Improved surface temperature estimates with MASTER/AVIRIS sensor fusion

Abstract

a b s t r a c tLand surface temperature (LST) is an important parameter in many ecological studies. The current Root Mean Square Error (RMSE) in standard MODIS and ASTER LST products is greater than 1 K, and for ASTER can be as large as 4 K for graybody pixels such as vegetation. Errors of 3 to 8 K have been observed for ASTER in humid conditions, making knowledge of atmospheric water vapor content critical in retrieving accurate LST. For this reason improved accuracy in LST measurements through the synthesis of visible-to-shortwave-infrared (VSWIR) derived water vapor maps and Thermal-Infrared (TIR) data is one goal of the Hyperspectral Infrared Imager, or HyspIRI, mission. The 2011 ER-2 Delano/Lost Hills flights acquired data with both the MODIS/ASTER Simulator (MASTER) and Airborne Visible InfraRed Imaging Spectrometer (AVIRIS) instruments flown concurrently. This study compares LST retrieval accuracies from the standard JPL MASTER temperature products produced using the temperature-emissivity separation (TES) algorithm, and the water vapor scaling (WVS) atmospheric correction method proposed for HyspIRI. The two retrieval methods are run both with and without high spatial resolution AVIRIS-derived water vapor maps to assess the improvement from VSWIR synthesis. We find improvement using VSWIR derived water vapor maps, with the WVS method being most accurate overall. For closed canopy agricultural vegetation we observed temperature retrieval RMSEs of 0.49 K and 0.70 K using the WVS method on MASTER data with and without AVIRIS derived water vapor, respectively.

Monitoring the Impacts of Severe Drought on Southern California Chaparral Species using Hyperspectral and Thermal Infrared Imagery

“…Temperature-emissivity separation required for retrieving LST used a water vapor scaling method described by [37]. Root mean square error of LST retrieval has been estimated at 1.2 K for ASTER data by [38] and at 0.7 K for MASTER data [39]. LST data were registered to the same 18 m orthoimage basemap and were resampled to 18 m spatial resolution using nearest neighbor resampling.…”

Reassessment of the temperature-emissivity separation from multispectral thermal infrared data: Introducing the impact of vegetation canopy by simulating the cavity effect with the SAIL-Thermique model

“…Many studies investigated the retrieval performances of TES, by focusing either on calibration (Grigsby et al, 2015;Hulley, 2011;Hulley and Hook, 2011;Jimenez-Munoz et al, 2014;Payan and Royer, 2004;Sobrino and Jiménez-Muñoz, 2014) or on retrieval accuracies (French et al, 2008;Gillespie et al, 2011;Göttsche and Hulley, 2012;Hulley et al, 2012b;Jacob et al, 2004;Jiménez-Muñoz et al, 2006;Jimenez-Munoz et al, 2014;Mira et al, 2009Mira et al, , 2011Sabol et al, 2009;Sobrino et al, 2007). These studies addressed several issues among which (1) the calibration of the ε-min -MMD relationship for different sensor spectral configurations, (2) the robustness of the ε-min -MMD relationship and the TES retrieval performances with regards to emissivity variations depicted by land surfaces, and (3) the impact of experimental errors (e.g., instrumental and atmospheric perturbations) on the accuracy of TES emissivity/temperature retrievals.…”

“…Gillespie et al (1998) first proposed a calibration for earth observation system (EOS)/advanced spaceborne thermal emission and reflection radiometer (ASTER) using five channels. Several studies have subsequently proposed calibrations for the spectral configurations of various sensors such as EOS/moderate resolution imaging spectroradiometer (MODIS) with three channels Hulley and Hook, 2011;Jimenez-Munoz et al, 2014), EOS/ASTER with three channels close to the MODIS ones (Hulley and Hook, 2011), hyperspectral infrared imager (HyspIRI) with six channels (Hulley, 2011), MODIS/ASTER simulator (MASTER) with five channels close to HyspIRI ones (Grigsby et al, 2015), or Meteosat second generation (MSG)/spinning enhanced visible and infrared image (SEVIRI) with three channels (Jimenez-Munoz et al, 2014). Overall, calibrated coefficients vary greatly from one sensor to another, with relative changes in A, B and C values up to 15-20% relative, and with subsequent changes in ε-min value up to 0.015 (respectively 0.025) for a MMD value of 0.1 (respectively 0.3).…”

Utilizing HyspIRI Prototype Data for Geological Exploration Applications: A Southern California Case Study

“…All analyses used data from the 2013 suite of flights as these were corrected by the AVIRIS and MASTER teams and were used to create the HyspIRI simulated products including at-surface reflectance at several spatial scales, land surface temperature and emissivity [35][36][37]. Apparent surface reflectance is generated using the ATREM atmospheric correction [35].…”

“…Noise approximating a HyspIRI VSWIR noise function was added to radiance data [36]. Temperature-emissivity separation takes advantage of concurrent water vapor estimates obtained from the AVIRIS sensor to improve land surface temperature [37].…”