Research On Estimation Of Gross Primary Productivity In Heihe River Aasin By Multi-model Integration

As the belt and road continues to deepen,sustainable development had gradually become a consensus among the countries along the line.As an important part of the Hexi Corridor,the change of ecological environment in Heihe River Basin had become a research hotspot at home and abroad,therefore.The gross primary productivity was a key parameter of material and energy flow,and an important indicator for the quality of biosphere and ecosystem.The accurate estimation of the gross primary productivity in Heihe River basin would not only accelerate the construction of ecological civilization in Hexi Corridor,but also play an important role in promoting the prosperity and development of the belt and road economic open area.In order to simulate the gross primary productivity in Heihe River Basin accurately,and explore its distribution characteristics and causes deeply,this paper taked the growth season of 2015 as the research time,and carried out three steps successively.Firstly,CASA model,VPM model and TL-LUE model which driven by multi-source data were used to estimate the gross primary productivity in Heihe River Basin.Then,multiple linear regression algorithm and support vector machine algorithm were used to integrate multiple single models to optimize the estimation of gross primary productivity in Heihe River Basin.Finally,the best model of gross primary productivity was determined by using quantitative analysis index,the change rules of gross primary productivity were revealled by using spatio-temporal distribution map,the influencing factors of gross primary productivity were studied by using fit scatter diagram.Among the above steps,the multi-source data included MODIS product data,meteorological data with different scales,landcover data of Heihe River Basin and elevation data,while the quantitative analysis index included judgment coefficient R~2,the bias coefficient Bias and the root mean square error RMSE.Based on the measured data of the eddy station and the MODIS GPP products,the accuracy of estimation which originated from the single model and the integrated model was evaluated respectively.After regression analysis and comparative analysis,the research results were summarized as follows:(1)The CASA model,VPM model and TL-LUE had higher applicability than MODIS GPP products in Heihe River Basin,and can be used to simulate the gross primary productivity in Heihe River Basin with 8-day time resolution accurately.For the whole basin,it is said that the judgment coefficients increased by at least 20%.and the root mean square errors reduced by at least 57%,while the bias coefficients reduced by at least 69%.(2)Integrating single model with multiple linear regression algorithm and support vector machine algorithm can reduce the estimation error of gross primary productivity effectively.For the whole basin,compared with MODIS GPP,it is said that the growth rates of judgment coefficient increased by 5%and 10%,and the reduction rates of root mean square error increased by 11%and 19%,while the reduction rates of bias coefficient all increased by 30%.(3)In the simulation of the gross primary productivity for maize,willow,reed and alpine meadow in Heihe River Basin,support vector machine integrated model performed best among all models.Compared with MODIS GPP,the increase range of judgment coeffi-cient was 7%to 527%,and the reduction range of the root mean square error was 56%to86%,while the reduction range of the bias coefficient was 90%to 99%.(4)In the Heihe River Basin,dominated by temperature and water content,affected by vegetation types and vegetation growth,the accumulative gross primary productivity in the growing season presented a spatial distribution which the highest in the middle reaches,the second in the upper reaches and the lowest in the lower reaches,while the gross primary productivity with 8-day time resolution presented a variation trend which increased first,then decreased,and peaked in late July.(5)In the Heihe River Basin,the gross primary productivity was affected by the growth environment and vegetation growth.With the increasing in temperature,water and vegetation growth,the gross primary productivity increased,too.It is said that the temperature,precipitation,leaf area index,normalized vegetation index and enhanced vegetation index can account for 84%,81%,97%,86%and 92%changes in the gross primary productivity,respectively.