Effects and Mitigation of Clear Sky Sampling on Recorded Trends in Land Surface Temperature

Friday, 18 December 2015: 08:45
3001 (Moscone West)
Thomas R Holmes1,2, Christopher Hain3, Richard de Jeu4, Martha C. Anderson5 and Wade T Crow5, (1)USDA ARS, Beltsville, MD, United States, (2)Science Systems and Applications, Inc., Lanham, MD, United States, (3)Earth System Science Interdisciplinary Center, COLLEGE PARK, MD, United States, (4)Transmissivity B.V., Noordwijk, Netherlands, (5)USDA ARS, Hydrology and Remote Sensing Lab, Beltsville, MD, United States
Land surface temperature (LST) is a key input for physically-based retrieval algorithms of hydrological states and fluxes. Yet, it remains a poorly constrained parameter for global scale studies. The main two observational methods to remotely measure T are based on thermal infrared (TIR) observations and passive microwave observations (MW). TIR is the most commonly used approach and the method of choice to provide standard LST products for various satellite missions. MW-based LST retrievals on the other hand are not as widely adopted for land applications; currently their principle use is in soil moisture retrieval algorithms.

MW and TIR technologies present two highly complementary and independent means of measuring LST. MW observations have a high tolerance to clouds but a low spatial resolution, and TIR has a high spatial resolution with temporal sampling restricted to clear skies. This paper builds on recent progress in characterizing the main structural differences between TIR LST and MW Ka-band observations, the MW frequency that is most suitable for LST sensing. By accounting for differences in diurnal timing (phase lag with solar noon), amplitude, and emissivity we construct a MW-based LST dataset that matches the diurnal characteristics of the TIR-based LSA SAF LST record. This new global dataset of MW-based LST currently spans the period of 2003-2013. In this paper we will present results of a validation of MW LST with in situ data with special emphasis on the effect of cloudiness on the performance.

The ability to remotely sense the temperature of cloud covered land is what sets this MW-LST datasets apart from existing (much higher resolution) TIR-based products. As an example of this we will therefore explore how MW LST can mitigate the effect of clear-sky sampling in the context of trend and anomaly detection. We do this by contrasting monthly means of TIR-LST with its clear-sky and all-sky equivalent from an MW-LST and an NWP model.