Addressing near-surface forecast biases: outcomes of the ECMWF project 'Understanding uncertainties in surface atmosphere exchange' (USURF)

The demand for more accurate near-surface weather forecasts is rapidly increasing, driven by various factors such as renewable energy applications, or the occurrence of more intense and frequent extreme events. In this context it is becoming increasingly important to reduce systematic biases in near-surface weather parameters, such as temperature, humidity or winds, which manifest at all forecast ranges. These biases are the result of a complex interplay between processes parametrized in the  atmospheric and surface columns of the forecasting system, which can lead to locally generated errors, and advection, which constitutes a non-local source of errors. Understanding the leading causes of these systematic errors, which often have complicated geographical patterns and temporal structure, is a necessary step to enhance the near-surface forecast accuracy and improve the realism of the model. This requires disentangling the role of individual processes by using a range of diagnostics for stratifying and attributing errors. The ECMWF internal USURF project, which was initiated in 2017, has coordinated efforts in this area, and its main results as well as the necessary model developments to reduce systematic biases in forecasts of near-surface weather parameters over continental regions are summarized here. This report was a special topic paper presented to ECMWF’s Scientific and Technical Advisory Committees in October 2020.