An Analysis and Case Study of Using Mobile Crowdsourcing for Weather Forecast Verification

Forecast verification is the subject of measuring the goodness of the forecast. In atmospheric science, the current practice of weather forecast verification is to compare observed data to past forecasts from weather services. The aim of this paper is to study a situation in which probabilistic rain forecasts are verified against crowdsourced data, in addition to station data. Weather reporting tasks were outsourced to participants of a mobile social application called SWUA in Bangkok. We show that crowdsourced data are able to make biases and weaknesses of forecast services more prominent using the Reliability diagrams and the Brier scores. The results should call for further attention to the application of crowdsourced data for weather forecast verification in the future.