- Luke Boxall
Capturing the Cold Model flip
The first few days of April witnessed a strong cold flip in the models over Germany having previously been solidly warm, in this blog I will share some guidance on how this could have been spotted.
The last weekend of March saw the models trend progressively warmer over Germany for Week 15, so that by the time the markets opened on Monday 1st April - the models were showing a solid warm signal over Germany (EC ENS 12z from the 31st March is shown below - this is the run that is available first thing - shows strong signal for above normal temps for Week 15 - 8th-14th April)
It's hard to argue against such a solid warm signal as this and as such the market was very confident on the idea of warm weather for week 15.
I remember looking at this signal first thing and thinking - "Looks warm next week" as well - it's the main signal the models were showing. However I then analysed the pattern the models were going for using the METSET tool (https://www.met-set.com) from the same run (EC ENS 12z from the 31st March - GFS also showed a similar picture)
There was strong support to move into a HNFZ pattern for week 15 - with HFZ pattern the next favourite - a difference of opinion between the forecasted pattern and forecasted temperature quickly became apparent. Below shows the pattern normals for HNFZ/HFZ:
Historically both HNFZ and HFZ are patterns that bring widespread cool risk (blue circles are below normal temperatures) - the fact that we had a strong signal in the 6-15day window for a sustained period for a cooler pattern flags up the risk that we actually end up trending cooler. Only 20% of the members went for a warmish pattern (HNFA/SZ) yet a good 60-70% of the members wanted a warm picture if you looked at temperature plumes - something had to give.
Sure enough by Weds 3rd April the model temperature had given way - a substantial cold flip had occurred in the models between 1st April and the 3rd April, the EC ENS 0z run from the 3rd April is shown below:
The forecast for week 15 now showed a strong below normal risk - which better correlated with the cooler risk suggested by the pattern - clearly showing the added value of having access to a pattern classification tool.
This example explains the added value of when the Pattern tool very clearly goes for a particular pattern type in the day 5-15 window. There was a similar situation about a month before in late February - after a fairly calm spell - the models were suggesting around 20-25GW of DE Wind to push in for the start of March.
This looked like a big enough wind signal as it was much higher signal at day 7-15 than anything else that had been seen up until that point - however when viewing through the METSET tool - there was 60-80% support for a WZ pattern to set in for week 10 - the pattern normal of DE Wind for WZ in early March is around 30GWs. It out-turned at 32GWs.
The pattern tool won't always go clearly into a specific pattern in the 6-15 Day window - but that doesn't mean there isn't added value from that - the current outlook for week 17 has no clear favourite pattern at all - multiple patterns with just 1 or 2 members going for it.
The rule of thumb in this situation being - unless there are any clear teleconnections to tell you otherwise - this lack of clarity in the predcited pattern suggests no confidence/no skill in the forecast at that range.