MSGIS Portfolio
Yiwei Wang
Remote Sensing
Green coverage area in Salt Lake City
Introduction:Vegetation plays an important role in people's life. The percentage of urban green area is an important index for every city. This index direct influence the quality of life. In my project, the study and research area was a part of Salt Lake City. The analysis was based on ENVI software, the main method was supervised classification. The result would be the percentage of urban green area in the part of Slat Lake City.
Method and Result:The method of this project includes three parts: preprocessing, analysis, and evaluation. The preprocessing method was to convert ".tif" file to ENVI standard. The analysis method were contrast stretching,supervised classification, and class statistic. The evaluation method was accuracy assessment. Figure.1 is a digital orthophotography of a part of Salt Lake City, Figure.2 shows how does it look like after classification. The green areas include grass and tree, so the total percentage of green urban area in Salt Lake City in my project was 31.5%. For evaluating result, the overall accuracy is 72.46%, the kappa coefficient is 0.6327, the user’s accuracy for grass, tree, building, road, water, and bare surface are 79.7%, 86.7%, 75.4%, 59.4%, 76.7%, 84.9%. The producer’s accuracy are 88.5%, 81.7%, 63.9%, 65.6%, 69.3%, 89.6%. The value of overall accuracy and kappa coefficient was moderate. In the supervised classification part, grass and tree were easily confused, because the color of these two classes are similar. Grass and bare surface were easy to confuse, grass grows on bare surface, it was hard to divide them very clearly.
Skills(Cartography and Graphic Design): The project used ENVI software to process image. The skills include rescale image data into brightness values can make a drastic difference in the way that the image appears,(ENVI) cluster pixels in a dataset into classes corresponding to user defined training classes, compute pixel count and percentage per class, basic statistics, and histograms for all bands on classifications,(EVIS, 2014) and assess by comparing the classification with some reference data that was believed to accurately reflect the true land-cover.( Yale University, 2015)
View the paper
Figure.1
Figure.2
References:
ENVI. (2014) Laboratory Exercises in Image Processing: Contrast Stretching