Ground Biomass Simulation Of The Grassland In Hejing County Of Xinjiang Based On Vegetation Index

The study of grassland biomass is an important aspect of grassland ecology and ecological economics,the acquisition of grassland vegetation dynamics and the real-time accurate the distribution of grassland biomass on the ground,is the precondition of reasonable use and protection of grassland resources.Hejing grassland in Xinjiang as the study area,this study of grassland as the research object,the Landsat image as the main remote sensing data,from July 2014 to August as the modeling data,field survey sampling point biomass data using RMB as binary model first,using Arcgis superposition analysis function,grading estimating grassland vegetation coverage,four different periods based on the center of gravity migration analysis method,study the variation characteristics.Secondly,based on the vegetation index NDVI,RVI and grassland biomass measured values of correlation analysis on the ground,respectively,set up a linear model,index model,the quadratic polynomial model,through the SPSS statistical analysis,model precision inspection,determine the applicable grassland aboveground biomass estimation of optimal model in the study area,using the optimal model of remote sensing inversion of grassland biomass on the ground.1.Based on pixels binary model estimate grassland vegetation coverage with the field measured values of linear correlation analysis,get the correlation coefficient is 0.7267,can satisfy the requirement of the vegetation coverage area scale survey,using this method to study the grassland vegetation coverage on Hejing is feasible.The overall distribution of grassland vegetation in the study area of northwest,southwest,east is relatively poor.The average area weighting of vegetation coverage in 4 different periods were respectively 4.2851,4.3042,4.4252,3.7524,which had a trend of increasing-increasing-decreasing.The overall average vegetation coverage was reduced by 0.5327 from 2010 to 2015,while the reducing of vegetation coverage in 2015 was largest compared with that in 2010,which was 15.20%.2.From the year 2000 to 2015,the conversation rate of coverage of 5 different levels from low to high were respectively 54.82%,60.08%,5.52%,266.82%,772.37% while the conversation rate of level III vegetation coverage was minimum,which was 5.52%,and the conversation rate of level V vegetation coverage was maximum,which was 772.37%.From the year 2000 to 2015,the percentage of level I vegetation coverage area had a trend of decreasing-increasing-decreasing and had maximum decreasing of 798164.10hm2,while the coverage area in 2015 was reduced by 57.23% than that in 2010;the percentage of level II vegetation coverage area had a trend of constant increasing,and had maximum increasing of 473682.07hm2,while the coverage area in 2015 was 159.75% of that in 2000.the vegetation coverage in the study area degraded by 1376348.49hm2 during 2000~2015,which accounted for 49.42% of the total area.The degraded area was 1261164.41hm2 larger than the increased area of vegetation coverage and 11.95 times as much as the latter.From the year 2000 to 2015,the center of level I vegetation coverage type moved 49.78 km to northwest while that of level III moved 4.24 km to northeast,and that of level II,level IV and level V moved to southwest,with respective distances as 38.07 km,53.00 km and 83.35 km.3.Based on NDVI and RVI two planting index,and the correlation between the measured 67 sampling points,6 kinds of regression model is established,including RVI-quadratic polynomial model and the grassland biomass on the ground of the highest correlation,the correlation coefficient reaches 0.911,forecasts determine coefficient is 0.830,forecast accuracy of 85.31%;Followed by RVI-linear model,the correlation coefficient is 0.908,forecasts determine coefficient is 0.829,the forecast precision is 78.52%;NDVI-quadratic polynomial model,correlation coefficient is 0.907,forecasts determine coefficient is 0.822,the forecast precision is 81.22%;NDVI-linear model,the correlation coefficient is 0.903,forecasts determine coefficient is 0.814,the forecast precision is 77.01%;NDVI-exponential model,the correlation coefficient is 0.877,forecasts determine coefficient is 0.768,the forecast precision is 72.86%;RVI-exponential model,the correlation coefficient is 0.854,forecasts determine coefficient is 0.728,the forecast precision is 70.11%;Through a P < 0.001.RVI-quadratic polynomial model is the model of the correlation coefficient,correlation factor and the predictive model precision is higher than other models,the study area is the optimal model of remote sensing monitoring grassland aboveground biomass,inversion model: y=62.121 x2+1146.7x-377.66,R2=0.830,n=67).Using the optimal model of the performance of the study area grassland biomass spatial distribution characteristics of the ground,is consistent with the spatial distribution characteristics of vegetation coverage in the study area.