Evaluating sunshine hour approximation for biometeorological indices

Abstract Sunshine influences a range of biometeorological parameters, from human health to the enjoyment of touristic destinations. Sunshine has therefore become an important meteorological variable in biometeorological indices and the causal analysis of the impact of climate on humans. However, ground-based recording of sunshine hours is sparse, particularly in African countries. Methods to estimate sunshine hours from more ubiquitously measured or modelled meteorological parameters offer the prospect to supplement these datasets, and are thus appealing for biometeorological studies. It is therefore important to critically explore and test the validity of these approaches at country-level resolution. In this study, we explore the accuracy of a method to estimate sunshine hours from solar radiation that is used in the Camping Climate Index for the United States, and test the suitability of the substitution for South Africa. We find that for each station in South Africa, at least 65% of the values estimated by this equation are completely erroneous, either returning negative values for sunshine hours or a total of sunshine hours for a day in excess of the duration between sunrise and sunset. Further comparison of the estimated data against measured data reveals that for at least half of the months of the year for each station, the differences in values are statistically significant. We therefore strongly caution against the use of this particular method for estimating sunshine hours in South Africa, and encourage similar tests before applying it elsewhere. More broadly, we advocate for careful ground-truthing of estimated meteorological variables before inputting them into biometeorological indices or models.