Towards Identification of Relevant Variables in the observed Aerosol Optical Depth Bias between MODIS and AERONET observations

Abstract

Comparison of the aerosol optical depth values from two datasets, observed by satellite remote sensing Moderate Resolution Imaging Spec-troradiometer (MODIS), and globally distributed Aerosol Robotic Net-work (AERONET), show that there are biases between the two data products. In this paper, we present a general framework to identify the possible factors influencing the bias, which might be associated with the measurement conditions such as the solar and sensor zenith angles, the solar and sensor azimuth, scattering angles, and surface reflectivity at the various measured wavelengths, etc. Specifically, we performed analysis for remote sensing MODIS Aqua-Land data set, and used ma-chine learning technique, neural network in this case, to perform mul-tivariate regression between the ground-truth and the training data sets. Finally, we used mutual information between the observed and the predicted values as the measure of similarity. The set is then iden-tified as the most relevant set of variables. The search consists of a brute force method as we have to consider all possible combinations of the regressors to the neural network. The computations involves a huge number crunching exercise, and we implemented it by writing a job-parallel program. 1 ar