Publications

Park, H; Im, J; Kim, M (2019). Improvement of satellite-based estimation of gross primary production through optimization of meteorological parameters and high resolution land cover information at regional scale over East Asia. AGRICULTURAL AND FOREST METEOROLOGY, 271, 180-192.

Abstract
Gross primary production (GPP) is a crucial factor in the carbon cycle especially for the absorption of carbon dioxide into the biosphere from the atmosphere. A large discrepancy between a satellite-based GPP product named MOD17A2H GPP and in-situ data measured at eddy covariance flux towers has been identified over East Asia where the biome types and climatic characteristics are heterogeneous with rugged terrain. For that reason, this study focuses on two potential major error sources in MOD17A2H GPP, which are the coarse resolution land cover information and inappropriate meteorological parameters. The finer resolution observation and monitoring global land cover (FROM-GLC) and the MODIS land cover product collection 5.1 (MCD12Q1) were used to describe biome types in detail, by combining spatial distribution from FROM-GLC and the phenological characteristics of land cover from MCD12Q1. Meteorological parameters were optimized using the 55-years Japanese reanalysis meteorological data (JRA-55). The light use efficiency of the MOD17 GPP algorithm was modified using the combined land cover information (FROM-MCD). Although the use of FROM-MCD and JRA-55 did not improve MOD17A2H GPP, the optimization of two meteorological parameters-daily minimum air temperature (TMIN) and vapor pressure deficit (VPD) significantly improved the GPP algorithm for East Asia. The results show that the root mean square errors (RMSEs) between the estimated and in situ GPPs decreased from 21.83 (gC/m(2)/8days) to 16.11 (gC/m(2)/8days) through optimizing the two parameters at 9 flux tower sites. The optimized TMIN and VPD thresholds in the MOD17 GPP algorithm were applied to the entire study area (i.e., East Asia) according to the Koppen-Geiger climate classes. The estimated GPP using the proposed approach was compared to GPPs from widely used process-based models (i.e., VISIT, BEAMS, and BESS), which confirmed that the proposed approach with locally optimized meteorological parameters improved on the underestimation of the MOD17 GPP algorithm for East Asia. The uncertainty of the VPDmin parameter was revealed to be larger than that of TMINmax.

DOI:
10.1016/j.agrformet.2019.02.040

ISSN:
0168-1923