GC23K-1246
Finding Blackbody Temperature and Emissivity on a Sub-Pixel Scale

Tuesday, 15 December 2015
Poster Hall (Moscone South)
David Jonathan Bernstein1, Jesse Bausell2, Shane Grigsby3 and Raphael Martin Kudela2, (1)Purdue University, West Lafayette, IN, United States, (2)University of California Santa Cruz, Ocean Sciences, Santa Cruz, CA, United States, (3)University of Colorado at Boulder, Boulder, CO, United States
Abstract:
Surface temperature and emissivity provide important insight into the ecosystem being remotely sensed. Dozier (1981) proposed a an algorithm to solve for percent coverage and temperatures of two different surface types (e.g. sea surface, cloud cover, etc.) within a given pixel, with a constant value for emissivity assumed. Here we build on Dozier (1981) by proposing an algorithm that solves for both temperature and emissivity of a water body within a satellite pixel by assuming known percent coverage of surface types within the pixel. Our algorithm generates thermal infrared (TIR) and emissivity end-member spectra for the two surface types. Our algorithm then superposes these end-member spectra on emissivity and TIR spectra emitted from four pixels with varying percent coverage of different surface types. The algorithm was tested preliminarily (48 iterations) using simulated pixels containing more than one surface type, with temperature and emissivity percent errors of ranging from 0 to 1.071% and 2.516 to 15.311% respectively[1]. We then tested the algorithm using a MASTER image from MASTER collected as part of the NASA Student Airborne Research Program (NASA SARP). Here the temperature of water was calculated to be within 0.22 K of in situ data. The algorithm calculated emissivity of water with an accuracy of 0.13 to 1.53% error for Salton Sea pixels collected with MASTER, also collected as part of NASA SARP. This method could improve retrievals for the HyspIRI sensor.



[1] Percent error for emissivity was generated by averaging percent error across all selected bands widths.