Development and Testing of a Decision Tree for the Forecasting of Sea Fog Along the Georgia and South Carolina Coast

Abstract

A classification and regression tree analysis for sea fog has been developed using 648 low-visibility (<4.8 km) coastal fog events from 1998–2014 along the South Carolina and Georgia coastline. Correlations between these coastal fog events and relevant oceanic and atmospheric parameters determined the range in these parameters that were most favorable for predicting sea fog formation. Parameters examined during coastal fog events from 1998–2014 included sea surface temperature (SST), air temperature, dewpoint temperature, maximum wind speed, average wind speed, wind direction, inversion strength, and inversion height. The most favorable range in SST for sea fog formation was 10.6–23.9°C. The most favorable gaps between air temperature and SST, dewpoint temperature and SST, and dewpoint temperature and air temperature were found to be –1.7– 2.2°C, 0°C, and 0–2.2°C, respectively. The most favorable range in maximum wind speed was 11.1–20.4 km h-1, and the most favorable wind directions were parallel to the coast or SST isopleths. The most favorable range in inversion height was 70.6–617.2 m, and the most favorable inversion strength was anything >6°C. Utilizing these eight predictors, a forecasting decision tree was created and beta tested during the 2016/2017 sea fog season. The decision tree successfully predicted sea fog on 17 of the 18 dates that it occurred (94%) and successfully predicted a lack of sea fog for 189 of the 194 days where sea fog did not occur (97%). Two of the six incorrect predictions appear to have extenuating circumstances. ABSTRACT (Manuscript received 11 December 2017; review completed 16 April 2018)