• Luke Boxall

A win for the seasonal models

I think it's fair to say that the seasonal models have done a remarkable job in predicting the dominant westerly regime that has persisted throughout and looks set to continue for Winter 2019-2020.

Image above shows the ECMWF seasonal outlook for DJF Winter 2019-2020 from Autumn 2019 and showed a remarkable consistent forecast for a strong Atlantic low pressure bias with Mainland Europe high pressure - this was also shown clearly on other forecasting systems as well such as the NCEP model and the UK MetOffice System.

The above image is what out-turned from 1-Dec-2019 until 28-Jan-2020 and shows the same Atlantic low pressure /Mainland Europe high pressure signature - and the forecast going forwards into February, at the time of writing, shows the continuation of a dominant Atlantic low pressure pattern across all computer models - as illustrated by the GloSea5 Monthly outlook for February issued 29-01-2020:

This isn't always the case though - often there is no model skill before November - although there is some forecast skill for model seasonal forecasts issued in November for the N Atlantic region - as shown below in the anomaly coefficient score from November for DJF Period in the ECMWF forecasting system - the same chart for October shows a tiny 0.2 spot in the mid-Atlantic

Which means normally we end up with wildly different forecasts from the Sept-Nov start dates - take last years Autumn forecasts for winter 2018-19 as an example:

What was different about this winter to get such a clear signal?

You can often get an understanding of the risks for the upcoming winter by exploring a number of teleconnections - last Summer a paper was released by UCL that explained why we were at risk of a Cold Jan/Feb 2020 due to the current state of the QBO (Quasi-Biannual Oscillation - Tropical Stratospheric winds) and the Solar Cycle:

However there are a number of other teleconnections to consider as well - with Sea Surface Temperature (SST) globally and the state of the Polar Vortex (Polar Stratospheric Winds) being more crucial - especially for the start of Winter-time.

When I gave my Winter Forecast back in September - it became apparent that there was one key area that seemed to be dominating the outlook. The IOD (Indian Ocean Dipole).

I often look at West Indian Ocean temperatures as there has been research to suggest that warm SSTs here - brings an increased risk of +NAO. Normally though it gets blurred out by other signals - primarily from the more dominant Pacific signals (El-Nino/La-Nina) - however this year - with a neutral Pacific and a very strong warm pool developing over the West Indian Ocean it was clear the main tropical signal was going to come from here - which is something I discussed with my clients along with this slide:

On this slide - I compared the forecasted precipitation from the ECMWF September forecast for DJF (top) to note the expectation that this IOD event was going to be big and drive the Winter's Weather - a teleconnection that the model often handles well (as shown on the bottom of the page that a Wet/dry signal over West/East Indian Ocean corresponds to a low pressure signal over N Atlantic/N Europe and high pressure over S Europe.)

Obviously this signal has a stronger influence in the regions near-by and has been responsible for the East-African Locust Plague from the recent heavy rains there

To the major drought/very hot weather that Australia has experienced for there summer

Back to the Northern Hemisphere - our fate was sealed when it become apparent a very strong westerly vortex was forming aloft (Polar Vortex) during the late Autumn - whilst a surface block feature did last longer through November than forecasts suggested - the strong vortex did eventually made connection to the surface and hasn't look back since

Going forwards

My worry going forwards is an over-reliance on seasonal models and also the IOD to predict the correct weather pattern - this winter saw one of the warmest IOD events on record - combined with the lack of strong Pacific Ocean anomalies meant this was the driving force and led to a huge global response - a teleconnection that the model handles well - it's not a common occurrence - so caution is advised in extrapolating these signals as "the thing" to use for next Winter's weather - however at least the IOD now has more recognition for the potential impacts that it can have on the global weather not just in the local vicinity.

Another important point to mention is the observed skill in the MetOffice GloSea Forecasting system in predicting the 10mb Zonal wind - this is a key index for measuring whether the Polar Vortex is strong or weak - strong vortex is strongly correlated to +NAO and westerly regimes whereas a weak vortex brings increased risk of strong blocks and cooler weather for mid-latitudes - so having skill in this index beyond the front 3 weeks is useful for forecasting the winter weather as we move into the season so a continued effort to keep an eye on the stratosphere is useful - providing you use the correct models for it i.e. ECMWF/GloSea rather than CFS/GFS which has much weaker skill.


The Weather Perspective Ltd