Weather models of the present and future
When we check the weather forecast, either on a smartphone app or on a site like weather.com, we usually see a list of high and low temperatures accompanied by little cloud, rain, or sun symbols, telling us what we can expect for each day’s weather. Have you ever wondered where these forecasts come from? No, they aren’t the predictions of a man in glasses and a lab coat lifting his licked finger up to feel the wind direction… These forecasts come from weather models!
Great. So, what is a weather model? In short, it’s a program run on a computer. Within this program, or model, we split the Earth into a bunch of boxes, like a checkerboard (see the figure below). Within each box, the model knows the average temperature, moisture, pressure, wind, etc. (we get this information from observations: weather stations, balloons, and satellites). From there, the model predicts the temperature, moisture, pressure, etc. within each box at a future time by using equations that tell us how these variables affect each other and evolve over time. That’s basically it—that’s how a weather model works.
As you might expect, the size of those boxes in a weather model is very important! The smaller those grid boxes are, the more details the model can “see.” In the example below, we see what terrain elevation looks like in two different models. The model on the left has much larger grid boxes, or coarser resolution, than the model on the right. The impacts of this difference in resolution are huge: the model of the left doesn’t have the Olympic or Cascade mountains here in Washington! The entirety of Western Washington is encompassed by only ~4 grid boxes, so the coarser model would never be able to simulate the complex, mountain-dependent weather we see here in the Pacific Northwest. The model on the right, however, features these mountain ranges in detail, and would thus be able to capture their weather impacts.
The average weather model used in forecasting today uses fairly large grid box sizes, somewhere between the two models shown above. This is a problem, and not just near mountains; cloud systems and thunderstorms can also be quite small in scale and are often misrepresented in models whose grids are too coarse. Storms like these are important atmospheric features not only because they affect local weather (through heavy rain, wind, and lightning), but because they have long-reaching impacts, especially when they occur in the tropics. Tropical storms, much like a rock thrown into a pond, have a “ripple” affect that can affect weather far away: yes, even here in the U.S. If current weather model grids are too coarse to capture these storms, then they can’t capture the storms’ impacts either, leading to forecast errors.
The solution to this—you guessed it—is to use higher-resolution models! This sounds quite simple, but there’s a catch: higher resolution, or more grid boxes, requires more computer power. Much more, in fact. However, with rapid advances in technology, we are finally able to run global models with very high resolution. My research aims to explore this “wild west of weather forecasting” by analyzing a global weather model that I have run using grid boxes that are less than 2 miles wide. That’s more than 65 million boxes covering the Earth! In the animation below, we see the incredible detail of the clouds produced in this model (That’s a model! Not a satellite image!). With this work, we hope to discover how the enhanced realism in this model can help improve weather prediction worldwide.
NIck Weber is a PhD candidate in the Department of Atmospheric Sciences. His research focus is long-term weather prediction, and how we might be able to improve forecasts with next-generation, high resolution global models.