Satellite-observed enhanced carbon sink with local greening in China

Trends in vegetation indices observed from remote sensing indicate changes in ecosystem productivity and land use change. Vegetation greening trends over eastern China during the growing period (GP) from 2001-2017 are captured using NDVI (0.0045±0.0025 per year) and EVI (0.0038±0.0020 per year) from MODIS MOD13C1 (Collection 6). The growth period of each grid (1 x 1 degree in latitude and longitude) was identified with logistic fitting applied to the EVI dataset with a temporal resolution of 16 days. In addition, LAI and FPAR (from MCD15A3H) and GPP (from MOD17A2H) also show significant increase trend over north part of the eastern China, but not in the south part.

Trends in column-averaged CO₂ dry air mole fraction (XCO₂) in these regions imply associated changes in net CO₂ fluxes. We examined the XCO₂ series from combined SCIAMACHY and GOSAT observations and a spatio-temporal kriging interpolation mapping method. Decreasing trends are shown in that region with mapping XCO₂ time series during the GP from 2003 to 2016, after subtracting a yearly increase over each gird. The interpretation of the XCO₂ trends in this area are complicated by nearby anthropogenic emissions from major urban centers in eastern China. We attempt to remove a yearly background increase in order to make the XCO₂ trends comparable to greening trends. We used XCO₂ from CarbonTracker 2017 (CT-XCO₂), an atmospheric inversion model, as an estimate of the yearly increase. In eastern China, the GP XCO₂ after subtracting the CarbonTracker results is decrease by 0.74±0.25 ppm/year in the south and 0.53±0.17 ppm/year in the north. The trends in the north of China may be partly explained by agricultural productivity increases in that region, but we found no similar agricultural productivity trends in the south of China.

This analysis highlights two challenges in remote sensing of carbon fluxes. First, co-located regions of changing urban anthropogenic emissions and biogenic uptake are difficult to separate. Second, the CarbonTracker model is under-constrained in Asia. We will address this second issue by comparing XCO₂ observations to CT-XCO₂ to examine biases. Identifying changes in the terrestrial sinks observed from satellites will provide more accurate constraints for anthropogenic emissions estimation and greenhouse gas model simulations.