Flux Footprint Analysis And Remotely Sensed Estimation Of Gross Primary Productivity At Southwest Mountainous Ecosystem Of China

Accurate monitoring GPP（Gross Primary Productivity）of terrestrial ecosystems is critical to understanding global carbon cycle and its responses to climate change.Mountains account for about 24%of world’s total land surface area,while China’s mountains are approximately65%of the land surface.As an important part of global climate change,it is necessary to assess the processes of carbon cycle over mountain ecosystems accurately and effectively.Taking the southwest mountainous district of China-Wanglang national nature reserve as an example,based on observed data of three flux stations over the study area,a two-dimensional footprint FFP（Flux Footprint Prediction）model was used to calculate flux footprints at multiple time scales,analyzing their characteristics and evaluating their spatial representativeness at multiple pixel scales（30 m,60 m,120 m,250 m,500 m,1000 m and 2000 m）.Combined with the spatial representativeness results of site fluxes,a vegetation index model that did not account for topography（Temperature and Greenness,TG）and a vegetation index model that considered topography（Mountainous Temperature and Greenness,MTG）were used to estimate long-term vegetation GPP,the long-term topographic effects on vegetation GPP estimation and its relationship with environmental factors were also explored.The main results of this study are as follows:（1）In terms of spatial variation of flux footprint,the flux footprint ranges of different sites over Wang Lang station were large（10～10³ m）,while their symmetry were low（usually less than 40%）.In terms of temporal variation of flux footprint,the overlapping footprints of different sites varied greatly at the daily scale（0%～88%）,while presenting a good overlap overall at the monthly scale（>83%）.The observed heights of deciduous broadleaf shrub,deciduous broadleaf forest and evergreen needleleaf forest sites over study area were 10 m,30m and 75 m,which obtained the best spatial representativeness of flux footprints at the pixel scales of 30 m,60 m and 1000 m,respectively.Therefore,it is especially necessary to focus on the differences in spatial representativeness of flux footprints while remote sensing models operating and products verifying over mountainous ecosystems.As for high temporal resolution scales of model operation and product production（e.g.,daily scale）,it is necessary to combine corresponding footprint characteristics to capture the temporal variations of flux observations.（2）After parameter calibration,compared with GPP estimation by TG model(GPP_TG),the MTG model estimated GPP(GPP_MTG)had a better consistency with EC GPP（R² increased by0.02～0.04 and d increased by 0.01～0.07）.GPP_MTG exhibited a higher spatial variation than GPP_TG,with an increasing coefficient of variation of 19%.Results showed that coupled terrain information could effectively improve the accuracy of models in estimating GPP of mountainous vegetation and captured its spatial changes.（3）During the growing season,GPP_MTG showed highly similar seasonal variations with Downward Shortwave Radiation（DSR）and Land Surface Temperature（LST）.Over the 21years of this study,GPP_MTG was more sensitive to time series radiation than GPP_TG（with a higher correlation coefficient of 0.10）and less sensitive to time series temperature（with a lower correlation coefficient of 0.04）.The sensitivity of GPP_MTG to time series DSR and LST varied significantly with the variation of terrain attribution（i.e.,elevation,slope and aspect）,indicating that long-term topographic effects could regulate the distribution of environmental factors（such as radiation and temperature）,which in turn affect the seasonal variations of vegetation photosynthesis over mountainous watersheds.