Study of spectral characterization of electro-optical sensors and apply it to analyze multispectral and hyper spectral payloads

Abstract

Reliable and accurate spectral information for Remote Sensing electro-optical payload is backbone for correct understanding of target properties from payload imageries. The work presented in this thesis is aimed to understand and determine correct spectral properties of electro-optical (EO) payloads and utilize for remote sensing applications. The signal received at the payload during its on orbit operation is the result of convolution of target, earth atmosphere and spectral characteristics of the payload itself. Payload characterization is very important for its effective and proper utilization after launch. After development of electro -optical payloads three main characterization (Optical, Radiometric and Spectral characterization) will determine its best usage. Spectral characterization will determine correct instrument response to input electromagnetic radiation. Payload spectral response is the integrated response of all its components like optics, filter and detector. One of the requirement is to determine accurate spectral parameters of the payload like central wavelength, bandwidth and absolute payload response for a given input Electro-Magnetic Radiation. This knowledge leads the researcher to understand its proper advantage by knowing the target properties in the specified spectral range. In this regard, the main objective of this thesis is to study the spectral characterization of multispectral and hyperspectral type of electro-optical payloads and utilize it for various important applications like spectral parameters computation, Spectral Band Adjustment Factor(SBAF) determination for IRS sensors like Resourcesat-2,2A, OCM2, Chandrayaan-1 HySI Payload etc. Further image correction algorithm is also developed for spectral overlap correction for Mars Color Camera (MCC). Moment Method is very useful to determine spectral parameters of the EO sensor as it consider the signal in full spectral range and almost full signal entering in the payload is considered for computation. It provides more realistic results for both multispectral and hyper spe ctral payloads considered in this thesis. Further the study provided in this thesis, examined the variability in sensor response due to intrinsic difference in Relative Spectral Response (RSR) between different Indian remote sensing sensors and its implication. The detailed study is carried out for compensation of difference in spectral response of two similar type of sensor using the Spectral Band Adjustment factor(SBAF) method. The study showed that it is very important for intercalibration of two sensors, and required to be considered then only results of different satellite can be brought to a common reference scale. Now a days intercalibration among various sensors is very important to assimilate their data. Using this technique, various sensor data corresponding to wide range of temporal and spatial scale can be generated which is extremely important to earth resource monitoring, time series analysis and climate change study. The study carried out in this thesis shows that for sensor having almost similar band also give sizable difference in absolute radiance for different type of targets. This study become more important for the sensors having smaller bandwidth because their difference in absolute radiance vary much higher and it highly depends on target spectral response. So it becomes clear that same SBAF compensation cannot be applied for different type of targets. Particularly in the case of Ocean measurement where the percentage of signal contribution from ocean is very small (approx. 10% albedo) in the total signal received at sensor. Another study presented in this research is for correction of spectral overlap in the images received by MCC. This camera contain RGB Bayer pattern detector and spectral response of Red, Green and Blue pixel shows large overlap (particularly in green and blue), which reduces spectral information content in the image. In the thesis a methodology is developed to correct the data for spectral overlap. Study shows that after correction, spectral information content in the i mage enhances and different features on Mars, like dust cloud, polar ice cloud can be differentiated in a better way.