A method is presented that attempts to isolate the relative magnitudes of various error sources present in common algorithms for inverting the effects of atmospheric scattering and absorption on solar irradiance and determine in what ways, if any, operational ground truth measurement systems can be employed to reduce the overall error in retrieved reflectance factor. Error modeling and propagation methodology is developed for each link in the imaging chain, and representative values are determined for the purpose of exercising the model and observing the system behavior in response to a wide variety of inputs. Three distinct approaches to modelbased atmospheric inversion are compared in a common reflectance error space, where each contributor to the overall error in retrieved reflectance is examined in relation to the others. The modeling framework also allows for performance predictions resulting from the incorporation of operational ground truth measurements. Regimes were identified in which uncertainty in water vapor and aerosols were each found to dominate error contributions to final retrieved reflectance. Cloud cover was also shown to be a significant contributor, while state-of-the-industry hyperspectral sensors were confirmed to not be error drivers. Accordingly, instruments for measuring water vapor, aerosols, and downwelled sky radiance were identified as key to improving reflectance retrieval beyond current performance by current inversion algorithms.
Library of Congress Subject Headings
Remote sensing--Data processing--Mathematical models; Imaging systems--Image quality; Reflectance--Measurement; Algorithms--Evaluation
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Klempner, Scott, "Statistical modeling of radiometric error propagation in support of hyperspectral imaging inversion and optimized ground sensor network design" (2008). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus