Semih Kuter

Inversion of top of atmospheric reflectance values by conic multivariate adaptive regression splines

Spatial technologies offer high flexibility to handle substantial amount of spatial data and wide range of modelling capabilities. Remotely sensed data are the most significant data source used in spatial technologies. However, it is often associated with uncertainties due to atmospheric effects (i.e. absorption and scattering by atmospheric gases and aerosols). Methods based on rigorous treatment of radiative transfer models still have some drawbacks in the inversion of top of atmospheric reflectance values to surface reflectance values on large numbers of satellite images. In this paper, our aim is to represent a more flexible (adaptive) approach for the regional atmospheric correction by employing nonparametric regression splines within the frame of inverse problems and modern techniques of continuous optimization. To achieve this objective, atmospheric correction models obtained through conic multivariate adaptive regression splines, which is an alternative method to multivariate adaptive regression splines by constructing a penalized residual sum of squares as a Tikhonov regularization problem, are applied on a set of satellite images in order to convert the top of atmospheric reflectance values into surface reflectance values. The results are compared with the ones obtained by both multivariate adaptive regression splines and a radiative transfer-based method.

Modern Applied Mathematics for Alternative Modeling of the Atmospheric Effects on Satellite Images

Nonparametric regression and classification techniques are mostly the key data mining tools in explaining real life problems and natural phenomena where many effects often exhibit nonlinear behavior. The remotely sensed earth data collected by earth-observing satellites is degraded due to the absorption and scattering of solar radiation by atmospheric gases and aerosols. In order to use these data for information extraction, they must first be corrected for the atmospheric effects. Recent methods based on radiative transfer modelling still have many challenges including achieving high accuracy and developing real-time processing capability of large numbers of satellite images acquired with high temporal resolution and Large Field of View instruments. In this chapter, two state-of-the-art nonparametric tools, Multivariate Adaptive Regression Splines (MARS) and its successor Conic Multivariate Adaptive Regression Splines (CMARS), are reviewed within the frame of an earth science example. Both methods are utilized for the atmospheric correction of five sets of MODIS images taken over European Alps. The Simplified Method for Atmospheric Correction (SMAC), a simplified version of 6S radiative transfer model, is also applied on the image data sets for the removal of atmospheric effects. The performance of the models was evaluated by comparing their results with the MODIS atmospherically corrected surface reflectance product in terms of RMSE. Although MARS and CMARS approaches produce similar results on the data sets, they both outperform SMAC.

An Alternative Method for Snow Cover Mapping on Satellite Images by Modern Applied Mathematics

Continuous monitoring of snow cover is very crucial since the extend and amount of snow are key parameters for many processes closely related to ecology and climatology. Measuring the extend and amount of snow by in situ measurements are not always practical and possible due to operational and logistic reasons. Since 1960s, images taken by earth-observing satellites have been extensively used to monitor snow cover, and many parametric and nonparametric image classification methods have been proposed and applied for snow cover mapping. In this study, a novel application of nonparametric regression splines is introduced within the frame of modern applied mathematics and remote sensing. Implementation of multivariate adaptive regression splines (MARS) in image classification for snow mapping on moderate resolution imaging spectroradiometer (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 represented in various perspectives. Performance of MARS in image classification is compared with the traditional maximum-likelihood (ML) method by using error matrices. 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 ML method.