Abstract
Because of the limited number of observation stations and the short time series of orbiting carbon satellite data, it is difficult to monitor CO2 concentrations (XCO2) at broad spatial scales for long time spans. Therefore, we are limited in accurately forecasting change in XCO2. Studies based on the approach of using satellite sensor-derived data as independent variables to model CO2 exchange show promising results for closed forest stands. There is a need to extend this approach to other land-cover types to monitor XCO2 at large spatial scales. In this study, Moderate Resolution Imaging Spectroradiometer (MODIS)-derived indices were used to model XCO2. Greenhouse Gases Observing Satellite (GOSAT) data and MODIS-derived indices in 2010 and 2011 were selected to construct XCO2 models during the growing season (May–October). We selected three ground stations to assess the accuracy of the modelled XCO2 for each month from 2011 to 2013. The accuracy of the results indicates that the average bias was 2.25, 4.53, and 4.43 ppm at the three ground stations, respectively, although the largest bias was 10.03 ppm (at Shangdianzi Station in June 2013). We also used GOSAT Thermal and Near Infrared Sensor for Carbon Observation (TANSO) point data in 2012 and 2013 as the observed data to assess the accuracy of the XCO2 models, and achieved a slightly favourable result for each month, except June. The overall conclusion of this study is that the proposed new approach to obtaining XCO2 at the regional scale needs to be perfected in the future.
Acknowledgement
We thank the GOSAT Project of Japan, NASA, and WDCGG for the use of their data in this study.