Abstract

Abstract. Modern coupled general circulation models produce systematic biases in the tropical Atlantic that hamper the reliability of long-range predictions. This study focuses on a common springtime westerly wind bias in the equatorial Atlantic in seasonal hindcasts from two coupled models – ECMWF System 4 and EC-Earth v2.3 – and in hindcasts also based on System 4, but with prescribed sea-surface temperatures. The development of the equatorial westerly bias in early April is marked by a rapid transition from a wintertime easterly, cold tongue bias to a springtime westerly bias regime that displays a marked double intertropical convergence zone (ITCZ). The transition is a seasonal feature of the model climatology (independent of initialisation date) and is associated with a seasonal increase in rainfall where a second branch of the ITCZ is produced south of the Equator. Excess off-equatorial convergence redirects the trade winds away from the Equator. Based on arguments of temporal coincidence, the results of our analysis contrast with those from previous work, and alleged causes hereto identified as the likely cause of the equatorial westerly bias in other models must be discarded. Quite in general, we find no evidence of remote influences on the development of the springtime equatorial bias in the Atlantic in the IFS-based models. Limited evidence however is presented that supports the hypothesis of an incorrect representation of the meridional equatorward flow in the marine boundary layer of the southern Atlantic as a contributing factor. Erroneous dynamical constraints on the flow upstream of the Equator may generate convergence and associated rainfall south of the Equator. This directs attention to the representation of the properties of the subtropical boundary layer as a potential source for the double ITCZ bias.

Highlights

  • The identification and reduction of systematic biases in coupled general circulation models (CGCMs) have been ongoing challenges in model development in recent times

  • We have used a set of seasonal hindcasts to investigate the origins of the systematic westerly wind bias that occurs on the Equator in the tropical Atlantic during spring

  • Coupled initialised hindcasts from European Centre for MediumRange Weather Forecasts (ECMWF) System 4 and EC-Earth v2.3 have been used alongside a set of hindcasts from System 4 with prescribed sea-surface temperature (SST)

Read more

Summary

Introduction

The identification and reduction of systematic biases in coupled general circulation models (CGCMs) have been ongoing challenges in model development in recent times. Despite significant development of CGCMs in the last two decades, large systematic biases remain in the simulated tropical climate – including the tropical Atlantic (Solomon et al, 2007; Davey et al, 2002; Richter and Xie, 2008; Toniazzo and Woolnough, 2014) These biases can have significant impacts on seasonal forecasts and future climate predictions. We analyse systematic biases in the tropical Atlantic in seasonal hindcasts ( sometimes referred to as reforecasts) obtained with the System 4 coupled seasonal prediction system of the European Centre for MediumRange Weather Forecasts (ECMWF) These are compared with hindcasts obtained by prescribing the time-dependent SST fields from observations in order to isolate biases originating from the atmosphere component of the model and those originating from air–sea coupling.

Approach and method
Annual cycle of climatological biases
Comparison of the biases in the three models
A focus on the onset of the westerly wind bias
Relationships with other model biases
Discussion: mechanisms of bias development
Summary and conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call