The Research Of Scale Effect Of Surface Reflectance And NDVI Of Remote Sensing Image

This paper takes part of the Shatian peninsula in Guangxi province as the researchobject and the selects the remote sensing images of Geoeye-1and Landsat7ETM+asthe data source. Firstly, use five kinds of scale effect evaluation indexes to analysis thescale effect of surface reflectance and normalized difference vegetation index (NDVI)of the whole study area in0.5m,2m and30m resolution, and then do more furtherresearch. In the study of the experimental area, this paper use0.5m panchromatic bandimage of Geoeye-1and field observation data to select seven typical features of thesample area and then carry out the analysis of the scale effect of surface albedo andnormalized vegetation index of different underlying surface types. In this paper, specificresearch contents and research results are as follows:(1) In the past, most of the research method of scale effect take one remote sensingsatellite imagery as the data source to study scale effect and when do a research of acertain scale there is often no corresponding spatial resolution of satellite data asreference. But this paper uses the two satellite actual access data as the reference remotesensing data in the scale effect study and divide the study of scale effect of surfacealbedo and normalized vegetation index into two part. One is the scale effect of thewhole area and the other is different underlying surface types of, which is from thewhole to relatively microscopic sight to carry on the comprehensive analysis and makesthe whole experiment more rigor and authenticity.(2) In the data preparation phase, in order to get the0.5m resolution image,thispaper makes full use of the Geoeye-1satellite data and use its panchromatic image with0.5m resolution and multispectral image with2m resolution for image fusion. The sixfusion transformation methods include HSV transform, Brovey transform,Gramm-Schmidt transform, principal components (PC) transform, color normalizedtransform (CN) and Pan-sharpening transform, and we use quantitative analysis andqualitative analysis to choose the best quality image fusion images with0.5m resolutionas the experimental data in this study. Because of the two remote sensors with differentspectrum characteristic, imaging geometry characteristic and the phase characteristic, there is necessary to do the normalization processing to reduce the difference ofexperiment research.(3) In the experimental results and analysis, from the perspective of macro andmicro we use five kinds of assessment indexes to do further research and analysis. Inthe study of different underlying surface type we select seven typical ground objects inthe study area, including mangrove forest, eucalyptus forest, artificial construction,cassava, bare land, arable land, water and so on. It is found that with the scale increasingall kinds of feature information gradually reduce, and the details of the selected sampleof grayscale gradually become less and less, so is even some information. And we alsouse spatial heterogeneity index to analyze the causes of scale effect.