PROJECT TITLE :
Joint Use of ICESat/GLAS and Landsat Data in Land Cover Classification: A Case Study in Henan Province, China
Lidar waveform features from the Ice, Cloud, and land Elevation Satellite/Geoscience Laser Altimeter System (ICESat/GLAS) and spectral options from Landsat Thematic Mapper (TM)/Enhanced TM Plus (ETM+) were used to discriminate land cowl classes for GLAS footprints in Henan Province, China. Fifteen waveform metrics were derived from GLAS knowledge while band ratios and surface spectral reflectance were taken from Landsat TM/ETM+. Random forest (RF) was utilized in feature selection and classification of footprints together with support vector machines (SVMs). The categories of classification included croplands, forests, shrublands, water bodies, and impervious surfaces. Compared with the employment of waveform or spectral options alone in land cowl classification, the joint use of waveform and spectral information as inputs improved the classification accuracy of footprints. An overall accuracy (OA) of ninety one% was achieved by either RF or SVM when features from both GLAS and Landsat sources were used increasing upon an accuracy of eighty five% if solely one supply was used. The high accuracy land cowl data obtained by the joint use of the 2 data sources may be used as further references in massive scale land cover mapping when ground truth is difficult to obtain. It is believed that the increase in accuracy is basically a result from the inclusion of the additional info of vertical structure offered by waveform lidar.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here