| Vegetation plays a crucial role in various ecological cycles on Earth,and it’s changing trends also reflect the process of natural environment evolution.Previous researchers often used normalized vegetation indices to study vegetation changes.However,due to various factors such as clouds,fog,and imaging conditions,image data m ay have noise and data loss,making precise analysis of NDVI changing trends and influencing factors difficult.Therefore,to effectively detect the trend of NDVI changes in Wenshan Prefecture and analyze its influencing factors,a weighted SG filtering method based on the Dixon test and spatiotemporal information fusion(SGDST)was proposed,which achieved high-quality reconstruction of MODIS/NDVI data in the region.Subsequently,the comprehensive residual trend method(In-RESTEND)and an NDVI change type classification method were used to detect the long-term NDVI change trend in the region.In addition,calculating the Pearson correlation coefficient between NDVI and temperature,precipitation,and population,and dividing the dominant influencing factors of NDVI in Wenshan Prefecture by season,can better reflect the impact of temperature,precipitation,and population on NDVI in Wenshan Prefecture in different seasons.The following conclusion was ultimately reached:Firstly,after comparing four fitting methods,it was found that the SGDST method had the least residual noise in the reconstructed image;When fitting curves,the SGDST method can accurately reconstruct the time series of consecutive missing data for multiple periods,effectively preserving the detailed features of local peaks and valleys,and the spline curve is closer to the original NDVI curve.In addition,in the quantitative evaluation of simulation data,compared with the other three methods,the SGDST method with RMSE and MAN le ss than 0.0853 pixels accounts for the largest proportion of the total fitted pixels and the highest fidelity.In terms of model comparison,the SGDST method has the highest distribution of AIC and BIC low-value regions,making the model more stable.In summary,the SGDST method is superior to the other three methods in the Wenshan Prefecture area and is more suitable for NDVI reconstruction.Secondly,the comprehensive residual trend analysis method outperforms linear regression equation slope,MK test,and RESTREND in detecting NDVI significance pixels.The results show that from 2010 to 2020,the NDVI of Wenshan Prefecture showed a main growth trend,with a main growth period from 2014 to 2017.The three counties in the northern part of Wenshan Prefecture had the largest growth rate.The decrease in NDVI is scattered in various counties and urban areas,with a very small proportion.From the perspective of change types,the NDVI changes in Wenshan Prefecture are mainly monotonically increasing,accounting for 92.8% of the total pixel area;The second type is the first ascending and then descending type,accounting for 6.79%;The continuous decline type accounts for approximately 0.82%.The proportion of other types is relatively small.Finally,in the study of NDVI dominant influencing factors,the pixel proportion of precipitation positive correlation influencing factors is the highest in the first three seasons.In winter,the pixel proportion of the temperature-positive correlation factor is the highest.The proportion of precipitation-influencing factors in spring,summer,and autumn is 81%,78%,and 52%,respectively,while the proportion of winter temperature-influencing factors is 77%.In the secondary classification,the main positive correlation factor of precipitation is the grid pixel,which accounts for 78% in spring,78% in summer,47% in autumn,and only 6% in winter.However,the temperature-positive correlation factor pixel accounts for 76% in winter.In the main urban areas with high population density,there are also pixels with a positive correlation between NDVI and population,indicating that population factors also have a positive impact on NDVI. |
PDF Full Text Request