On the causes of systematic forecast biases in near-surface wind direction over the oceans

TitleOn the causes of systematic forecast biases in near-surface wind direction over the oceans
Publication TypeTechnical memorandum
Date Published06/2020
Secondary TitleECMWF Technical Memoranda
AuthorsSandu, I, Bechtold, P, Nuijens, L, Beljaars, A, Brown, A

The demand for more accurate forecasts of near-surface weather is growing at a rapid pace, especially in the context of increased use of renewable energy. Although improvements in numerical weather prediction systems have lead to improved forecasts of near-surface weather parameters over the years, systematic biases remain. Here we examine one of the most longstanding biases in ECMWF forecasts of near-surface weather, the bias in surface wind direction over the oceans. Since the 1990’s the forecasted surface wind direction is rotated clockwise (anticlockwise) with respect to scatterometer (ASCAT) observations in the Northern (Southern) Hemisphere. Using a conditional analysis of the short-range forecast errors against ASCAT observations, we demonstrate that unstable boundary layers contribute the most to the remaining mean bias in surface wind direction over the oceans (3-5 degrees), although stable boundary layers that occur in the mid-latitude storm tracks during summer also contribute to some extent. Focusing on a typical trade wind region upstream of Barbados and using sensitivity experiments with the Integrated Forecasting System, we demonstrate that the surface wind direction bias in unstable boundary layers is related to the representation of turbulent and shallow convection momentum transport, and in particular to an apparent lack of friction in the lower part of the cumulus layer. We also show that both the forecast and analysis wind profiles are strongly sensitive to the momentum transport by shallow convection in the lowest 1.5 km in this region, which suggests that the analysis is only weakly constrained by wind observations in the trade wind boundary layer (at least prior to the assimilation of Aeolus observations). An evaluation of both forecast and analysis wind profiles with in-situ observations such as those provided by the EUREC4A field program, which recently took place around Barbados, would therefore be very valuable in the future.

PDF icon Download (4.56 MB)