Fog probabilistic forecasting using an ensemble prediction system at six airports in Iran for 10 fog events

Abstract

Low visibility conditions due to fog affect air traffic and, in some cases, are the leading cause of aviation accidents. Accurate forecasting of fog can lead to a significant reduction of human and financial losses. Numerical weather prediction (NWP) models have limitations in simulating fog accurately for operational purposes. However, recent studies have confirmed that ensemble-based forecasts are effective in fog forecasting. In this study, a multi-physics ensemble prediction system (EPS) was used to simulate the occurrence of several radiation, cloud base lowering (CBL), and advection fog events that occurred at Arak, Ardebil, Hamedan, Orumiyeh, Rasht and Shahrekord airports located across Iran, during January and December 2015. The multi-physics EPS including a reference deterministic forecast consists of 16 different configurations of the weather research and forecasting (WRF) model and was run from 12 UTC 12 January to 12 UTC 16 January 2015 and from 00 UTC 27 December to 12 UTC 31 December 2015 to simulate fog occurrences at six airports. SW99 and G2009 visibility algorithms were applied to the EPS outputs to predict the fog. The advantages of probabilistic fog forecasting in this study were shown by comparing the reference forecast and the ensemble-based forecasts of fog events. The results showed that by considering a probability threshold of 37.5%, 50% and 62.5% for the ensemble forecasts, the EPS outperformed the deterministic fog forecasts obtained from a deterministic reference forecast. Such that, with 37.5% and 50% probability thresholds, the equitable threat score (ETS) was much higher than the ETS for the reference deterministic fog forecast. Also, the EPS with 37.5% and 50% probability thresholds could correctly predict 3 and 4 (out of 10) more fog events compared with the reference deterministic forecast.