Task Title: Land Surface Temperature Diurnal Analysis (Diurnal) to Validate the Performance of GOES-R Advance Baseline Imager PI: Konstantin Vinnikov, Senior research Scientist Institute: University of Maryland/ Department of Atmospheric and Oceanic Science/ CICS Period of Performance: 1 July 2012 – 30 June 2013

1. Background: The proposed research is a continuation of work on the current Task that is going to be fulfilled to the end of June 2012. The goals of this current Task are: to statistically evaluate angular anisotropy of LST (Land Surface Temperature) using available simultaneous land based and satellite, GOES-8 and GOES-10, year 2001 observations at locations of SURFRAD stations; to develop an algorithm for correcting GOES-R retrieved clear sky LST for angular anisotropy of LST field; to take into account angular anisotropy of LST in validation of GOES-R observed LST. There have been few related modeling and case study type investigations on this subject, which include empirical evaluation of “Anisotropy in the land surface temperature as observed with two geostationary satellites” by Peter Romanov and Dan Tarpley. They concluded that “at least 2-3 K temperature change may be caused by changing the relative solar-satellite azimuth from backscatter to approximately 90º. The difference between temperatures observed in the backscatter and forward scatter may be higher. The major reason for the temperature differences is the effect of shadows”. A climatological approach is applied here. The main requirement of the angular correction algorithm is to convert satellite observed directional LST, T(z,zs,az), that depends on satellite viewing angle z, sun zenith angle zs, and azimuth az into the scalar, unbiased, energy effective value of LST, θ, that can be used in land surface energy balance computation.

2. Proposed Work: We postulate that there are two main kernels in the angular anisotropy of LST. The first kernel, related to angular anisotropy of land surface absorbed solar radiation, manifests itself in the daytime LST data, only. Effect of shadows is the main component of this kernel and it depends on satellite viewing and solar, zenith and relative azimuth, angles. The other kernel manifests itself in the nighttime dependence of LST on viewing zenith angle and is caused by angular dependences of broadband infrared land surface emissivity and atmospheric infrared radiation. It is analogous to land surface limb darkening phenomenon. This effect can be studied using nighttime LST observations when it can be distinguished from the daytime shadowing effect. The infrared kernel works also in the daytime, when its effect is superimposed on the solar kernel. It is expected that the current period of this project will be completed with developing the first prototype of the algorithm for angular correction of GOES-R satellite retrieved LST. Angular correction should be also included in the GOES-R LST validation algorithm. The approach, kernels, parameterizations, structure of the algorithm, and strategy of its application are going to be enhanced and tested using independent data during the proposed period of work. 3. Goals: The goals of the proposed work are in comprehensive testing and improving of the developed first version of LST angular correction algorithm using independent satellite observed data.

4. Data Description: GOES Surface and Insolation Products (GSIP) contains GOES-E and GOES-W LST retrievals with temporal resolution of one hour and spatial resolution of 1/8 degree (~14km). Data from GOES-E has a 15-minute shift compared to GOES-W data. No angular correction is included into the LST retrieval algorithm. One full year of observation (May 2010-April 2011) for Northern Hemisphere Extended Scan Sectors of satellites GOES-13 (EAST) and GOES-11 (WEST) will be obtained from the CLASS/NOAA web site http://www.class.noaa.gov/ . Systematic, but seasonally/diurnally distributed LST error resulted from the 15-minute shift in observation time of two satellites. This error has the same order of magnitude as effect of angular anisotropy. To be comparable, LST observed by GOES-13, will be interpolated into the observation times of GOES-11. This should be done at each spatial grid at each time. Mean diurnal and seasonal variations of LST will be evaluated and taken into account in the interpolation algorithm.

5. Method

The proposed statistical model to approximate angular dependence of satellite observed LST can be expressed by the next simple equation:

T(z,zs,az)/T(0) =1+A·φ(z)+D·ψ(z,zs≤90º,az), (1)

where: T(0) is LST in the nadir direction at z=0; φ(z) is “infrared kernel” (related to angular anisotropy of infrared radiation near land surface); ·ψ(z,zs≤90º,az) is “solar kernel” (related to angular anisotropy in solar radiation near land surface), ψ(z,zs≥90º,az)≡0; A and D are the coefficients that should be estimated from observations. These coefficients depend on land cover structure. The equations applied to estimate A and D from instantaneous LST observations of two satellites, GOES-EAST (W) and GOES-WEST (W), at the same location are:

TE-TW ≈BW+A*(TW+BW)·φ(ZE)-TE·φ(ZW). (2)

TE·[1+A·φ(ZW)]-(TW+BW)·[1+A·φ(ZE)]≈D·[(TW+BW)·ψ(ZE,ZS,AZE)-TE· ψ(ZW,ZS,AZW)].(3)

BW is constant bias of GOES-WEST retrieved LST compared to GOES-EAST retrieved temperature which is assumed to be unbiased. First step: Ordinary least squared with a few iterations (to resolve weak nonlinearity) are applied to the nighttime observations to estimate BW and A from (2). Second step: ordinary least squares technique and daytime observation are used to estimate D in the equation (3). Different analytical expressions of kernel functions are going to be tested. θ - An unbiased, angular corrected, energy effective, value of LST can be estimated from the following equation: (4) The double integral in the right side of expression (4) is constant for the most of reasonable analytical approximations of φ and ψ.

6. Subtasks and Expected Outcome  Collect and quality control full year of overlapping GOES-11 and GOES-13 GSIP archived LST data for Northern Hemisphere Extended Scan Sectors. Adjust GOES-13 LST data to GOES-11 times of observation. Expected result: One year GOES-13 LST data set adjusted to GOES-11 times of observation.  Validate and improve parameterization of IR limb darkening kernel in the empirical model of LST angular anisotropy using nighttime overlapping observations of GOES-11 and GOES-13. Expected result: Improved parameterization of the infrared kernel.  Validate and improve parameterization of solar radiation related kernel in the empirical model of LST angular anisotropy using daytime overlapping observations of GOES-11 and GOES-13. Expected result: Improved parameterization of the solar kernel.  Statistical evaluation of accuracy of proposed LST angular anisotropy correction algorithm. Expected result: Conclusion on accuracy of the proposed algorithm.  Accomplish experimental blending historical GSIP GOES-11 and GOES-13 LST data using proposed algorithm for LST angular anisotropy correction. Expected result: Conclusion on usefulness and recommendation on blending LST data retrieved from GOES-E and GOES-W satellites.