Forecasting eclipse weather

By Peter Clark                                    Department of Meteorology homepage

Our weather is driven by energy radiating from the sun. The annual cycle of the earth’s tilt with respect to the sun is so profound that we identify our seasons by their profoundly different weather. The daily cycle is less profound but still has a clear impact on the weather we experience near the surface, primarily through the warming up during the day and cooling at night of the land surface.

The solar eclipse later this week (Friday 20 March) will only last a few of hours in total, and at any one place the significant effects on the radiation we receive from the sun will only last an hour. It is worth asking the question ‘will this have a significant effect on our weather?’. Indeed, do we need to take account of it in our weather forecast?

To answer this, the Met Office will be running a second version of their numerical weather prediction model for the UK that takes account of the impact of the eclipse on the incoming short-wave radiation from the sun. It is a tribute to the re-usability of computer code that, to enable this, I have been able to re-use code first written by Stephen Cusack in the Met Office to study the impact of the eclipse in 1999! Unfortunately, while the physics of eclipses hasn’t changed since then, the Met Office forecast model has completely changed, so some work has been needed to adapt this code to the new model.

The capabilities of the forecast model have changed too. Back then, the model horizontal grid length was about 12.5 km; now it is 1.5 km over the UK. This is the size of the boxes we divide the atmosphere into to compute the impact of the laws of dynamics and thermodynamics that drive the weather. It is very relevant in this case because we expect the impact to be on the temperatures and hence winds nearest the surface, which we know are strongly influenced already by small-scale features like hills, coasts and changes in land surface.

(The model is actually known as UKV to denote the fact that away from the UK it has variable resolution- see

We have tested this by pretending that the eclipse took place on a day in February (the 9th). The impact on near-surface temperature near the peak of the eclipse can be seen in Figure 1. The pattern looks very complex! This is because a lot of the country was covered by low cloud at the time. Here the impact is very small; 0.1-0.2 degC cooler. However, some areas are not cloudy, in particular around Yorkshire, parts of Norfolk, south Wales and eastern Scotland, where the impact is typically 2 degC cooler – more in places. This illustrates the importance of cloud on the day! If there is a lot of cloud cover, most of the solar radiation gets reflected back to space anyway. Note that the impact over the sea is negligible.

2015 03 13 Peter Clark Fig 1

Figure 1. Forecast temperatures (in °C) 1.5 m above the ground at 1000 UTC on 9 February (from a forecast starting 0300 UTC); left, temperatures if there had been an eclipse with the same path as on 20 March; centre, the standard forecast; right, the difference between the two, in degC.

Are there any other effects? We expect the changes in temperature to produce small changes in pressure and also changes in the turbulent mixing near the surface, both of which affect the wind. Figure 2 illustrates the predicted impact on a portion of the UK. The impact is really small – less than a metre per second. It is greatest in the sunny areas, and the flow then is influenced by the hills such as the North Yorkshire Moors, but is noticeable beneath the cloud as well. Without detailed analysis, it is hard to say why, but one explanation may be that the turbulent mixing associated with the cloud has been changed by the reduction of solar heating at the top of the cloud.

2015 03 13 Peter Clark Fig 2

Figure 2. Forecast winds  (speed and wind vectors) at 10 m, as Figure 1, for an area of the UK. Difference arrows are not drawn where the speed difference is less than 0.1 m/s.

All being well, the Met Office will be including this change in a parallel run of their UKV forecast model from the run starting 0300 UTC on Thursday 19 March, providing a 36 hour forecast, and we will have a good idea of the likely impact well ahead. Thereafter, it will run every 6 hours till the eclipse is over. If time permits, they will also include it in their global model, though the impacts will be less discernible in this much lower-resolution model.

Additional links

The eclipse takes place on Friday morning, 20 March. In Reading, first contact is at 0824 UTC, the maximum obscuration (85%) is at 0930 UTC, and last contact is at 1040 UTC. Times and maximum occultation for other localities, and more details of the eclipse path, can be found at NASA’s eclipse website.

The University of Reading Department of Meteorology is running a large citizen science experiment during the eclipse, the National Eclipse Weather Experiment or NEWEx. There are links on this page to results, displayed in near-real time, which will include frequent (5 minute cadence) weather observations from across the UK as well as weather balloon ascents across the path of the eclipse from Reykjavik in Iceland, to Thorshavn, Faeroe Islands (in totality), to Lerwick (Shetland) and Reading. We invite all interested parties, including schools, to take part in this unique experiment; see notes for schools.




This entry was posted in Eclipses, Measurements and instrumentation, Numerical modelling, Space, Weather forecasting and tagged . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *