Synergies between VSWIR and TIR data for the urban environment: An evaluation of the potential for the Hyperspectral Infrared Imager (HyspIRI) Decadal Survey mission

Abstract

This study provides an introduction to the HyspIRI mission a National Research Council “Decadal Survey” mission that combines a 213 channel visible, near−infrared and shortwave infrared (VSWIR) imaging spectrometer with an 8 channel multispectral thermal infrared (TIR) instrument and evaluates some of its potential in urban science. Potential synergies between VSWIR and TIR data are explored using analogous airborne data acquired over the Santa Barbara metropolitan region in June, 2008. These data were analyzed at both their native spatial resolutions (7.5 m VSWIR and 15 m TIR), and aggregated 60 m spatial resolution similar to HyspIRI. A spectral library of dominant urban materials (e.g., grass, trees, soil, roof types, roads) was developed from field and airborne-measured spectra using Multiple-Endmember Spectral Mixture Analysis (MESMA) and used to map fractions of impervious, soil, green vegetation (GV, e.g., trees, lawn) and non-photosynthetic vegetation (NPV). Land Surface Temperature (LST) and emissivity were also retrieved from the airborne data. Co-located pixels from the VSWIR and TIR airborne data were used to generate reflectance/emissivity spectra for a subset of urban materials. MESMA was used to map GV, NPV, soil and impervious fractions at the different spatial resolutions and compare the fractional estimates across spatial scales. Important surface energy parameters, including albedo, vegetation cover fraction, broadband emissivity and surface temperature were also determined for and evaluated for 14 urban and natural land-cover classes in the region. Fractions were validated using 1 m digital photography. Fractions for GV and NPV were highly correlated with validation fractions at all spatial scales, producing a near 1:1 relationship but with a < 10% overestimate of GV from MESMA. Similar, high correlations were observed for impervious surfaces, although impervious was significantly underestimated in most urban areas and soil overestimated. Comparison of fractions across scales showed high correlation between GV and NPV at 7.5 and 60 m resolution, suggesting that HyspIRI will provide accurate measures of these two measures in urban areas. An inverse relationship between vegetation cover and LST was observed. Albedo proved to be highly variable and poorly correlated with LST. Broadband emissivity was far less variable with high emissivity surfaces (~ 0.95) including vegetation, water and asphalt, and low emissivity surfaces (< 0.95) including selected roof types, beach sands and senesced grasslands. Residential and commercial areas showed a general pattern of increasing LST with increasing impervious fraction with the highest impervious fractions mapped in commercial areas, roads and roofs. Fine scale spatial structure in cover fractions and LST demonstrated important departures from a simple inverse relationship between GV and LST, even at 60 m. The results demonstrate the utility of HyspIRI data for urban studies and provide an insight of what will be possible on a global scale when HyspIRI data become available.