Application of MM5 in China: Model evaluation, seasonal variations, and sensitivity to horizontal grid resolutions

The rapid growth of energy consumption in conjunction with economic development during past decades in East Asia, especially China, caused severe air pollution problems at local and regional scales. Understanding of the meteorological conditions for air pollution is essential to the understanding of the formation mechanism of air pollutants and the development of effective emission control strategies to reduce air pollution. In this paper, the Fifth Generation National Center for Atmospheric Research (NCAR)/Pennsylvania State University (PSU) Mesoscale Model (MM5) modeling system is applied to simulate meteorological fields during selected six 1-month periods in 2007/2008 over a triple-nested modeling domain covering East Asia, the eastern China, and Shandong Province at horizontal grid resolutions of 36-, 12-, and 4-km, respectively. MM5 generally reproduces well the observations in the eastern China but performs worse in the western China and northeastern China. Largest biases occur in 2-m temperatures (T2) and wind speed and wind direction at 10-m in haze months (i.e., winter) and daily mean precipitation (Precip) in non-haze months (i.e., summer), due to limitations of the model in simulating snow cover and convective precipitation. Meteorological predictions agree more closely with observations at urban sites than those at the coastal and mountain sites where the model performance deteriorates because of complex terrains, influences of urban heat island effect and land/sea breezes, and higher elevations. Model results at 12-km in Shandong Province show an overall better performance than those at 4- or 36-km while the results at 4-km show worst performance due to inaccurate land use and the model’s incapability in simulating meteorological processes at a fine scale.

► MM5 reproduces observations in eastern China but performs worse in other regions.
► T2 & precip. has largest biases, due to inaccurate snow cover & convective precip.
► Predictions agree better with observations at urban sites than other sites.
► Results at 4-km are not the best due to inaccurate land use and model limitations.