Correcting citizen-science air temperature measurements across the Netherlands for short wave radiation bias

Citizen-science thermometer measurements have the potential to provide information about surface air temperature fields on scales smaller than is typically quantified by the official monitoring network. As such, national meteorological services are becoming increasingly interested in these measurements as a possible source of data for use in weather monitoring or forecasting. However, in order for the information to be used, biases in the data need to be assessed. The most important source of bias is the potential overheating of the thermometer due to inadequate shielding or exposure. Previous research has indicated that information about the nature of the instrument and its exposure is important for correcting this bias. However, in the majority of cases this information is unavailable for amateur stations. In this paper a statistical correction for short wave radiation bias is developed for the air temperature data recorded at 159 Weather Observations Website (WOW) stations across the Netherlands during the period 2015-2016. Generalized additive mixed modelling (GAMM) is used to quantify and correct for short wave radiation bias in the hourly measurements using a background temperature field generated from the official 34 automatic weather stations along with satellite-derived short wave radiation estimates. It is demonstrated that the corrected WOW data add local detail to the hourly temperature field, which may provide a useful source of data to supplement official measurements.