Study On Simulation Methods Of Alpine Grassland Net Primary Productity In Three Rivers Source Region Of Tibetan Plateau, China

Net primary productivity (NPP) is a crucial part of the terrestrial carbon cycle, and plays a significant role in global change and the carbon balance. Three Rivers District, which is the birthplace of the three major rivers, is a vital source of hydration. It is very sensitive to climate change and its ecological environment is very fragile. Therefore, the study of the dynamic changes of the space-time will be of great importance for understanding the surface carbon sinks and how the NPP of alpine grassland responses to climate change.In this study the Three Rivers area is the study area. With the use of remote sensing and GIS techniques, this study analyzed the temporal and spatial distribution of the annual accumulated temperature above O℃, mean annual temperature and annual precipitation in this area. In this study, the NPP simumlation results of MIAMI model, Classification-Index model (CIM) based on the comprehensive and sequential classification system (CSCS), CASA, the linear regression model and MOD17A3data based on BIOME-BGC model were compared and analysed. In order to simulate the NPP of real grassland and improve the the simulation accuracy by modifying the CIM, in this study we focused on the grassland and excluded the non-grass area, based on the CSCS, with reference to IGBP data and the vegetation map. By selecting the main affecting factors, combined with the MODIS remote sensing data and the CIM model, a new CIM correction model was proposed. By using this CIM correction model we simulated the NPP of different grassland types and compared the spatial variation between2005and2006.This study achieved following results:(1)the thin-plate spline interpolation method was used to interpolate the meteorological data and by comparing the NPP simulation results of MIAMI, CIM, CASA models and MOD17A3data, which were NPPM, NPPL, NPPC, and NPPB, on the correlation coefficient (R), NPPB> NPPC> NPPM> NPPL, while on the root mean square error (RA), NPPL<NPPM<NPPB<NPPC;(2) NDVI and the ratio of NDVI and NDVI of the corresponding grass type (NN) were selected to modify the CIM model in the study area, and the best CIM correction model was NPP=291.52·ln(x)-1017.1, x=CIM·NN;(3) By the simulation of the CIM correction model, the NPP of study area in2006was less than2005. According to the simulation of different grassland types in the study area, the average NPP of cold temperate-subhumid montane meadow steppe was higher than other types, while the figid-humid tundra, alpine meadow class had the lowest NPP among all the types.In this study only the grassland in the area had been researched. Combining the remote sensing data with the CIM model, not only the simulation accuracy was improved, but the model could also be applied to simulate the actual grassland NPP, thus increase the practical value of the CIM model.