Evaluating Human-Processed Forecasts Provided to the Aviation Industry in South Africa

Abstract

The main objective of this study is to determine whether human-forecaster at the main airports of South Africa add value to the raw numerical weather prediction model output when they provide forecast services to the aviation industry. Data set pairs of weather observations made at these airports and terminal aerodrome forecasts were constructed for the three forecast systems: the human-forecaster, persistence forecast and the raw output from the 12 km resolution the Unified Model administered by the South African Weather Service. These three data set pairs were independently evaluated by the in house developed forecast verification system of the South African Weather Service. A Monte Carlo method was used to determine the significance of the verification results obtained from calculating the proportion correct, hit rate, false alarm ratio, critical success index, Heidke skill score and the Pierce skill score of the various forecasts. In general, it was found that the forecaster-adapted forecasts are superior to the raw model output, thus providing the evidence that the aviation industry may benefit most from forecasts routinely issued by South African Weather Service forecasters.