Temporal And Spatial Dynamics Of Alpine Grassland Biomass And Grassland Livestock Balance And Its Influential Factors In The Three River Headwaters Region

The Three River Headwaters Region(TRHR)is the headstreams of Yangtze River,Yellow River and Mekong River.In China,this region is one of the regions that eco-environmental security,which is a most sensitive issue.Due to climate change and irrational utilization,the grassland in TRHR is experiencing degradation,which poses serious threats to the ecological security of the TRHR and even the whole nation.Therefore,in the TRHR region,the development of remote sensing monitoring of grassland biomass in TRHR,the evaluation of current situation of animal husbandry,and the analysis of factors that influencing grassland biomass are of great importance to the regional ecological conservation practice.In this study,factors that are closely related to the grassland Aboveground Biomass(AGB),such as soil,meteorology and vegetation variables are used in single-factor and multi-factor modeling,to explore the utilization of artificial neuron network(ANN)in the estimation of grassland biomass.By comparing model accuracy,the optimal model is identified and used to estimate grassland AGB in TRHR region during 2001-2016.Then the spatial and temporal variations of AGB for different grassland types,climatic zones,basins and administrative areas are analyzed and discussed.Based on the analysis of proper livestock carrying capacity and proper stocking density,and the actual number of livestock penned at the end of year,the status of grassland livestock balance in TRHR are analyzed,driving forces of such balance are studied.Results show that:1)The relationships between Landsat vegetation indices and AGB show that NDI7 has a better correlation with AGB,with a higher R2 of 0.3480 than NDVI(R2=0.3).The exponential models based on NDI7(y=227.57e3.102 x,R2=0.3704,RMSE=575.781 kg DW/ha)and NDVI(y=192.33e2.7613 x,R2=0.3644,RMSE=576.211 kg DW/ha)have higher accuracy.2)Single-factor models based on one of the factors that are in close association with AGB can only reflect 16.24 %-44.9% portion of the grassland AGB variation in TRHR.Inversion model based on normalized difference vegetation index(NDVI)reflects 44.9% portion of the AGB variation(RMSE=787.7 kg DW/ha),which is the highest in these single-factor models.The various multi-factor linear AGB regression models are based on different combination of factors,including grassland coverage,height,longitude/latitude and NDVI,etc.Fitting accuracy of these models ranged between 40.14% and 58.03%,with RMSE of 577.4-689.8 kg DW/ha.Fitting accuracy of various multi-factor nonlinear regression models ranged between 46.63-62.10%,with RMSE of 552.7-654.5 kg DW/ha.Compared to these models,ANN-based model is much more suitable for inversion modelling of AGB(R2=0.7699,RMSE = 429.1 kg DW/ha).3)Monthly averaged AGB are 594.56-751.43 kg DW/ha in the grassland growing seasons during 2001-2016,with a maximum AGB of 1272.2-1739.6 kg DW/ha during the peak period of growth.Spatial distribution of AGB shows significant regional differences,with a decreasing trend from southeast to northwest.The AGB reaches 3000 kg DW/ha in some locations of southeastern area,and is very low(less than 500 kg DW/ha)in some locations of northwestern area.The hydrothermal factors have significant effects on the AGB in TRHR,with apparent differences among the different grassland types,basins,and administrative areas.4)In the TRHR,during the past 16 years,the grassland AGB value is mostly stable,with minor restoration,with 53.74% areas of stabilization,37.31% areas having a restoration tendency,and 8.95% areas having deteriorating tendency.There are significant differences among the different grassland types,climatic zones,basins and administrative areas.In general,by the analysis of the sustainability in grassland biomass changes,39.06% areas in TRHR are predicted to have an uncertain trend,28.57% areas in TRHR are predicted to hold a stable AGB value,29.96% areas in TRHR are predicted to have a restoring trend and 5.43% areas in TRHR are predicted to have a deteriorating trend.Thus,grassland AGB in TRHR shows an overall stable state in general,with some local deterioration.5)During 2001-2016,proper carrying capacity of grassland in TRHR is 130.76×104-178.26×104 SU,and the average carrying capacity is 146.31×104 SU.The proper stocking density is 0.53-0.72SU/ha,and the average proper stocking density is 0.60 SU/ha.There are large differences in proper carrying capacity and in proper stocking density among different grassland types and administrative areas in TRHR.The spatial distribution of grassland stocking density shows a gradually decreasing trend,from southeast to northwest in TRHR,and the spatial heterogeneity is obvious.6)There is no overgrazing in in both Golog Prefecture and Yushu Prefecture during 2004-2015,but the grassland livestock balances in these two prefectures have apparent differences.The Balance of Grassland and Livestock Index(BGLI)of Golog Prefecture have a continuously decreasing trend in the past 12 years,and stays on a grassland livestock balance state since 2009.While the actual carrying capacity of Yushu Prefecture is increasing in the past 12 years,the carrying capacity status keeps balanced only during 2004-2010,the BGL are 20.66%,30.73% and 49.7%,respectively in 2011,2014 and 2015,the grassland appears to be obviously overgrazing.7)The annual mean temperature and precipitation show increasing trend in TRHR in the past 16 years.The correlation between precipitation of May and grassland AGB is the highest,with an r of 0.53;there is a high correlation between temperature of October and grassland AGB,with an r of 0.58.The results from grazing pressure analysis indicate that Banma,Zeku,Jiuzhi,Dari and Gande counties have a high grazing pressure.The grazing practices in Qingshui River and Xinghai counties only have a relatively small effect on the grassland AGB.The implementation of ecological protection projects effectively improved the ecological environment in TRHR,it has a positive impact on the protection and restoration of grassland eco-environment,but there are still severe challenges in current grassland ecological protection.