The sensitivity of land emissivity estimates from AMSR-E at C and X bands to surface properties

Abstract

Abstract. Microwave observations at low frequencies exhibit more sensitivity to surface and subsurface properties with little interference from the atmosphere. The objective of this study is to develop a global land emissivity product using passive microwave observations from the Advanced Microwave Scanning Radiometer -Earth Observing System (AMSR-E) and to investigate its sensitivity to land surface properties. The developed product complements existing land emissivity products from SSM/I and AMSU by adding land emissivity estimates at two lower frequencies, 6.9 and 10.65 GHz (C-and X-band, respectively). Observations at these low frequencies penetrate deeper into the soil layer. Ancillary data used in the analysis, such as surface skin temperature and cloud mask, are obtained from International Satellite Cloud Climatology Project (ISCCP). Atmospheric properties are obtained from the TIROS Operational Vertical Sounder (TOVS) observations to determine the small upwelling and downwelling atmospheric emissions as well as the atmospheric transmission. A sensitivity test confirms the small effect of the atmosphere but shows that skin temperature accuracy can significantly affect emissivity estimates. Retrieved emissivities at C-and X-bands and their polarization differences exhibit similar patterns of variation with changes in land cover type, soil moisture, and vegetation density as seen at SSM/I-like frequencies (Ka and Ku bands). The emissivity maps from AMSR-E at these higher frequencies agree reasonably well with the existing SSM/I-based product. The inherent discrepancy introduced by the difference between SSM/I and AMSR-E frequencies, incidence angles, and calibration has been assessed. SigCorrespondence to: H. Norouzi (hnorouzi@citytech.cuny.edu) nificantly greater standard deviation of estimated emissivities compared to SSM/I land emissivity product was found over desert regions. Large differences between emissivity estimates from ascending and descending overpasses were found at lower frequencies due to the inconsistency between thermal IR skin temperatures and passive microwave brightness temperatures which can originate from below the surface. The mismatch between day and night AMSR-E emissivities is greater than ascending and descending differences of SSM/I emissivity. This is because of unique orbit time of AMSR-E (01:30 a.m./p.m. LT) while other microwave sensors have orbit time of 06:00 to 09:00 (a.m./p.m.). This highlights the importance of considering the penetration depth of the microwave signal and diurnal variability of the temperature in emissivity retrieval. The effect of these factors is greater for AMSR-E observations than SSM/I observations, as AMSR-E observations exhibit a greater difference between day and night measures. This issue must be addressed in future studies to improve the accuracy of the emissivity estimates especially at AMSR-E lower frequencies.

Assessment of the consistency among global microwave land surface emissivity products

“…High emissivity-based MPDI values greater than 0.06 are found for low soil moisture and low NDVI values in all products. This is in line with previous studies that suggest that the contrast between horizontal and vertical microwave signals is higher in desert regions with almost no vegetation (e.g., Norouzi et al, 2011;Prigent et al, 2006). Soil moisture has been found to decrease the emissivity across the frequencies (Basist et al, 1998).…”

“…Previous studies have shown that differences in channel frequencies and incidence angles between AMSR-E and SSMI/I channels may lead to around 0.01 error in the emissivity retrieval (Norouzi et al, 2011). TMI and SSM/I have similar emissivity values especially at 10, 19, and 37 GHz.…”

“…The comparison of emissivity and NDVI as indicators of vegetation density and surface condition revealed that the differences between horizontal and vertical emissivity values (e H − e V ) have positive correlation with vegetation and soil moisture values (Norouzi et al, 2011;Prigent et al, 2006) using AMSR-E and SSM/I values. Monthly temporal correlation between e H − e V and NDVI values were calculated for 5 years globally for each pixel.…”

“…The products show better agreement at lower frequencies which involve less scattering and deeper penetration into the canopy unlike highfrequency brightness temperature which penetrates less and reflects the top of the canopy microwave radiating temperature. The use of the canopy skin temperature, in the rain forest region, to approximate the canopy effective temperature for low and high frequencies in the emissivity retrieval can lead to discrepancies among the products that are frequency dependent (Norouzi et al, 2011;Prigent et al, 1999). The effective optical depth of the canopy depends on the vegetation water content, intercepted water in rain forests, and the vegetation structure and type.…”

“…Global land emissivity retrieval first was developed by Prigent et al (1998) when brightness temperatures from the Special Sensor Microwave Imager (SSM/I) were used. Other available products later were proposed from other sensors, such as the Advanced Microwave Scanning Radiometer -Earth Observing System (AMSR-E) (Norouzi et al, 2011;Moncet et al, 2011), the Advanced Microwave Sounding Unit (AMSU) (Karbou et al, 2005), and the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) (Furuzawa et al, 2012). To retrieve land emissivity values, those studies did not necessarily use the same ancillary data, radiative transfer model, and assumptions to account for the atmospheric contribution.…”