Analysis Of Vegetation Changes And Driving Factors In Guangdong Province Based On The Reconstructed NDVI

Previous studies have shown that:vegetation is a very important part of the earth's surface system,it has a very close relationship with the water cycle,carbon cycle and climate change in the earth,and it can also reflect human activities on the surface.In short,vegetation can reflect extremely rich information,which is very suitable for the study of environmental evolution.In remote sensing data,normalized differential vegetation index(NDVI)is one of the most commonly used data,which is used to characterize the spatial distribution of vegetation.As we all know,Guangdong is the province with the largest permanent resident population and the largest Gross Domestic Product(GDP)in China.In addition,with the decrease of available land,the relationship between people and land has become increasingly tense.Therefore,it is of great significance to study the interaction between NDVI and temperature,precipitation,population,GDP and other factors in it,as well as the temporal and spatial changes of NDVI and its driving factors,so as to get the goal of sustainable development of population,economy and environment in Guangdong.However,the detection of NDVI change and analysis of driving factors is affected by the quality of NDVI and the detection method of NDVI change.Therefore,to better detect the change of NDVI in Guangdong and analyze its driving factors,this study proposes some new methods,they are shown as follow:(1)the NDVI filtering reconstruction method combining"weighted iterative smooth spline"(ISSPW)and"weighted iterative filtering method based on time series decomposition"(IWSd),which can reconstruct high quality MODIS(moderate resolution imaging spectroradiometer)NDVI data in Guangdong Province;(2)integrated residual trend method(In-RESTREND)method and a simple NDVI change type division method,they can better detect the long-term NDVI changes in Guangdong Province.In addition,this study also includes other works which are shown as follow:(1)the correlation coefficients between the reconstructed NDVI and temperature,precipitation and population are calculated respectively,and the first-level and second-level zone of the NDVI dominated factors in Guangdong are classified by their maximum correlation coefficients;(2)The influence and interaction of temperature,precipitation and population on reconstructed NDVI in2000-2018 are quantified by using geographic detector;It also quantifies the influence of temperature,precipitation,population,traffic data,land cover data and gridding data of population and GDP on the reconstructed NDVI in Guangdong.Finally,the following conclusions are obtained as follow.Firstly,compared with weighted iterative savitzky Golay filtering,ISSPW,time-series decomposition based iterative filtering(ISd),IWSd,iterative Whittaker filtering(Iwhittaker),Iwhittaker after parameter optimization and combination of IWSd and ISSPW,the image reconstructed by combination of ISSPW and IWSd(ISSPW+IWSd)has the least noise;The sample curve reconstructed by ISSPW+IWSd fits the original NDVI better and describes more details of the curve;Moreover,ISSPW+IWSd get the largest correlation coefficient and the smallest root mean square error between the simulated noise and the original data,which indicates that its denoising effect is better and more stable.It also show that ISSPW+IWSd is suitable for NDVI reconstruction in Guangdong Province.Second,compared with RESTREND,In-RESTREND and(Time Series Segmentation and Residual Trend,TSS-RESTREND)both can detect more NDVI decreasing pixels.RESTREND underestimates the NDVI reduction from 2000 to2018.The significant decreasing pixels detected by In-RESTREND is 4.8 times of those detected by RESTREND.It shown that In-RESTREND can be used to detect the change of NDVI in Guangdong Province.According to the detection of In-RESTREND,the growth of NDVI in Guangdong Province is mainly in 2006?2018.The growth of NDVI are mainly distributed in northern Guangdong,western Guangdong and some cities in the Pearl River Delta;The decrease of NDVI mainly occurred in 2000?2005,which was mainly distributed in the Pearl River Delta.In terms of change type,the change type of NDVI in Guangdong is mainly monotonous growth type,accounting for 50.06%of the total area of Guangdong Province,followed by the first decrease and then increase type,accounting for 27.35%;The third is the intermediate weakening type,accounting for 9.39%;The other types account for a little area of Guangdong Province.Thirdly,in the four seasons,the dominant factors pixels mainly consists of population-dominated pixels.The positive and negative correlation pixels of dominant factors consists of the negative correlation pixels dominated by population.Population dominated negative correlation pixels are distributed in the whole province,while population-dominated positive correlation pixels are mainly distributed in Shenzhen,Huizhou,Guangzhou,Foshan and Heyuan.Among the dominant factors pixels,the area of temperature-dominated pixels rank the second.The temperature-dominated pixels are mainly distributed in northern and eastern Guangdong in summer.The area of precipitation-dominated pixels is the smallest,and it is mainly distributed in northern Guangdong in autumn.From the perspective of scenario hypothesis,insight into the pixel classification of the maximum correlation coefficient can give researchers and policy makers enlightenment:1.In the high value district of NDVI,the adverse human impact on NDVI should be reduced.2.In high density district of population,NDVI is positively related to population,indicating that population have a positive influence on NDVI.Further enlightenment is that people and nature can get along harmoniously,and in high-density population areas,it is possible to realize the sustainable development between population,resources and environment.Fourth,at the province scale and in the four seasons of 2000-2018,the influence of population is stronger than those of temperature or precipitation,and the influence of population shows a linear growth trend,while the influence of temperature and precipitation shows a linear weakening trend.In terms of interaction,the interaction between temperature and population is stronger than those between precipitation and temperature and those between precipitation and population.At the city scale,population is still the dominant factor,followed by temperature,which mainly affects more cities in spring and summer.The number of cities affected by population in autumn and winter is more than those in spring and summer;It is worth noting that NDVI of Pearl River Delta region is affected by both population and temperature in spring and summer,In spring,summer,autumn and winter,the NDVI of most cities in northern,Western and eastern Guangdong are mainly affected by population.However,the curve of influence of precipitation at city scale fluctuates greatly and its linear trend is weak.In terms of the multi factor analysis of NDVI change,at the provincial scale,secondary and tertiary industries in GDP(GDP₂₃)has the greatest impact on NDVI change,followed by land use layout.At the city level,the major factor is land use layout,and the combination of GDP₂₃ and climate factors(temperature and precipitation)affect the change of NDVI.Among them,the layout of construction land in land use layout has a greater impact on NDVI change.Combined with the results of NDVI change,it is found that in 2000?2005,the population influence(i.e.,Mainly land use layout change)mainly played the role of destroying vegetation(i.e.,NDVI reduction),and the main negative effects were concentrated in the Pearl River Delta.From 2006?2018,population influence mainly plays a role in promoting the growth of NDVI,and NDVI has increased in the whole province.