Prioritize...
Most of this section is what I consider to be background knowledge. That is, it's not quite an 'Explore Further' section... rather the opposite. This section introduces the idea behind using a computer to solve the equations that govern how the atmosphere works. I am not going to assess you on the details of this material, but you should be familiar with it so that when we get to the meat of the lesson, you will understand what you are looking at.
Read...
Numerical weather prediction (NWP) is the term given to the process of predicting the future state of the atmosphere using a computer. Meteorologists call a list of instructions that produces a virtual weather forecast a computer model, or just a model for short. So, just how can a computer be made to forecast the weather? To answer that question, we have to talk briefly about some fairly complicated topics -- some fairly mathematical in nature. However, the following discussion is solely for your general understanding of how computers are used to predict the weather. Now, let's get down to business.
The first thing that you should realize is that a computer knows nothing about the "weather." That is, there is not a special "weather program" that can be loaded onto the computer so that it can analyze weather maps and make forecasts in a way that you or I do. Instead, the computer uses mathematical equations and the current state of the atmosphere to calculate what the atmosphere might look like at some future time (often called a "prog"). The exact nature of these equations is far beyond the scope of this course. Suffice to say, however, some of the equations are pretty straightforward and all of the inputs to the equations are easily measured, while others are far more complicated and must be approximated (the technical term is parameterized) by more simple functions or values.
The end result of all of this mathematical manipulation can be quite astounding. Take the animation below as an example. This numerical simulation is based on a set of equations that describe how simple water waves behave (called the shallow-water wave equations). In the simulation below, these equations are used to simulate water in a square, imaginary "bathtub." The waves that you see are generated by an occasional "drip" that splashes down on the surface and sends ripples bouncing off the sides of the tub. Looks pretty real, doesn't it? This animation demonstrates very clearly that for some processes, simple mathematical equations (yes, I know "simple" is relative) can produce astoundingly realistic results. Similar wave models can even be used to predict the propagation of a tsunami across the Pacific Ocean. Check out this animation showing the tsunami that devastated Japan on March 11, 2011.
Understanding Model Terminology
As I have stated before, your first step when looking at data is to figure out the when, what, and where. This is especially key with model data because you have to not only know what you are looking at but when. It is important to understand that for a computer prediction of the atmosphere, the data represent what the computer predicts for the atmosphere at a single, specific time, called the valid time. So if a prog is "valid" at 12Z on April 23, 2009, then the features that you observe on the prog are predicted to be in those exact positions at exactly 12Z (on April 23, 2009). The second vital piece of time information that you need to glean from the forecast prog is the initial time. The initial time is the time that the computer model was started (or more precisely, the time that the data used to start the model was collected). Most of the major forecast models are run at 4 times during the day -- 00Z, 06Z, 12Z, and 18Z. Therefore, when we refer to the "00Z model run", we are referencing the computer model that was started using atmospheric data collected and fed into the simulation at 00Z.
Finally, the difference between the initial time and valid time is called the forecast lead time. The forecast lead time is basically how far in the future the model is trying to predict. The larger the forecast lead time, the less accurate the model tends to be. So, you will often hear meteorologists refer to model data in the following format: "the 48-hour prog of the 12Z April 23, 2009, model run...". This means that this prog was valid at 12Z on April 25, 2009 (48 hours after the initial time). You also might hear, "the 36-hour prog valid 00Z April 25, 2009...". This means that if the 36-hour prog was valid at 00Z on April 25, then the model must have been started (its initial time) at 12Z April 23 (36 hours before).
Model "Analysis" vs. Model "Forecast"
You will notice that some websites provide a model output that begins with a "zero-hour forecast". These plots show the state of the atmosphere that was used to initialize the model run. The correct terms for this data are model analysis or initialization. All other outputs from the model (describing some future state of the atmosphere) are model forecasts. To avoid strange looks from other meteorologists, don't interchange these two terms (analysis and forecast). It is not correct to say "tomorrow's model analysis shows...." Instead, refer to all model outputs as forecasts, except the initialization data. Furthermore, regardless of the data source, know that the term analysis is used to describe only current or past data, while the term forecast is used to describe only future atmospheric conditions.
In the next section, we will look at some of the common types of numerical weather models, each of which has its advantages and disadvantages. If you are going to use data from weather models, you need to be aware of these differences.
Read on.