The Application Of ETM Data In Estimating The Area Of Small-region Vegetation In The City

Vegetation coverage , the ratio of vegetation occupying a unit area , is a very important parameter in the development of climate and ecological models . On-ground filed work surveys of vegetation coverage are time consuming and expensive and produce low-precision results . Estimation of vegetation coverage from remote sensing data may be a more efficient approach . In this study , two methods were used to estimate the urban vegetation coverage from remote sensing data : automatic classification of merged TM images and SPOT images ; interpretation of mixed pixels of TM images.On the basis of geometric correction for remote sensing images data , detailed character analysis was conducted for the TM images . Several image transformations which are linear scale transformation , ratio processing , principal components transformation , tasseled-cap transformation and minimum noise fraction rotation (MNF transformation) were then implemented . Two indexes was calculated to estimate the best bands union for color combination , one is Optimum Index Factor (OIF , the sum of standard deviation divided by the sum of correlation coefficient.) , the other is the determinant of the co-variance matrix . It can be seen from the result that for color combination the original optimal bands were TM 4, 3, 7 and TM 4, 3, 5 , the best mixed images were MNF1 , BR and NDVI.After being enhanced in many ways , the SPOT panchromatic band were merged with TM 4,3,7 through five different data fusion algorithms , they were weighing fusion , principal components transform , K-T transform , IHS transform and Brovey fusion . And then maximum- likehood classifier was used to have these fusion images classified automatically . We estimated the fusion images in both subjective factors and objective factors . The objective parameters involved entropy , average gradient , spectral divergence , correlation coefficient and the accuracy of the classification . The result of the estimation demonstrated that the fused images had higher spatial resolution while maintaining the basic spectral contents of the original TM images , and the visual effect and the accuracy of the classification bad beengreatly enhanced . Among the five fusion images the Brovey image had the highest classification accuracy , it was 99.5% , and it was high enough for the acquirement of the vegetation coverage of the city.Three methods of interpreting mixed pixels were used to TM images , linear method , logistic nonlinear method and neighbor-field-based mixed pixel interpretation method , to extract the vegetation information , especially the small-region vegetation . The result showed that,. making full use of the spatial information of the structure , the neighbor-field-based mixed pixel interpretation method made the best effect than the other two methods , it got a accuracy of 96.7%.It can be showed from this research , image fusion and mixed pixel interpretation can both be used in extracting the small-region vegetation information of the city and can acquire satisfactory accuracy . To this area being studied , the optimal fusion method was Brovey method and the best mixed pixel interpretation method was neighbor-field-based mixed pixel interpretation . The technology of image fusion and mixed pixel interpretation bave wide prospects in forestry remote sensing.