- Get link
- X
- Other Apps
Written by: Brandon Molyneaux
Before proceeding, forecasts and decision-making regarding Irma should be redirected to the National Hurricane Center, as they provide the official forecasts regarding Irma. Be sure to also consult your local officials if you are in the path of Irma.
We are beginning to approach 100 years of the development of the first weather model developed by Lewis Fry Richardson (created in 1922). It was the first attempt at computer "predicting" the weather, but what Richardson didn't realize at the time was that his model had imperfections that caused a pressure change of 145mb in 6 hours.
Richardson extrapolated the instantaneous pressure change, assuming that it remains constant over a long period of time. The reasoning behind this was because his model didn't take into account gravity waves. For those who are wondering why this is significant, gravity waves help the atmosphere get into balance - without taking into account gravity waves, the model showed the pressure change mentioned above.
Over the past week or so, the GFS and Euro have been putting out solutions for Irma, but the reason why meteorologists keep saying "we don't know the solution more than 5 days out" is because there is no single model that can accurately forecast such phenomena accurately (key word: accurately). Sure, we can post model data regarding Irma that goes out 200+ hours out showing landfall in the Chesapeake, but we all see how that verified. Sure, we can post hour 360 hour GFS snowfall accumulation maps, but 99% of the time they won't verify.
You're trying to predict the future using variables we have now. There are many things models have to take into account, such as placement of troughs, as they're writing output for forecasters to digest. As the event gets closer, confidence increases with the impacts of the event. Likewise, the further out the event, the less confident the forecaster will be.
A way to increase this confidence is by using ensemble models, which, to simply put, are models that have slightly tweaked physics to give a different output. For example, the ensemble member #2 may have Irma going out to sea, while ensemble member #6 has Irma riding up the west side of Florida (again, consult NHC for the official forecast).
To simply put it, there is no weather model that can correctly and accurately forecast Irma. Forecasters take in all of the data from several different models and their respective ensemble members and try and understand the why and not take it for face value.
Before proceeding, forecasts and decision-making regarding Irma should be redirected to the National Hurricane Center, as they provide the official forecasts regarding Irma. Be sure to also consult your local officials if you are in the path of Irma.
We are beginning to approach 100 years of the development of the first weather model developed by Lewis Fry Richardson (created in 1922). It was the first attempt at computer "predicting" the weather, but what Richardson didn't realize at the time was that his model had imperfections that caused a pressure change of 145mb in 6 hours.
Richardson extrapolated the instantaneous pressure change, assuming that it remains constant over a long period of time. The reasoning behind this was because his model didn't take into account gravity waves. For those who are wondering why this is significant, gravity waves help the atmosphere get into balance - without taking into account gravity waves, the model showed the pressure change mentioned above.
Since then, we have come a long ways in numerical weather prediction. We now have the Global Forecasting System (GFS), EMCWF ("The Euro"), NAM (North American Model), CMC ("The Canadian"), and plenty more. However, each model has it's strengths and weaknesses. It's the job of the forecaster to go through and digest the model, understand the why behind it's output, and then translate that information and provide a forecast for the public and decision-makers.Impressive view of the Massive #storm over #SD from #GOES16 vis... w/Gravity Waves. Possible 3in. #HAIL. Imagery SRC: @CODMeteorology #SDWx pic.twitter.com/f2d0nekgLQ— Kyle Gentry (@ksgWXfan) August 21, 2017
Let's use the NAM as an example - it is a non-hydrostatic model. A non-hydrostatic model is generally used to resolve the "fine details" such as land-sea breeze circulations. (see: AMS definition of non-hydrostatic model)Look, it's cute, but this just isn't what you do with models. You ignore that 0z NAM run entirely, you don't forecast off it. Nothing. pic.twitter.com/jmAw9FlwlF— crankyweatherguy (@crankywxguy) September 8, 2017
Over the past week or so, the GFS and Euro have been putting out solutions for Irma, but the reason why meteorologists keep saying "we don't know the solution more than 5 days out" is because there is no single model that can accurately forecast such phenomena accurately (key word: accurately). Sure, we can post model data regarding Irma that goes out 200+ hours out showing landfall in the Chesapeake, but we all see how that verified. Sure, we can post hour 360 hour GFS snowfall accumulation maps, but 99% of the time they won't verify.
You're trying to predict the future using variables we have now. There are many things models have to take into account, such as placement of troughs, as they're writing output for forecasters to digest. As the event gets closer, confidence increases with the impacts of the event. Likewise, the further out the event, the less confident the forecaster will be.
A way to increase this confidence is by using ensemble models, which, to simply put, are models that have slightly tweaked physics to give a different output. For example, the ensemble member #2 may have Irma going out to sea, while ensemble member #6 has Irma riding up the west side of Florida (again, consult NHC for the official forecast).
To simply put it, there is no weather model that can correctly and accurately forecast Irma. Forecasters take in all of the data from several different models and their respective ensemble members and try and understand the why and not take it for face value.