An evaluation method of the absorption process with a new radiation code for CCSR/NIES AGCM

It is required the radiative forcing needs to be estimated within an accuracy of about 0.1 W/m² by the progress of optical modeling of the earth's atmosphere and the progress of computers. Under this situation there is a large demand for an accurate yet rapid radiation transfer scheme applicable to atmospheric dynamics modeling. The broadband radiative transfer code “mstrn8”, which was developed by Center for Climate System Research (CCSR) about ten years ago and was implemented in the CCSR/NIES atmospheric general circulation model (AGCM), has an error of 10% in the radiative heating rate calculation and start having a difficulty to meet this demand. The most difficult part of accurate and efficient radaiative transfer calculation is to evaluate light absorption by atmospheric molecules. It is known that CCSR/NIES AGCM has a cold bias around the tropopause that seems to be caused by the error of the mstrn8 code. In this study, we develop an updated code, named "mstrnX", by improving the gaseous absorption process to solve these difficulties met by mstrn8. At first, the absorption line database is replaced by the latest versions of the Harvard-Smithsonian Center, HITRAN2004, and the program of continuum absorption is also replaced by MT_CKD_1.0. By this update, the light absorption by water vapor is significantly changed in the longwave region and absorption by the Chappuis band of ozone is substantially revised in the shortwave region. The effects are extended up to 0.3 K/day in the troposphere in the longwave region and up to 0.1 K/day in the upper stratosphere in the shortwave region. Furthermore, the heating rate calculated by mstrn8 has errors about 5.2% in the longwave region and 3.6% in the shortwave region as compared to the result evaluated with all the absorption lines of major seven gas species, so that a new criterion for selecting absorption lines is proposed in this study. An optimization method is adopted in mstrn8 to decrease the number of integration points for wavenumber integration using the correlated k-distribution method and to increase the computational efficiency in each band. The integration points and weights of the correlated k-distribution are optimized for accurate calculation of the heating rate up to altitude of 30 km in mstrn8. We extend the maximum altitude up to 70 km for accurate calculation of the heating rate. For this purpose we adopted a new non-linear optimization method of the correlated k-distribution and studied an optimal initial condition and the cost function for the non-linear optimization. For the application of global warming experiments, the integration points and weights in each band have to be determined to generate plausible results both in standard and doubling CO₂ conditions. With these improvements, the new radiation code computes the flux with errors less than 1.5 W/m² and heating rate with errors less than 0.2 K/day in the troposphere and the lower stratosphere; 1.0 K/day in the upper stratosphere in longwave region and less than 0.2 K/day in shortwave region. The cold bias in the simulation by the CCSR/NIES AGCM is largely decreased by the improved radiation code developed in the present study.