1. LIS and CRTM Coupling Jing ‘Lily’ Zeng 1,4, Ken Harrison 1,2, Sujay Kumar 1,3, Christa Peters- Lidard 1, John Eylander 5, Weizhong Zheng 6, Fuzhong Weng 6, Mike Ek 6 JCSDA 7th Workshop on Satellite Data Assimilation University of Maryland at Baltimore County (UMBC) May 12-13, 2009 1 – Hydrological Sciences Branch, NASA/GSFC 2 – Earth System Science Interdisciplinary Center (ESSIC), Univ. Maryland, College Park 3 – Science Applications International Corporation (SAIC) 4 – Earth Resources Technology, Inc. (ERT) 5 – Air Force Weather Agency 2WG/WE 6 – Environmental Modeling Center, NOAA

2. Outline LIS-CRTM Coupling Fu-Liou Flux Model Testing with Cloud Optical Property (COP) data AFWA has need for radiative flux determination, which is not a current or planned capability of NESDIS CRTM Pg. 2

3. LIS-CRTM Coupling Develop a fully coupled prototype in LIS that enables the CRTM forward model integration, drawing on the earlier work of NCEP/NESDIS developing the CRTM ‘Stand-Alone’ Driver that was provided to the LIS team. The coupling will enable future radiance data assimilation. Pg. 3

4. NCEP/NESDIS ‘CRTM Stand-Alone Driver’ Input: Atmospheric and land surface states from GDAS files Output: Tb and surface emissivity for satellite “window” channels Comparison to TOVS BUFR radiance dataset Limitations: Clear sky  Objective of coupling is to drive CRTM with LSM surface data as provided through LIS Pg. 4

5. New LIS Support for Radiative Transfer Model (RTM) Pg. 5

6. RTM-related Extensible Interfaces Pg. 6 CRTM_init LIS-supported land types (IGBP, UMD, USGS, GDAS) matched to CRTM land types CRTM_forcing Atmospheric forcing state assigned to CRTM atmospheric data structure To supply Atmospheric Forcing State to an RTM, ‘3D’ Meteorological Inputs now are supported in LIS Formerly ‘2D’ (only surface layer taken) 3D Meteorological Inputs implemented as a Supplemental Forcing Current testing with 64 Atmospheric Layer-GDAS (as in CRTM Stand- Alone Driver): Temperature, Pressure 2 absorbers (water vapor, ozone) Clear-sky (if clouds, need cloud optical properties,) Noah_sfc2crtm maps to CRTM properties: Land_Temperature Soil_Moisture_Content Canopy_Water_Content Vegetation_Fraction Soil_Temperature Snow (Temperature, Depth, Density ) CRTM Stand-Alone Driver logic largely adopted, e.g., CRTM soil moisture content = Noah 1 st layer soil moisture If (snow depth exceeds threshold) then Snow_Coverage = 100%; else Land_Coverage = 100%

7. Status LIS architecture changed to support addition of an ‘RTM’ CRTM added as the first RTM, with mappings/logic largely based on the NCEP/NESDIS CRTM Stand-Alone Driver At a user-specified reporting frequency, for CRTM- supported sensors, brightness temperatures and emissivity in different channels can be simulated Pg. 7

8. Planned Improvements over the Next Year Review current use of CRTM defaults Add support to compare RTM-simulated and satellite- observed radiance (by channel) Validate microwave simulation of emissivity and brightness temperature (Tb) and investigate sensitivity of channels to land surface conditions Add Radiance Data Assimilation run mode to exploit LIS Data Assimilation capabilities Pg. 8

9. Fu-Liou Flux Model Testing with COP Data Fu-Liou flux model testing (CRTM at present time does not have radiative flux capability) Pg. 9

10. Science Requirements Fu-Liou Radiative Flux with COP Capability to make multiple wavelength (narrow and broad bands) calculations for SW/LW flux. Capability to calculate direct and diffuse flux and provide dynamic radiative properties. Pg. 10

11. Discussion of Flux calculation algorithm – AGRMET vs. Fu-Liou AGRMET SWLW SW flux is calculated according to Shapiro (1987). It uses analyzed cloud amounts and types for four layers within the cloud but not in the atmosphere from AFWA’s CDFSII model to calculate the surface emissivity and reflectivity No aerosol absorption. LW for clear sky is calculated according to Idso (1981) Cloud transmission is parameterized according to cloud type (model input is cloud type). Linear combination of clear and cloudy with cloud fraction. Fu-Liou SWLW Consider two radiative streams in the upper and lower hemispheres. Expand the phase function in four terms. Multiple layers, 35 layers, each layer is homogeneous, there is no diffuse radiation from the top and bottom of this sublayer; adjustable number of cloud layers. Incorporate a  – Eddington method to account for the forward peak in the context of the four-stream approximation. Include aerosol absorption. LW for clear sky is physically based. Cloudy sky is physical based.(same input as the clear sky) consider two radiative streams in the upper and lower hemispheres. Expand the phase function in four terms. Partial cloudy sky: linear combination. Pg. 11

12. Advantages of the  –Four Stream RTM Four stream: two upward and two downward Gaussian integration angles. K-distribution: faster integration vs. line by line. g, ω, and β are determined by r eff, which is provided in AER data. Comparison without adjusting any parameters (no calibration). Solar irradiance spectrum above atmosphere and at surface. Pg. 12

13. Data and Method Field observations: SGP(CF) Period:2007/04-2007/08 Data: Flux observations: SW and LW Flux calculation input: cloud effective radius, cloud liquid water content, aerosol optical thickness, atmospheric profiles Model simulations Fu-Liou simulations with SGP(CF) data AGRMET archived 3hr output Observed diffuse/direct to determine sky condition Totally cloudy: diffuse/direct >=0.85 (>90%) Totally clear: diffuse/direct 90%) Pg. 13

14. SW Flux Calculation Clear Sky with Fu-Liou Code intercept : -9.4± 2.8 slope : 0.95 ± 0.11 correlation: 0.99 5 days obs.Day 20080608 Diurnal variation of Fu-Liou calculated SW flux (using the SGP cloud, aerosol obs and vertical profile) and SGP observed SW flux. Pg. 14

15. The Fu-Liou calculation is using SGP cloud, aerosol and vertical profile observations. SW Flux Calculation Cloudy and Partial Cloudy Sky with Fu-Liou Code 3 obs days intercept : 15.3±7.6 slope: 1.03±0.28 correlation: 0.93 intercept :11.42±7.8 slope:1.04±0.17 correlation: 0.94 72 obs days Pg. 15

16. Input Aerosol Optical Thickness from SGP Obs SGP observed aerosol optical (AOT) thickness is biased in cloudy skies. AOT in cloudy sky AOT in partial cloudy sky AOT in clear sky Pg. 16

17. Comparison of AGRMET, Fu-Liou Calculated SW with SGP Observed SW Flux b: 30.91±29.1 a : 0.90±0.36 Correlation: 0.98 b: -3.97±105.1 a: 0.43±0.77 Correlation:0.53 5 days3 days 72 days b: -6.71± -11.04 a:0.86±0.25 Correlation:0.90 b: 17.19± 24.01 a: 0.88±0.30 Correlation: 0.99 b: 8.26±23.83 a: 0.35±0.61 correlation: 0.66 5 days 3 days 72 days b:20.83±12.62 a: 0.84±0.24 Correlation: 0.93 Fu-Liou flux is calculated with GOES derived Cloud Optical Properties (COP) cloud observations Pg. 17

18. Conclusions The four-stream RTM code can give more accurate calculation of SW flux in cloudy and partial cloudy conditions with accurate input of cloud and aerosol observations. Pg. 18

19. Future Work Solve the problem of missing input cloud optical depth. Understanding AFWA model formulation and uncertainty in the input data (understand causes of discrepancy). How to increase the efficiency of the four-stream Fu-Liou code to make it fit AFWA application requirement. Do validation in snow cover region (ARM/North Slope Alaska); do validation in winter season. Continue using the COP data in the flux calculation. Validation of the COP flux calculation vs. ARM. Coupling of Fu-Liou code with LIS-AGRMET. Pg. 19