Summary Results from:
Intercomparison of Versions 4, 4.1 and 5 of the MODIS Land Surface Temperature and Emissivity Products and Validation with Laboratory Measurements of Sand Samples from the Namib Desert, Namibia
As they relate to the validation of MOD21
Authors: Hulley, G.C., Hook, S.J.
Source: Remote Sensing of Environment 113 (2009) 1313-1318
Link to: Access Publication
Abstract:
Eight new refinements were implemented in the MODIS Land Surface Temperature and Emissivity (LST&E) product suite when transitioning from version 4 (V4) to version 5 (V5). The refinements were designed to improve the spatial coverage, stability, and accuracy of the product suite. Version 4.1 (V4.1) is an interim collection which uses V5 input products (MOD02, MOD03, MOD07, MOD10, and MOD35), but the LST&E retrieval algorithm is unchanged from V4 in which the split-window and day/night temperature retrieval algorithms are only partially incorporated, and not fully incorporated as in V5. A test dataset for the V4.1 product was produced by MODAPS for a 3-month period from July through September 2004, and after an initial evaluation period, it was decided to generate the V4.1 product from mission period 2007001 onwards as a continuation of previous years of V4 data. This paper compares MODIS retrieved surface emissivities between V4, V4.1 and V5 using the level-3 MODIS daily LST&E product, MOD11B1.Comparisons of MOD11B1 retrieved surface emissivity during the Jul-Sep 2004 test period with lab measurements of sand samples collected at the Namib desert, Namibia result in a combined mean absolute emissivity difference for bands 29 (8.55μm), 31 (11μm) and 32 (12μm) of 1.06%, 0.65% and 1.93% for V4, V4.1 and V5 respectively. Maximum band 29 emissivity differences with the lab results were 4.10%, 2.96% and 8.64% for V4, V4.1 and V5 respectively. These results indicate that over arid and semi-arid areas, users should consider using MODIS V4 or V4.1 data instead of V5. Furthermore, users should be careful not to develop time series from a mixture of product versions that could introduce artifacts at version boundaries.