First peek at the weather for the Wellington Round the Bays 2012

What’s the weather going to be like for the Wellington Round the Bays race on 26th Feb 2012? Personally I’m hoping for nice cool southerlies, as opposed to strong northerlies or humid northeasterlies, and not too much rain. We’re still ten days away which is really stretching it in terms of what weather forecasts can deliver, but I thought some crystal ball gazing would be interesting. Numerical computer models at the European Centre for Medium Range Weather Forecasting (ECMWF) routinely run out as far as 10 days (and further), so they offer a first guess at what the weather might have in store.

Of course it is best not to take any of these forecasts too literally. “Weather is a mixture of pattern and chaos”, and the further out forecast models are run the more the chaos tends to take over, so a forecast for 10 days ahead has to be taken with a big pinch of salt. The sensible approach is to look for the pattern in the data to see what we can learn from it, rather than get carried away by small details which may be the result of chaos.

All the charts below are for 1pm local time on Sunday 26th Feb, and they paint the following picture: A high over the Tasman Sea extending a ridge over the upper North Island, and a trough moving up the South Island. A very common weather pattern which would suggest freshening northerly winds for Wellington on Sunday, followed some time later by a southerly change. Probably not that far off a climatological average!

It is difficult to make any guesses about rainfall based on these charts because they do not depict moisture, and  it would depend on the timing of the trough. Figure 1 is the deterministic forecast, which has the trough over the lower South Island. From this we might guess that it would still be dry in Wellington. However, a look at the ensemble data in figs. 2 and 3 shows that the ensemble has a spread of opinions about where any trough might be. From fig. 3 we can see that there is reasonable confidence in the location of a high over the Tasman Sea (low values in the ensemble standard deviation), but much lower confidence in the pressure pattern over New Zealand (high values in the standard deviation). The low confidence in the pressure pattern over New Zealand is probably because the ensemble members all have different ideas about what the trough might do. It might even have passed by Wellington by 1pm.

For comparison, the Australian Bureau of Meteorology ACCESS-G model also runs out 10 days and portrays a similar story, but with a different timing on the trough. In fig. 4 there is a high out to the west of the country, while the trough lies over the central North Island. This suggests cool southerlies and showers for Sunday morning! I think I prefer to believe the ACCESS model compared to the ECMWF – Cool southerlies and some light showers would suit me perfectly.

It’ll be interesting to watch these forecasts evolve as we draw closer to the big day. Will they hang on to the same story of a high to the west and a trough moving north over the country? Or will a different prediction emerge? Keep an eye on the ECMWF and Weatherzone websites for the latest data. As we draw closer to the day other model data will also become available (the American GFS model goes out 7 days, see the MetVUW site), and of course more detailed forecasts will start to appear on the MetService website.

Click on any of the charts below to see the full size image.

Fig. 1 ECMWF H+240 MSLP/850hPa wind, 26th Feb 2012 00Z. (Image courtesy European Centre for Medium Range Weather Forecasting)
Fig. 2 ECMWF H+240 MSLP and ensemble standard deviation, 26th Feb 2012 00Z. (Image courtesy European Centre for Medium Range Weather Forecasting)
Fig. 3 ECMWF H+240 MSLP ensemble mean and normalised standard deviation, 26th Feb 2012 00Z. (Image courtesy European Centre for Medium Range Weather Forecasting)
Fig. 4 BOM ACCESS-G H+240 MSLP/THK/Precip 26th Feb 2012 00Z. (Image courtesy Weatherzone)


There are many centres around the world running global weather prediction systems, or “global models”. The most well known are in the US (at NOAA) and the UK (at the Met Office and at the ECMWF). Not all model data is available to the public free of charge though. It just so happens that the ECMWF puts a limited data set on its website for free, hence the above analysis is based on what’s publicly available from the ECMWF. The ACCESS-G (G for Global) model is run at the Australian Bureau of Meteorology and is available via the Bureau website, and also in very useful form on the Weatherzone website. As far as I know the ACCESS-G is a deterministic model.

The ECMWF runs computer simulations of the weather for the whole globe twice a day. Observations from ground stations, weather balloons and satellites are fed into the computer so it knows the current state of the atmosphere, then it runs forward in time to see how the weather will evolve. This is the “deterministic” model. The problem is that the weather evolves chaotically – Tiny variations in the initial state that the computer begins from, can grow into very big variations in terms of outcomes. And of course we can never know exactly the initial state of the atmosphere of the entire globe. This puts limits on how far we can usefully run deterministic models into the future. Of course you could let it run as long as you like, but after a few days or a week, or ten days, it will be completely wrong!

The idea of an ensemble is to run lots and lots of simulations together, starting from slightly different initial states. Sometimes it turns out that varying the initial state a little bit doesn’t make much difference to how the simulation evolves, so the variation in the outcome is small. So we might not know everything about the current state of the weather, but on certain days it doesn’t matter too much anyway. This means we can feel more confidence in the model output. On other days the opposite is true! The ensemble mean is the average of all the individual ensemble members. The standard deviation tells you how large the spread of all the ensemble members is. The larger it is, the less confidence there is in the forecast. The “normalised standard deviation” is roughly the same thing. See the ECMWF help page for the details.

All forecast models are co-ordinated against UTC time (Universal Time Co-ordinated), which is basically the same as GMT. 00Z is just shorthand for 00 UTC or 00 GMT – Midnight GMT time, or 1pm in New Zealand (during NZ daylight saving time). The valid time (VT) on the charts above is the date and time that the model data in the chart represents. “H+240” and “t+240” means the chart is valid for a time 240 hours after the model run was initialised. All the model runs above were initialised with observations from Thurs 16th Feb 2012 at 00Z. Hence the H+240 chart will be valid for a date and time exactly 10 days ahead, which is 26th Feb 2012 00Z.

MSLP = Mean Sea Level Pressure, in hectopascals (hPa).
850hPa wind speed = wind speed at a certain pressure level, an altitude of 850hPa which is about 1500 metres.
THK = Thickness, the difference in height between the 850hPa and 500hPa pressure levels. Lower values of thickness implies a colder atmosphere.
Precip = Precipitation, such as rain, snow, drizzle.

“Weather is a mixture of pattern and chaos” is a favourite phrase of New Zealand weather celebrity Bob McDavitt.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s