| The Daning river basin is located in the core of the three gorges reservoir area and has sufficient water and heat conditions,so it is suitable for the growth of vegetation.It is surrounded by high mountains in the east,north and west and is connected to the Changjiang River in the south,so the ecological environment is relatively independent.In recent years,the area's vegetation cover has changed greatly and the ecological environment has become more sensitive because of the three gorges project.At present,the changes of regional vegetation are mostly concentrated in large areas and a few studies in small areas only focus on the changes of low time bucket,so we seldom consider the dynamic change process.This driving force focuses on the effects of weather,topography and other natural factors.Human factor research involves less and mostly adopts correlation analysis and linear regression methods,so it is unable to accurately describe the relationship between vegetation coverage and many complex factors.As one of the most efficient and accurate models of machine learning,XGBoost can effectively analyze the relationship between dependent variables and complex independent variables,so that it has great potential in ecological comprehensive analysis.Therefore,this paper took the Daning river basin as the research object,and used MODIS NDVI remote sensing data,and obtained the vegetation coverage change process from 2000 to 2015 by using the obtained data for mixed image dichotomy and trend analysis.At the same time,we combined the correlation analysis for it with the main climate factors(temperature and precipitation)response characteristic and application of XGBoost model analysis in the study area climate factors(temperature and precipitation),human activities(population,GDP),and terrain factors(elevation,gradient,slope direction)to analyze the the influence of the spatial distribution of vegetation.The main research work and conclusions of this paper are as follows:(1)According to the analysis of the temporal and spatial variation of NDVI,the average NDVI of the Daning River Basin was 0.72 from 2000 to 2015,and the NDVI value of 96.19% was between 0.5 and 0.83.The overall NDVI of the basin was an increasing trend,and the area of positive change of NDVI accounted for 96% in the total area of the basin;the increase of NDVI in the middle and lower reaches on the Daning River was more obvious;by comparison,all the months of the NDVI month interannual variation in the basin.As with the annual mean annual variation,there was a clear increase trend.(2)By comparing the soil,vegetation measured spectra and remote sensing methods to estimate the NDVI value,the measured spectra was more suitable for the dynamic study of vegetation coverage in small watersheds,and the automated and efficient calculation of vegetation coverage was accomplished by means of Python script.(3)From the past 16 years,the annual vegetation coverage of the Daning River Basin was between 0.74-0.86,and the overall average vegetation coverage was 0.81.The plant coverage rate was increasing steadily during this period.During the study period,the medium-high coverage in the basin gradually changed to the spatial distribution dominated by high vegetation coverage.More than 95% of the regional average annual vegetation coverage was improving,and the improvement in the lower reaches of the river valley was more obvious;the degraded areas were scattered and there was no obvious spatial law.Human production and life affected the volatility of vegetation.(4)From the time correlation about the influence of temperature and rainfall on the change of vegetation cover in the Daning River Basin,the annual temperature and vegetation coverage showed a weak negative correlation,and the temperature decline corresponded to the increase of vegetation coverage.The increase of vegetation coverage had a weak inhibitory effect on temperature;According to the volatility correlation proposed in this paper,it was found that there was strong volatility correlation between precipitation and vegetation coverage,indicating that precipitation in the basin had a strong positive correlation with the increased rate of vegetation coverage.Therefore,there was no direct linear correlation between the increase of vegetation cover in the Daning River Basin and meteorological factors such as temperature and precipitation.The reason for the increase of coverage was most likely caused by human ecological restoration and protection policies.(5)From the impact of quantitative factors on vegetation cover,the XGBoost model was used to analyze the vegetation cover.In the spatial autocorrelation,the vegetation coverage during the study period was mainly affected by the precipitation conditions.The correlation between the predicted vegetation coverage and the true value of the model was 0.780,and the RMSE was 0.019,which had higher simulation accuracy.According to the best accuracy of the model,the importance of the seven quantitative factors to vegetation cover was precipitation,population,GDP,elevation,slope,temperature and aspect. |
PDF Full Text Request