Four editions of the High Resolution Transmission （HITRAN） databases （HITRAN96, HITRAN2K, HITRAN04, and HITRAN08） are compared by using a line-by-line （LBL） radiative model in the long-wave calculation for six typical atmospheres. The results show that differences in downward radiative fluxes between HITRAN96 and HITRAN08 at the surface can reach a maximum of 1.70 W m-2 for tropical atmospheres. The largest difference in heating rate between HITRAN96 and HITRAN08 can reach 0.1 K day-1 for midlatitude summer atmosphere. Uncertainties caused by line intensity and air-broadened half- widths are also evaluated in this work using the uncertainty codes given in HITRAN08. The uncertainty is found to be 1.92 W m-2 for upward fluxes at the top of the atmosphere （TOA） and 1.97 W m-2 for downward fluxes at the surface. The largest heating rate caused by the uncertainty of line intensity and air-broadened hMf-width can reach 0.5 K day-1. The differences in optical depths between 1300 and 1700 cm-1 caused by different HITRAN versions are larger than those caused by the uncertainties in intensity and air-broadened half-width. This paper suggests that there is inaccurate representation of line parameters over some spectral ranges in HITRAN and more attention should be paid to these ranges in fields such as remote sensing.
Characteristics of cloud overlap over Eastern Asia are analyzed using a threeyear dataset （20072009） from the cloud observing satellite CloudSat. Decorrelation depth Lis retrieved, which represents cloud overlap characteristics in the simulation of cloudradiation processes in global climate models. Results show that values of L in six study regions are generally within the range 03 km. By categorizing L according to cloud amount in subregions, peak L appears near subregions with cloud amount between 0.6 and 0.8. Average L is 2.5 km. L at higher altitudes is generally larger than at lower lati tudes. Seasonal variations of L are also clearly demonstrated. The sensitivity of cloud radiative forcing （CRF） to L;y in Community Atmosphere Model 3.0 of the National Center for Atmospheric Research （CAM3/NCAR） is analyzed. The result shows that L can have a big impact on simulation of CRF, especially in major monsoon regions and the MidEastern Pacif ic, where the difference in CRF can reach 4050 W m2. Therefore, accurate parameterization of cloud vertical overlap struc ture is important to CRF simulation and its feedback to climate.
