South Dakota State University
Radiometric Calibration and Uncertainty Analysis of Landsat MSS Sensors Using a Radiance Based Approach on Pseudo-Invariant Calibration Sites
Abstract
dc:description.abstractThe primary objective of this project was to develop a consistent radiometric calibration for the five Landsat Multispectral Scanner (MSS) instruments in terms of producing comparably calibrated Top of Atmosphere (TOA) radiance values. This calibration process will allow the entire MSS data set, as contained within the USGS archive, to be accessed based upon a consistent radiometric scale. Initially, a consistent cross-calibration of all MSS sensors was achieved using co-incident/near co-incident scene pairs as acquired during the spatial and temporal overlap of these satellites sensors. Next, several sets of two nearly co-incident scenes were matched based on use of Pseudo-Invariant Calibration Sites (PICS). Regions of interest were sampled to estimate the relationship between the gains of the two sensors. The cross-calibration to Landsat-5 MSS for each sensor was calculated by combining the results of individual cross-calibrations between sensor pairs. For each sensor, changes in both bias and gain were measured and statistically evaluated. An example of the significance of the results is seen in MSS bands 1 and 2 where prior to calibration sensors exhibit errors up to 16% and 17% respectively. Additionally, temporal shifts (often considered as 'sensor degradation') can modify calibration up to 1% for band 1 and 2% for band 2 over a degradation') can modify calibration up to 1% for band 1 and 2% for band 2 over a sensor's lifetime. Once these calibration parameters are applied, test scenes as sampled from the entire MSS data archive are radiometrically consistent to within 1% for band 1, 2% for band 2, 3% for band 3 and 5% for band 4. Once the cross-calibration connection has been established for a pair of Landsat MSS sensors, the possible uncertainty and ranking of the uncertainty types associated with the cross-calibration process was studied. An error analysis shows that the dominant uncertainty for the cross-calibration was associated with atmospheric uncertainty in nearly coincident scene pairs. Calibration uncertainties range from 4% to 11 % due to the atmosphere; bidirectional reflectance distribution function (BRDF) and filter spectral effects together can add an additional 2% uncertainty.
Degree
thesis:*- Name thesis:degree_name
- Master of Science (MS)
- Level thesis:degree_level
- Thesis - University Access Only
- Discipline thesis:degree_discipline
- Electrical Engineering and Computer Science
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Karki, Sadhana
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://openprairie.sdstate.edu/etd2/1804