Remote Estimation Of Leaf Net Photosynthetic Rate Using Hyperspectral Reflectance

Photosynthesis,which means green plants absorb optical energy,transfer carbon dioxide in atmosphere into its organic matter and release oxygen in the meanwhile,is the most important physiological activities for vegetations'growth and the fundamental basis of carbon cycle in ecosystem.Traditional photosynthetic research is time-consuming with field sampling.With the development of hyperspectral remote sensing technology,not only the vegetation classification precision is greatly improved,but also make it feasible to estimate physiological and biochemical parameters of vegetations in large scale,quickly,dynamically and non-destructively.As an important part of the carbon cycle in ecosystem,photosynthesis of vegetations is the focus of remote sensing research.Net photosynthetic rate has been proved important for measuring photosynthetic capacity directly of vegetations,for the study of vegetation productivity,it is similar to gross primary productivity and net primary productivity.Leaf is the main photosynthetic organs,the higher net photosynthetic rate it has in the same condition,the better the leaf structure and function are.Therefore,it is of great significance to estimate leaf net photosynthetic rate of vegetation leaves via remote-based method for the evaluation of vegetation carbon sequestration and productivity.In this study,based on the leaves hyperspectral reflectance and photosynthetic data,several chlorophyll-related vegetation indices,eg:NDVI,EVI2,NDRE,WDRVI,CI,CIgreen,CIrededge,MTCI,multiplied by PARin were used to construct the leaf net photosynthetic rate inversion model of 3 crops:rice,rapeseed,wheat and 3 garden plants:citrus,ginkgo,camphor tree,in different environmental conditions.Comparing the precision and the sensitivity of models with different chlorophyll-related vegetation indices,the most optimal corresponding VI model was figured out,respectively.Meanwhile,analyzed the impacts of different conditions on the net photosynthetic rate retrieval,including nitrogen concentration,growth stage,moisture,incident photosynthetically active radiation.Finally,united inversion model for different vegetations was construct and the precision differences between models were compared.The results are as follows:(1)It was feasible to estimate leaf net photosynthetic rate by using chlorophyll-related vegetation indices multiplied by PARin.As for the rice,the rapeseed and the camphor tree,the retrieval models were high in accuracy with R~2 above 0.79;then for the wheat,the citrus and the ginkgo,the accuracy of retrieval models were relatively lower with R~2 above 0.53,which still could meet the need of remote-based inversion precision.(2)Different environment and phenological conditions did have a certain influence on net photosynthetic rate retrieval,for instance,rice with high nitrogen concentration had higher light use efficiency than those with lower nitrogen concentration,so the inversion model of rice should take the nitrogen concentration into consideration;and for the rapeseed,there were significant differences in photosynthetic characteristic between rapeseed during different growth stage,in pod stage,the light use efficiency of rapeseed leaf was lower,and the full time inversion model's accuracy was little lower than single period's;and for the wheat,there were also significant differences in photosynthetic characteristic between wheat in different water conditions,the inversion model of limited water feed was with higher precision than that of normal water condition,and the unified inversion model's accuracy decreased a lot compared to that in simple water condition;as for the garden plants,eg:citrus,ginkgo,there were significant differences in photosynthetic characteristic between them under different incident photosynthetically active radiation,leaves of citrus and ginkgo turned out to be high in light use efficiency in low incident light condition,so the inversion models of citrus and ginkgo should take the incident light intensity into consideration.(3)According to the 6 vegetations retrieval results,the vegetation indices which considered red edge reflectance,eg:CI,NDRE,CIrededge,MTCI,etc.performed best among the 8 chlorophyll-related vegetation indices,with higher determination coefficient and lower root mean square error,and could be used for unified retrieval in different environmental conditions and in different vegetation types.