Subsurface emission effects in AMSR-E measurements: Implications for land surface microwave emissivity retrieval

Abstract

[1] An analysis of land surface microwave emission time series shows that the characteristic diurnal signatures associated with subsurface emission in sandy deserts carry over to arid and semiarid regions worldwide. Prior work found that diurnal variation of Special Sensor Microwave/Imager (SSM/I) brightness temperatures in deserts was small relative to International Satellite Cloud Climatology Project land surface temperature (LST) variation and that the difference varied with surface type and was largest in sand sea regions. Here we find more widespread subsurface emission effects in Advanced Microwave Scanning Radiometer-EOS (AMSR-E) measurements. The AMSR-E orbit has equator crossing times near 01:30 and 13:30 local time, resulting in sampling when near-surface temperature gradients are likely to be large and amplifying the influence of emission depth on effective emitting temperature relative to other factors. AMSR-E measurements are also temporally coincident with Moderate Resolution Imaging Spectroradiometer (MODIS) LST measurements, eliminating time lag as a source of LST uncertainty and reducing LST errors due to undetected clouds. This paper presents monthly global emissivity and emission depth index retrievals for 2003 at 11, 19, 37, and 89 GHz from AMSR-E, MODIS, and SSM/I time series data. Retrieval model fit error, stability, self-consistency, and land surface modeling results provide evidence for the validity of the subsurface emission hypothesis and the retrieval approach. An analysis of emission depth index, emissivity, precipitation, and vegetation index seasonal trends in northern and southern Africa suggests that changes in the emission depth index may be tied to changes in land surface moisture and vegetation conditions.

Land surface microwave emissivities derived from AMSR-E and MODIS measurements with advanced quality control

“…Although the assumptions inherent in these formulations are major simplifications of the actual properties of terrestrial surfaces, this construct has been shown to be a useful first‐order quantification of penetration effects, wherein α ν represents spatial and temporal variations in penetration. The assumptions are tested in the quality control process using the model fit error [ Galantowicz et al , 2011].…”

“…Wherever the criteria for creating class‐based emissivities (section 3.4) are met, they override the default. The PTS products are used for all grid points that pass a series of tests for reliability of the PTS retrieval and significance of microwave penetration, considering data sample sizes and measures of the quality of the fit between the data and the penetration model [ Galantowicz et al , 2011]. By these criteria, the PTS products are selected for a substantial portion of the global land, responding to arid conditions well beyond deserts.…”

Quantifying Uncertainties in Land-Surface Microwave Emissivity Retrievals

“…In addition to factors highlighted in this study, many other ones, such as instrument errors (e.g., [18]), calibration errors, differences in locations and resolutions of the satellite footprints, spatial and temporal sampling errors, and representativeness of surface properties over highly heterogeneous and dynamic surfaces (e.g., desert) [13], [19], [20], all contribute to the complexity of the retrieval process and the associated errors. It is therefore necessary to recognize the limitations in these instantaneous retrievals, and to develop strategies to alleviate their impact while exacting as much useful information as possible.…”

“…We speculate this might be related to the strong diurnal cycle in the variation of the surface temperature and the microwave penetration depth, which makes it tricky to represent the effective emission characteristics in the retrieval process [13], [17], [19], [20].…”

A framework for the retrieval of all-weather land surface temperature at a high spatial resolution from polar-orbiting thermal infrared and passive microwave data

“…Second, temperature retrieval from PMW measurements yields subsurface temperature, which is different from LST retrieval from TIR measurements (skin temperature). Nevertheless, subsurface temperatures can be converted to skin temperatures using the thermal diffusion equation (Gao et al, 2010;Moncet et al, 2011;Galantowicz et al, 2011). Third, the accuracies (approximately 5-6 K) of subsurface temperature retrieval from PMW measurements are worse than those (approximately 1-2 K) of LST retrieval from TIR measurements.…”