Written by cuweathernerd    Tuesday, 03 January 2012 20:28   
Chasing 101: Section 1:2 Forecasting: Models

This is the next post in a series of discussions on how to chase. It is meant to be entry-level, to help those interested in chasing do so more safely and productively. It is also meant to be a discussion forum for more veteran chasers to share what they know.

While understanding what other meteorologists are saying is critical and almost certainly will be the foundation of your own chasing forecast, it is also imperative that you have at least rudimentary forecasting skills if you do not want your chasing to be reactionary (that is, waiting for a warning to jump into action).

Forecasting is an art and a science, and it takes a long time to truly learn -- it depends on intricacies -- down to the kind of crops, where lakes and hills are, winds left over from thunderstorms a thousand miles away the day before -- lots and lots of data. The good news is that we have ways to get lots and lots of data quickly.

Specifically, we use different instruments: weather satellites, radar, thousands of surface observations, and, most importantly, vertical soundings of the atmosphere made by weather balloons. All this data is diagnostic -- it tells us about the state of the atmosphere as it is right now.

Even better news is that we have computer models, sets of formulas and assumptions that use this diagnostic information to help make forecasts much more accurate. Models take all this data in, and they form a picture of the atmosphere at the current time, called the analysis frame. Once they have done that, they run many, many calculations to make a prognostic forecast -- to see how these initial conditions may behave at some point in the future.

These models are very, very complex, and are generally run on some of the most powerful computers in the world. Even so, these models normally take hours to run through all the calculations.

Each time a new set of initial conditions is used, the prognostic forecast will change -- so we call each iteration of the model a run. Sometimes, you will see the phrase prog used its place, or as a stand in for the word forecast. Normally, there will be 4 our more runs per day of any given model -- so new data is pretty much at hand.

As an amateur forecaster, these models will be the absolute cornerstone of your forecast workflow. That will limit, in some ways, your ability to make the best forecast, but in terms of time invested, it is by far the best option. Using experts alongside the models is an excellent way to make a great forecast for your own chase.

With that in mind, let's talk briefly about the three models you are likely to use or encounter, what they are used for, and where you can find their outputs.

The first is the Global Forecast System, or the GFS. This, as its name suggests, is a model which covers the globe. It also is a long-range model, forecasting up to 16 days in the future. This comes at the cost of resolution both temporally (time between each frame of the model) and spatially (the physical size of 'pixels').

Because the GFS is such a long term model, you will use it mainly to identify setups in the coming days. There are other long term models (ECMWF, GEM, etc), but their outputs are harder to find, so we will focus on the GFS for now.

In general, this is where the convective outlooks from the SPC will aid you -- and other discussion forums will help. You probably won't use this as much as the other models the day of the chase, though that is certainly not always the case.

Notably, the GFS helps to be a judgement tool of the performance of other models. By comparing the models, you can see how well each is doing relative to the others. Low variability between different models is a good sign the computers are doing well forecasting a situation.

The heart of most discussions will be the NAM, or North American Mesoscale Model. Like the GFS, it runs four times a day. It is actually a specific set of assumptions and formulas running through the WRF, the weather research and forecast model. Lots of different model set ups exist for the WRF, but the NAM is by far the most common you will encounter and use.

The NAM provides an 84 hour forecast -- 3 and a half days out. So it is a mid-range model. It also only is run for North America. This allows it to have higher spatial resolution, which is useful in identifying conditions that lead to mesoscale phenomena like severe weather. In general, this is where most of your forecasting will be done, and it will be the model most discussions are based around.

The NAM has its quirks and can make mistakes, so leaning on it exclusively is a dangerous possibility. This is where experts and discussion come into play.

The final model we are going to discuss is the High Resolution Rapid Refresh, or HRRR. As you might have guessed, this model has the highest resolution, temporally and spatially, of the three. It produces a run every hour, and forecasts for the next 16 hours in one hour increments (and for some parameters every 15 minutes!). It also has extremely high spatial resolution, on the order of 4km. The HRRR is in part another form of the WRF.

The HRRR is extra cool in that it takes in radar data as the heart of it's modeling, and resolves data at the scale of individual storms itself. This makes it very different from the NAM and the GFS.

It also outputs data that is very easy to use -- data like simulated radar maps and maps of the most likely areas for convective initiation, for instance.

All this makes it amazingly valuable in chasing, and it is the tool that, if you learn to use properly, that I think will increase your chase forecasts the most. As such, a whole post will be dedicated to it in the future.

A note about time in meteorology. Often, the features we study cover many time zones. This gets confusing fast, so we use Greenwich Mean Time (GMT)/Coordinated Universal Time (UTC)/Zulu time (Z). These are all the same time, and use a clock in England as the basis of the time.

It is the same time GMT at all points across the globe. So if it is 18Z, it is 18Z in Russia, the US, and Uganda, everywhere. This is very helpful, but it isn't the most intuitive. Complicating things more, it is always in 24 hour format, which you probably don't use daily. Further, things like daylight savings time add another twist.

To get your local time, you add or subtract a number of hours. Here in the US, we subtract. In central standard time, the time used across most of tornado alley, we subtract 6 hours or 5 hours if daylight savings time is in effect.

This is a little counter-intuitive, but it is imperative to using models well, since all the timestamps on models are in UTC. With a little practice, it will come naturally. If how I said it is confusing, a quick Google search should clear things up for you.

There are lots of other models, but for our purposes, these are the ones we will most likely use. The good news is that all are freely available online. The website I like to use for the GFS and NAM output is twisterdata.com. This is another one of the sites you want to bookmark.

There are several things I like about the site:

* Very clean, visually appealing outputs
* One click to produce a model sounding at any given location
* Set up by chasers for chasers
* Easy to compare models

Other people use the College of DuPage's site, Nexlab. You can access it here: http://weather.cod.edu/forecast/. It is also very good, but I learned using twisterdata, so I'm more comfortable there. Take a look at both sites, and figure out which you prefer!

For the HRRR, the website of the model itself is the best choice. http://ruc.noaa.gov/hrrr/. Go ahead and bookmark those too.


Mesoscale - Pertaining to atmospheric phenomena having horizontal scales ranging from a few to several hundred kilometers, including thunderstorms, squall lines, and fronts.

Resolution - The temporal and spatial scale of data. High resolution data tends to cover a limited domain as a result of computing limitations or physical costs.

Sounding - The data produced from (normally) a weather balloon launch that gives the vertical profile of wind, temperature, and moisture in the atmosphere. These are invaluable in evaluating stability, and will be covered in depth later.


The above is an article written by Reddit user cuweathernerd and has been modified by Ryan Lehms. The original article and discussion can be found here:


Last Updated ( Tuesday, 06 August 2013 18:54 )

Your are currently browsing this site with Internet Explorer 6 (IE6).

Your current web browser must be updated to version 7 of Internet Explorer (IE7) to take advantage of all of template's capabilities.

Why should I upgrade to Internet Explorer 7? Microsoft has redesigned Internet Explorer from the ground up, with better security, new capabilities, and a whole new interface. Many changes resulted from the feedback of millions of users who tested prerelease versions of the new browser. The most compelling reason to upgrade is the improved security. The Internet of today is not the Internet of five years ago. There are dangers that simply didn't exist back in 2001, when Internet Explorer 6 was released to the world. Internet Explorer 7 makes surfing the web fundamentally safer by offering greater protection against viruses, spyware, and other online risks.

Get free downloads for Internet Explorer 7, including recommended updates as they become available. To download Internet Explorer 7 in the language of your choice, please visit the Internet Explorer 7 worldwide page.