Completion of gaps in 10-min air temperature series from automatic weather stations in Belgium

Air temperature is historically monitored in Belgium by a network of climatological stations. Most of these stations are till nowadays mainly manual: they rely on a volunteer observer who records every morning the daily extreme air temperatures by reading liquid-in-glass thermometers (including mercury thermometers) placed in Stevenson screens. Since 2020, the UNEP Minamata Convention on Mercury bans all production, import and export of observing instruments containing mercury. In the recent years, various alternatives to mercury thermometers have therefore been considered and tested by the Royal Meteorological Institute of Belgium (RMI). Parallel measurements have been performed with several synoptic weather stations in Belgium. Based on these comparisons as well as on practical arguments (e.g., the price as well as the ease of installation, maintenance and operation), a compact low-cost automatic weather station was selected by RMI to transition away from the conventional manual climatological stations. In 2022, almost 50 of these compact stations have already been installed and around 50 further ones are planned in the coming months. These stations provide 10-min observations in real-time through wireless data transmission. 

In practice, the communication of these compact stations is however not perfectly robust as data are regularly missing due to frequent down-time of usually short duration. As these gaps in the time series can be problematic for end-users applications, a procedure to automatically provide estimations for all missing data is needed. In the complete data processing chain (i.e., from the sensor to the end-user), this automatic data completion step is performed just after the real-time data quality control (i.e., quality tests that directly eliminate obvious errors). It is followed by a final data validation (i.e., detailed but delayed and partly manual verification of the data to ensure the quality of the climate archives). Observations and estimations are labeled by distinct quality flags in the climate archives.

This contribution provides an overview of the recent developments regarding the automatic completion of gaps in 10-min air temperature series. Various approaches including inverse distance weighted interpolation (IDW), principal component analysis (PCA), linear interpolation and linear regression have been compared in order to select the method to be implemented in the operational data processing chain.