CRTM Non Local Thermodynamic Equilibrium (NLTE) ModuleDevelopment and Implementation for CRTM Version 2.1 Yong Han, Yong Chen and Paul van Delst JCSDA 9th workshop, May 24-25, 2011
Outline • NLTE effect • NLTE model implementation • NLTE model validation • Remaining issues • Summary
NLTE Effect Kinetic temperature Vibrational temperature ~ 3 mb Example of CO2 4.3 μm vibrational temperature
Fast NLTE Model i+1 i • J+1 • • • J • - Mean temperature from 0.005 To 0.2 mb - Mean temperature from 0.2 To 52 mb - Sensor zenith angles at the Earth surface, corresponding airmass = 1.0, 1.25, …, 3.75 and 4.0 - Sun zenith angles (= 0, 40, 60, 80, 85 and 90) - Regression coefficients, derived from LBLRTM and UMBC 48 profiles.
Fitting Error RMS Difference Mean Difference Very small fitting error
CRTM NLTE simulation vs observation example 1Solar zenith angle = 30o, sensor zenith angle = 0.7o AIRS
Differences of simulated and observed radiances in solar zenith angle range from 0o to 30oat four sensor scan angles The two channels have similar weighting functions Daytime
Differences of simulated and observed radiances in solar zenith angle range from 90o to 120oat four sensor scan angles The two channels have similar weighting functions Night Time
Issues • CRTM underestimates the NLTE correction term by up to 1 – 1.5 K • Factors that may cause the error: • The LBLRTM (v 11.7) used for this work applies the same vibrational temperature profile to all lines of a given state, ignoring the isotopolgue number of the line. The estimated error is up to 0.2 K • Error in LBLRTM CO2 v3 line strengths, as suggested by the night-time AIRS-model comparison plot, in which the NLTE effect is zero but the difference is still significant (underestimate about0.5K) ; the difference could be amplified due to NLTE during the day.
Summary • The NLTE can have a largeeffect (~10K) in daytime. • The fast NLTE module can significantly improve the accuracy of daytime shortwave radiance calculations. • The CRTM/NLTE model underestimates the NLTE affected radiance by about 1.5 K. • We are closely working with AER to address the remaining errors. • Possible error sources • LBLRTM (v11.7) does not take into account that different isotopologues have different vibrational temperature profiles (up to 0.2K). • LBLRTM CO2 ν3 line strength error as suggested by the nighttime AIRS M-O spectral difference.