Evaluating WRF-CHEM MULTI-SCALE Model In Simulating Aerosol Radiative Properties Over The Tropics - A Case Study Over India

Abstract

We evaluated the performance of WRF-Chem multi-scale model over the tropics, to simulate the regional distribution and optical properties of aerosols, and its effect on radiation over India for a winter month. The model is evaluated using measurements obtained from upper-air soundings, AERONET sun photometers, various satellite instruments, and pyranometers. The simulated downward shortwave flux was overestimated when the effect of aerosols and clouds, on radiation, was?2 neglected. The simulated downward shortwave radiation was 1 to 20 Wm closer to the observations when we included aerosol-cloud-radiation interaction in the simulation. The model usually underestimated particulate concentration for the few observations available. This is likely due to turbulent mixing, transport errors and the lack of dust emission/scheme and the secondary organic aerosol treatment in the model. The model efficiently captured the broad regional hotspots such as, higher aerosol optical depth over the northern parts of India, especially over the Indo-Gangetic basin and lower aerosol optical depth over southern parts of India. The regional distribution of aerosol optical depth agreed well with the AVHRR aerosol optical depth and the TOMS aerosol index pattern. The magnitude and wavelength-dependence of simulated aerosol optical depth was also similar to the AERONET observations across India. The difference in surface shortwave radiation between two simulations that included and neglected aerosol-radiation (aerosol-radiation-cloud) interactions?2 were as high as ?25 (?30) Wm?2. The spatial variations of these differences were also compared with the AVHRR observation. This study suggests that the model is able to qualitatively simulate the distribution of particulates and its impact on radiation over India; however, additional measurements of particulate mass and composition are needed to fully evaluate the model performance.