Atmospheric aerosol has already been well acknowledged as one of the largest sources of uncertaintyin climate models. However, most current operational numerical weather prediction (NWP) models rarely resolve the aerosol processes explicitly. Professor Aijun Ding and his group demonstrate that atmospheric aerosol is one of the important drivers biasing daily temperature prediction, and forecast errors are rapidly magnified over time in regions featuring high aerosol loadings.
By conducting an observation minus forecast (OMF) analysis with 3-year daily National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) data during 2016-2018, the authors found significant bias and expanding forecast errors in the daily forecast of lower-tropospheric air temperature in regions influenced by anthropogenic or natural aerosol. For example, strong negative temperature biases exist in central-southern Africa, Amazon, northern India and eastern China, and positive biases are over the southern Oceans in the high latitudes of the Sothern Hemisphere, the northern Atlantic, Siberia and eastern UnitedStates (US). The monthly-averaged bias in the 24-h temperature forecast varies between±1.5℃ in regions influenced by atmospheric aerosol. These biases agree well with the overall distribution of aerosol, which can be well explained by various types of aerosol, such as sea salt, mineral dust, and carbonaceous aerosol from biomass burning or fossil fuel combustions in different regions. The forecast bias in temperature has been proven to be related to aerosol in WRF-Chem simulations.
Fig. 1 The maximum monthly averaged OMF Tbias for the GFS 24-h forecast during 2016-2018 and the global distribution of the averaged MISR aerosol optical depth (AOD) during 2016-2018.
Case studies have shown that forecast errors are magnified in regions featuring high aerosol loadings instead of high cloud cover.Various kinds of aerosol caninfluence the lower-tropospheric air temperature viaeither aerosol-radiation (ARI) interaction or aerosol-cloud interaction (ACI), and ARI effect on temperature is highly dependent on the albedo of the underlying surface.To improve the overall forecasting skills with respect to the impact of aerosol,the great importance of aerosol and meteorology-chemistry coupling should be highlightedin the NWP application.
Fig.2 Relationship of temperature forecast bias Tbias with ARI and cloud bias over landand sea.
Xin Huang is the first author of this paper and Aijun Ding is the corresponding author. This work was published in Science Bulletin and was supported by the National Science Foundation of China (41725020 and 41922038).
Huang, X. & Ding, A., Aerosol as a critical factor causing forecast biases of air temperature in global numerical weather prediction models, Science Bulletin, https://doi.org/10.1016/j.scib.2021.05.009, 2021.