effectively reduce the BCC_AVIM model simulation bias and the assimilation results reflect more reasonable spatial and temporal distributions. Diurnal variation information in the observation data has a significant effect on the assimilation results. Assimilating LST data that contain diurnal variation information can further improve the accuracy of the assimilation analysis. Overall, when assimilation is performed using observation data at 6-hour intervals, a relatively good assimilation result can be obtained, indicated by smaller bias (<2.2K) and root-mean-square-error (RMSE) (<3.7K) and correlation coefficients larger than 0.60. Conversely, the assimilation using 24-hour data generally showed larger bias (>2.2K) and RMSE (>4K). Further analysis showed that the sensitivity of assimilation effect to diurnal variations in LST varies with time and space. The assimilation using observations with a time interval of 3 hours has the smallest bias in Oceania and Africa (both<1K); the use of 24-hour interval observation data for assimilation produces the smallest bias (<2.2K) in March, April and July.

Key words: land surface data assimilation, land surface temperature, MODIS, diurnal variation