Estimation of Net Primary Productivity of Desert Ecosystem Based on Multi-scale Remote Sensing Data Fusion
Net Primary Productivity (NPP) refers to the amount of the organic dry matter accumulation of the green plants per unit time and per unit area,is the rest of organic matter amount produced by photosynthesis after deducting the autotrophic breathing.In recent years,the NPP has gradually become a hotspot of research on resources and environment in arid areas,accurate estimating the spatial and temporal variations of NPP is of great significance to understand climate change and carbon cycle in arid areas.Taking Wucaiwan area in eastern junggar basin in Xinjiang as the research area,this paper estimated the NPP value of the study area during 2010 through the improved CASA model using meteorological data and multi-scale remote sensing data,and analyzed its spatial and temporal distribution.The results of this paper showed that:① NPP can be estimated just using remote sensing data and ground meteorological data.It showed strongly superiority for this model and its application can be enhanced.② The MODIS NDVI and TM fuse together to produce high-time-resolution TM data which contributes to the study on high-spatial-resolution NPP data.③ The results showed that the total annual NPP in the study area was 6.571 78 Mt · a-1 in 2010.The spatial changes of NPP were reasonable,and it decreased from northwest to southeast.Moreover,seasonal variations of NPP were also large.It was about 57.35 ％ of the total annual NPP in three month of June,July and August,which shows that the vegetation grew vigorously.However,the NPP value was very low,which was almost zero,in December,January,February,because vegetation almost stopped growing.④ Using the ground measured NPP data to verify the estimated NPP data by CASA model,the average relative error is about 0.167,so the average verification precision is up to 83.3％.The correlation coefficient between NPP observed data and NPP estimated data is 0.952 which is significant at the 0.01 level.
国家科技期刊平台
NSTL
万方数据
维普
