18 August 2017

Numerical Weather Model Adjusted For Eclipse Forecast

Posted by Dan Satterfield

There are a lot of experiments and observations that will be made during the eclipse on Monday, but I just heard of one tonight that is right up my alley! First, I have to explain some things. One of the better high-resolution weather models we use every day is called the HRRR for High-Resolution Rapid Refresh model. This model does an excellent job and has sophisticated physics that runs at a resolution of just 3 kilometres.

Now the  bad news. This model takes a lot of time to run on a super computer and it only makes a forecast for the next 18 hours! Since this model is run every hour, it does give us forecasters a good idea of how stable the forecast is (A large change from run to run indicates poor predictability). But here is the best news. A similar model to the HRRR has been adjusted to “know” that the eclipse on Monday is happening!


This image from ESRL shows the difference between the regular HRRR model and the model that “knows” of the eclipse. This was a simulated test run to look at how the model would react. Click on the link below for more details on the change tot he HRRRx model.

The other model is called the HRRRx and this model is run experimentally at the Earth Systems Research Lab (ESRL) in Boulder, Co. It is run out to a forecast of 36 hours at least once a day and several times out to at least 18 hours. It has been updated to know that an eclipse is occurring, and it will adjust the incoming solar radiation as it happens! Not only that, but the model is set to do this with any eclipse from Monday all the way through 2050! Eventually, the HRRRx model will likely become the operational HRRR model, but since it’s in test mode, it can be more readily tinkered with than the front line operational NOAA models.

We expect that the temperature will drop as the energy from the sun gets blocked on Monday and the greater the eclipse, the bigger the drop. Areas in the west where the air is dry will see the biggest temperature fall  (3-7°C?) with more humid areas seeing less of a drop since water vapor is a powerful greenhouse gas. That temperature drop will also have an impact on clouds, precipitation and winds, and since the model will know that the energy is dropping it should adjust the forecast for it. It will be very interesting to compare the model runs on Monday with the non-eclipse ready models!

We will start seeing model runs for the eclipse with this model on Sunday! Since a lot of us will be watching the weather closely, it will be worth watching closely for sky conditions at eclipse time. I will be in Nebraska and will post more about the great eclipse over the weekend.