The model evaluation metrics for smoke AOD forecasts currently comprise of mean bias error (BIAS), root mean square error (RMSE), correlation coefficient (CORR) and fractional gross error (FGE) as shown below, with c and o referring to the forecasted and observed quantity respectively. The AERONET AOD closest in time to the model outputs within a 30-minute window are matched in computing the evaluation metrics.

The monthly and annual statistics are computed retrospectively. The seasonal statistics are derived for February – April (FMA), May – July (MJJ), August – September (ASO) and November – January (NDJ) to best group the dominant regional monsoonal circulations with regional burning processes observed annually.