Atmospheric correction and image classification on MODIS images by nonparametric regression splines

Thesis Type: Doctorate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Civil Engineering, Turkey

Approval Date: 2014




In this study, two novel applications of nonparametric regression splines are introduced within the frame of remote sensing: (i) For the first time ever, atmospheric correction models are generated for moderate resolution imaging spectroradiometer (MODIS) images by using multivariate adaptive regression splines (MARS), and its recently introduced version, conic multivariate adaptive regression splines (CMARS). The obtained models are applied on the predefined test areas of twenty four different MODIS scenes. Simplified model for atmospheric correction (SMAC) algorithm, a radiative transfer-based approach, is also employed on the same test data. The performance of the MARS, CMARS and SMAC models are assessed against MODIS surface reflectance products. (ii) Additionally, implementation of MARS algorithm in image classification for snow mapping on MODIS images is demonstrated within a well-elaborated framework. The relation between the variations in MARS model building parameters and their effect on the predictive performance are presented in various perspectives. Performance of MARS in classification is compared with the traditional maximum-likelihood method. For the atmospheric correction, results reveal that MARS and CMARS approaches over perform SMAC method on all test areas. In classification, significant improvement in the classification accuracy of MARS models is observed as the number of basis functions and the degree of interaction increase. On three image sets out of four, the MARS approach gives better classification accuracies when compared to maximum-likelihood method.