Study Of Vegetation’s Spatial And Temporal Distribution And Influencing Factors On The Tibetan Plateau

In recent decades, the simple methods such as spatialization, quantification, multiscalization, and time series analysis were used to study the structure, function, and the change process of the local vegetation and environment. It is summarized as two points:(1) Following the temporal dimension, the methods used before were primarily the descriptions of simple one-piece linear regression method belonging to classical statistics;(2) Following the spatial dimension, the methods used before were primarily the simple assessment of the spatial autocorrelation, Moran’s I, and the semi-variance function. This simple analysis were not compellent. In particular, rare applications were used to assess the vegetation greenness trend. On the other hand, the interactions between the vegetation greenness and the climate parameters were at its primary stage. Different geographical positions can result in different influence degree of the climate change. That is to say, rare works to study the spatial and temporal change, the spatial heterogeneity, and the spatiotemporal correlation synthetically by using quantitative approaches in temporal dimension and spatial dimension based on the remote sensing data. This paper studied the changes of vegetation greenness on the Tibetan Plateau in temporal dimension and spatial dimension based on the GIMMS AVHRR NDVI data from 1982 to 2006 and the MODIS NDVI data from 2001 to 2010 by one-piece linear regression method, piecewise linear regression method, spatial autocorrelation analysis, semi-variance function analysis and the fractal dimension analysis method, respectively. And with the help of the MERRA data(annual average temperature, annual total rainfall, annual total solar radiation) and the daily climate data of China Meteorological Station(annual average minimum relative humidity, Annual average daily maximum temperature, Annual average daily minimum temperature, annual total sunshine duration, annual average relative humidity, annual average temperature, annual average wind speed and annual total precipitation), this paper evaluated the characteristics of the spatial and temporal changes of the past thirty years comprehensively. This paper has completed the following contents and conclusions.(1) One-piece linear regression method is used to analyze the time series data. The research is carried out from aspects of the changes of the whole vegetation, the vegetation types and the regional vegetation. Experimental results reveal some tips. First is the whole vegetation greenness has the trend of increasing. Second is the vegetation types with a higher greenness have a higher sensitivity under the influence of climate change. Third is the effect of human being on vegetation is very significant in the study of regional vegetation.(2) In view of the condition that there are some abrupt changes in the time series data, a piecewise linear regression method is proposed to monitor vegetation greenness trend. The results show that the method may be superior to the one-piece linear regression method for the assessment of vegetation greenness method, especially where there are abrupt changes in the trend resulting from climate anomalies and disturbances.(3) The spatial autocorrelation analysis, the semi variance function analysis and the fractal dimension analysis of vegetation are carried out. The results show that vegetation has positive spatial autocorrelation and shows a clear high value and low value aggregation region. And the structural factors dominate in the total spatial variation(about 70%). The spatial distribution of vegetation is mainly distributed along the southeast to northwest direction, which is caused by the geomorphic features and the trend of the mountain here. The effect of local climatic conditions, local terrain and landforms, and local disturbances human activities on the spatial distribution of vegetation is increasing year by year.(4) In view of the non-stationary phenomena in the research data, a low-order quadratic polynomial fitting method for each pixel is proposed here to achieve the purpose of de-trending. The results show that the method successfully achieves the expected goal.(5) Study the annual MERRA data and the annual AVHRR data, and the correlation between the growing season daily climate data of China Meteorological Station and the growing season MODIS data, results show that at different spatial scales and different time scales, the correlation between vegetation and climate variables is basically consistent. And different vegetation types have different main climate driven variables.