An Exploration, for the Upper Indus Basin, of Elevation Dependency in the Relationships Between Locally Observed Near Surface Air Temperature (SAT) and Remotely-Sensed Land Surface Temperature (LST)

Tuesday, 16 December 2014: 9:30 AM
Nathan Daniel Forsythe1, Hayley J Fowler2, Stephen Blenkinsop2, Chris G Kilsby2, David R Archer3, Andrew J Hardy4 and Tomas du Chemin Holderness5, (1)Newcastle University, Newcastle Upon Tyne, NE1, United Kingdom, (2)Newcastle University, Newcastle Upon Tyne, United Kingdom, (3)JBA Consulting Engineers and Scientists, Skipton, United Kingdom, (4)Aberystwyth University, Institute of Geography and Earth Sciences, Aberystwyth, United Kingdom, (5)University of Wollongong, SMART Infrastructure Facility, Wollongong, Australia
The distribution of ground-based observations of near-surface air temperature (SAT) is extremely skewed toward low elevation areas. Land surface temperature (LST) remote sensing data products -- from thermal and infrared wavelength satellite imagery -- provide spatial coverage independent of elevation, although they only provide values for “clear sky” conditions, the prevalence of which may be influenced by elevation-dependent factors. It is thus imperative for researchers studying EDW to characterise the relationship between observations of “all-sky” SAT and “clear-sky” thermal/infrared (TIR) LST in order to overcome the extreme sparseness of SAT observations at high elevations.

Drawing on local SAT observation data from both manned meteorological stations and AWS units covering an elevation range from 1500 to 4700m asl in the Upper Indus Basin, coupled with cloud climatologies from MODIS and global reanalyses, this study develops “clear-sky” and “all-sky” comparative, site-based climatologies of:

[a] ground-observed SAT

[b] reanalysis SAT and LST (skin surface temperature)

Relationships between these climatologies and corresponding clear-sky/TIR satellite-retrieved LST are quantitatively assessed in the context of elevation-dependency and cloud cover prevalence. The implications of these relationships are discussed in the context of efforts to develop a multi-decadal TIR LST data product. While multi-decadal and even centennial trends are calculated from station-based observations of SAT, the relatively short record lengths of satellite-borne instruments used to produce currently available TIR LST data products better lend themselves to characterisation of interannual variability than trend calculation. Thus progress is detailed on EDW-driven efforts to validate such an LST product for the Himalayan region using historical imagery from the second and third generation of the Advanced Very High Resolution Radiometer (AVHRR/2, AVHRR/3) instrument flown on NOAA satellite platforms since the mid-1980s through present day. Progress and remaining challenges are quantified in terms of skill and bias of AVHRR LST with respect to MODIS LST as well the intrinsically coupled AVHRR cloud product with respect to its MODIS analogue.