Figure 1.
The map of the study region with land cover type and location of eddy-covariance towers.
Land cover map was derived from the MODIS Land Cover product (MOD12Q1).
Table 1.
Site name, location, vegetation type, and available years of the EC data used for developing and validating the ET model.
Figure 2.
Interannual variability of meteorological variables, evapotranspiration (ET) and water balance over the DEA region.
Meteorological variables includes mean annual air temperature (A), total solar radiation and sunshine hours (B), mean relative humidity (C), and total precipitation (D). The air temperature, solar radiation and relative humidity derived from MERRA are calculated over the grassland area of the DEA. The field air temperature, sunshine hours, relative humidity and precipitation are the average value at meteorological stations. The evapotranspiration is the average value of RT ET over the DEA region (E). The precipitation derives from GPCP dataset (D). The water balance is the difference between GPCP precipitation and RT ET (F).
Figure 3.
Spatial pattern of trends in meteorological variables, evapotranspiration, water balance and NDVI over the DEA region.
(A)–(C) show spatial pattern of trends in mean annual air temperature, solar radiation, and average relative humidity from MERRA datasets, respectively. (D) shows the precipitation trends of GPCP dataset. (E) and (F) show the spatial pattern of trends in mean annual evapotranspiration and water balance respectively. (G) shows the trends in NDVI of growing season. The insets show the frequency distributions of corresponding trends.
Figure 4.
Observed (ETo) and predicted ET (ETp) with the model calibration (A) and validation (B) datasets.
The short dashed lines are 1∶1 lines and the solid lines are linear regression lines.
Figure 5.
Maps of multi-year (1982–2009) mean annual evapotranspiration (ET), precipitation, and water balance.
(A)–(C) show the regression tree ET, GPCP precipitation and water balance derived from GPCP precipitation, respectively.
Figure 6.
Seasonal, growing season, and annual trends of regional average evapotranspiration (ET), GPCP precipitation (P(GPCP)), and water balance derived from GPCP precipitation (W(GPCP)).
Figure 7.
The correlation coefficients between mean annual evapotranspiration and NDVI of growing season (A), mean annual GPCP precipitation (B), mean annual temperature (C), and annual solar radiation (D).
The grids with insignificant correlations were marked in gray, and other colors indicate significant correlations (p<0.05).
Figure 8.
Interannual variability of RT ET, PT-JPL ET, MODIS ET and LandFlux-EVAL synthesis ET datasets.
The LandFlux-EVAL synthesis ET datasets included four ensemble products from diagnostic ET data sets (LandFlux-EVAL Diagnostic ET), land surface models (LSMs) ET (LandFlux-EVAL LSMs ET), ET reanalyses dataset (LandFlux-EVAL Reanalyses ET)) and all the three categories (LandFlux-EVAL synthesis All ET).