Abstract
Meeting the accuracy demand of products derived from spaceborne laser altimetry requires post-launch calibration/validation procedures. Over general surfaces, such as the ones land or ice offer, calibration methods that suite undulating topography should be applied. Whereas the estimation of calibration parameters usually refers to the derivation of an error model, the problem here extends beyond that. Not knowing in advance the offsets between the recorded and actual footprint locations implies that the offsets and the calibration parameters should be estimated jointly. In addition, concerns arise regarding the reliability and quality of the estimates, namely avoiding convergence to a false solution and avoiding estimates that are highly correlated and with a low level of accuracy. The correlation and the accuracy can indeed be estimated in retrospect, but the preference is to learn in advance what factors affect them and under what circumstances they can be optimized. This paper analyzes the calibration problem along the lines of correctness and reliable estimation and shows that the error recovery model provides ample information that can lead to a solution that is both efficient and an integral part of a calibration model. Analysis of the model shows that over natural topography relatively short sites are sufficient for achieving a reliable calibration. The paper then shows that by applying optimization criteria on the estimation, rules for predicting the quality of the solution can be derived as well as general measures for evaluating sites considered for calibration.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have