Study on stability of reflectance characteristics of natural features for calibrating remote sensing data

Abstract

The development of satellite technology in Malaysia is to ensure the availability of satellite imaging for the requirement of local remote sensing user. Apparently, low-quality images were produced from local remote sensing satellite. The vicarious calibration technique can be adopted to provide post-calibration opportunity to overcome this problem. However, this approach has constraints in defining well establishing target bodies. Currently, reflectance based calibration approach had been used in the measurement but using only several test site as a calibration target. Malaysia with its tropical condition should be capable to provide natural bodies that stable enough to serve as calibration target. The main purpose of this study was to develop a spectral library for selected spatial features. The spectral library will serve as a reference for the calibration of the image data generated by the camera. Visible and near infrared (NIR) bands from Landsat 7 and SPOT XS images were used in this study to obtain reflectance value. These reflectance values were collected on the selected targets that are water bodies, dark dense vegetation, manmade features and cloud body. Reflectance curve of each features were compared with ASTER spectral library as reference. From the comparison it shows that reflectance curve from Landsat 7 and SPOT XS give almost similar reflectance curve with reference spectral library. The range of reflectance value for water features is from 0.018 to 0.028, vegetation features range from 0.01 to 0.5 and man-made features range from 0.05 to 0.4. Cloud features give range of reflectance values from 0.2 to 1. Statistical distribution for spectral reflectance was determined to calculate the mean, variance and standard deviation. These values were used to determine the minimum and maximum values of spectral reflectance for each feature. Graphs were plotted to show the interval of each features based on bands. From these graphs, there are maintain range of interval level which indicate that the measured reflectance is reliable enough for information extraction of the spectral library.