Although it is what it is, we all look at the weather from different perspectives. Farmers look with hope at rain predictions and worry about extreme weather events like frosts. Urban planners design cities that are adapted to specific weather conditions, and their decisions have to hold for decades, shaping the day to day of thousands of people. Researchers are more interested in reproducing certain conditions and processes, to understand them better.
At Predictia, we deal with climate data to provide these insights and many more. However, there’s a shared feature across many of the services we provide: the Weather Research and Forecasting model (or WRF for short). A state-of-the-art weather model, that covers scales ranging from meters to thousands of kilometers. So in this post we want to talk a little bit more about this model, and the world of possibilities it opens up.
A versatile and robust model
The first thing to know about WRF is that it is numerical. That means it relies on mathematical models of the atmosphere and oceans to predict the weather, taking into account current weather conditions. It was developed in the late 90s, as a collaborative partnership among the National Center for Atmospheric Research (NCAR) and the National Oceanic and Atmospheric Administration (NOAA). The model is licensed in the public domain, which has brought together a big community around WRF that keeps constantly improving it. At last count, over 30,000 registered users in more than 150 countries.
The model outputs cover a wide range of scales: from meters to thousands of kilometers. This opens the door to reproduce very localised weather phenomena (such as Urban Heat islands), as well as larger scale phenomena (like hurricanes). This wide variety of possibilities is made possible by two dynamical solvers: the ARW (Advanced Research WRF) core and the NMM (Nonhydrostatic Mesoscale Model) core. In addition, WRF counts with different modules to cover different applications. WRF-Urban, for example, takes into account urban parameters like building heights or heat storage to reproduce urban environments. The WRF-Chem module couples meteorology and chemistry, to study the emission, transport, mixing, and chemical transformation of trace gases and aerosols. These are just two examples, as WRF is a complex model, full of modules that enable us to develop a wide arrange of applications.
Modelling past events for more resilient cities
In July 2019, Europe suffered one of the worst heat waves on record, setting all-time high temperature records in Belgium, Germany, Luxembourg, the Netherlands, and the United Kingdom. In France, near 1,500 deaths were attributed to the heatwave. Some metropolitan areas like Paris suffered the most, with cities getting hotter than their rural surroundings. It’s what’s known as the urban heat island effect. What would have changed if cities were built differently? Green rooftops, tree planting, green parking lots, a more efficient use of energy… Measures such as these are being implemented as mitigation strategies throughout cities. But climate change adaptation is not an “anything goes” situation. What constitutes an effective measure in a city can be a maladaptation on another. One way of assessing the fitness of these measures is creating digital twins of cities: digital models of the cities, where different urban parameters can be changed, and estimate the effects such changes have.
In this approach, the WRF model has a lot to offer. Remember the urban module we talked about earlier? The average building height, details on the terrain use, changes in traffic levels...All these details can be codified as WRF parameters. By reproducing the current urban landscape in WRF, we can validate the model's configuration, checking the outputs against past weather records. And once the model is verified, it’s time to toy around with it, translating potential changes in the urban planning to changes in the parameters of the model. These enables urban planners to see how their planned measures would play out in their specific context, before doing any real changes to the city.
High-resolution weather forecasting
Traffic jams due to heavy rain, rises in electricity consumption due to air conditioning demand, crops turned unviable by an unexpected frost. Weather affects our day-to-day lives, and having forecasts as accurate as possible benefit our whole societies. In addition, although the weather forecast we usually see in TV give just one forecast for each location, the weather conditions across the city vary greatly, specially across metropolitan areas of cities like Paris, Madrid or London. however, for many businesses need more detail. A good example is the service we offer to Madrid Metro, the rapid transit system for Madrid, Spain's capital.
WRF enables us to provide fine-grained weather forecasts, covering the different stations across their underground network. This provides the institution with some key information:
- Temperature: while keeping a stable temperature within underground stations is important for passenger comfort, heating and cooling this infrastructures takes up a lot of energy. By having detailed temperature forecasts across the network, Metro Madrid is able to develop an projection of the energy needed, helping reduce energy consumption,
- Precipitation: heavy rains represent one of the big disruptors of metro services, as water may flood stations, not only delaying the service or stopping it altogether, but causing serious damage to the infrastructure. Therefore, having detailed predictions on precipitation levels across the network informs the grid managers in advance, so they can put the water pumps to work.
- Humidity levels: when joined with drops in temperature, humidity levels in the air represent a hazard for rail-based transport, due to sheets of frost forming on the rails.
We join these forecasts with a warning module, so when a value overpasses validated thresholds, this service allows Metro Madrid to stay on top of what’s coming, to provide a better service.
These examples are just a gist of what's possible thanks to WRF. Other examples include location studies for renewable energies (solar & wind) or forecasting of sector-specific indexes (like apparent temperature for tourism, on-ground snow thickness for transport, or soil temperature for underground facilities). And this is just us. Researchers, companies and public institutions around the world use WRF for many more applications. All of this, powered by one model in the public domain, and the supportive community behind it. Isn't it amazing?