NPP Estimation Of Qinghai Tibet Plateau Alpine Grassland Based On Multi Model

Qinghai Tibet Plateau?QTP?is known as the"third pole"of the world because of its high altitude and large area,which has a major impact on regional climate and the global environment.Accurate estimation of NPP is the premise of NPP research on climate change.Because of the special permafrost and alpine environment of the Qinghai Tibet Plateau,the mainstream NPP estimation model is not applicable here,and the spatial-temporal resolution of the estimation results is also quite different.Therefore,it is urgent to improve the NPP estimation model or to find the most suitable NPP estimation method in different scenarios.In this paper,Biome-BGC model,BEPS model and Random Forest are selected from mechanism process model,Coupling model and Machine Learning.The mechanism of Biome-BGC is strong,but the unique freeze-thaw cycle process in permafrost region of QTP is not considered in the simulation process,and it is only applicable to stations.In this study,the Biome-BGC is improved;the advantages of BEPS are that it can couple remote sensing data;random forest can process input samples with high-dimensional characteristics,and the default value can get better results.The NPP of Alpine Grassland in the QTP from 2000 to 2018 was estimated by these three models.The results were compared from the estimation accuracy,input parameters,spatial-temporal resolution and ease of use of the model.The conclusions are as follows:The accuracy of NPP estimation is improved by the improved Biome-BGC.The verification accuracy of annual NPP estimation results is in the order of Random Forest,improved Biome-BGC and BEPS.The time resolution of Random Forest estimation is year,the time resolution of BEPS output is 8 days,and the time resolution of Biome-BGC output is every day.Among the three models,Random Forest is the easiest to use,BEPS is in the middle,and Biome-BGC is the most demanding.The selection of the model depends on the needs.If a higher time resolution is needed,the Biome-BGC will be used,if the requirement of faster estimation speed for time resolution is not high,the Random Forest will be used,and the BEPS will be selected based on the combination of the two.The temporal and spatial distribution characteristics of NPP in Alpine Grassland of QTP are as follows:From 2000 to 2018,the annual NPP of Alpine Grassland in the QTP estimated by the Biome-BGC model is 128.86 gC·m^-2·a^-1,and the total NPP was 253.83TgC·a^-1.The average annual value calculated by BEPS is 141.39 gC·m^-2·a^-1,and the total amount is 278.51 TgC·a^-1.The average annual value of RF is 119.91 gC·m^-2·a^-1,and the total is 236.21 TgC·a^-1.From 2000 to 2018,the annual NPP of Alpine Grassland in QTP decreased from southeast to northwest.The NPP estimated by the three models in the study period showed an increasing trend,with an average annual growth rate of 0.2 gC·m^-2·a^-1,0.34 gC·m^-2·a^-1and 1.39 gC·m^-2·a^-1,respectively.In the estimation results of Biome-BGC,the NPP increased by 32.30%in alpine grassland,17.02%in BEPS and 25.87%in RF.The annual variation of NPP in Alpine Grassland of QTP can be divided into four stages.In the first stage,NPP increased gradually,and in the second stage,NPP developed steadily and reached the highest value.In the third stage,NPP decreased rapidly and the rate of decline was higher than that of the first stage.In the fourth stage,NPP is relatively stable,basically 0.In the interannual response of NPP to climate factors,24.69%of NPP was temperature significant,4.54%of NPP was precipitation,and 7.04%was radiation.The annual response of NPP to climate factors is as follows:the vegetation range begins to expand in 31-61 days,and temperature and precipitation control the growth of Alpine Grassland on the QTP;the correlation coefficient is very high in 91-121 days due to the joint effect of temperature,precipitation and radiation;the influence of radiation and temperature on the growth of alpine grassland is weakened in 121-151 days,and the influence of precipitation on vegetation growth is gradually From 181 to 211 days,the growth of alpine grassland gradually changed from temperature and precipitation control to temperature and radiation control;from 271 days,the range of vegetation began to gradually reduce;from 331 to 361 days,the range of vegetation retreated to the southernmost part of the Western Sichuan region and the southern Valley region of Tibet,which was affected by both temperature and water drop.