Global climate models are one of the best tools we have for understanding how the climate may evolve in the future. They are incredibly complex pieces of software that combine the latest scientific thinking into how the atmosphere, the ocean, land surface, sea ice are all going to interact and evolve in response to changing greenhouse gas concentrations and the overall warming of the planet.
Now because such complex software for a global climate model to produce a simulation that goes out for several decades, some compromises need to be made. And one of these is that global climate models typically produce their output at a regional level of resolution. Think the northeast US or Western Europe global climate model simulations may only produce one number to describe the climate and trends for that entire region. And now we know in an area like the northeast US, the climate actually varies quite a bit. Depending on if you’re near the ocean or up in the mountains, you’re going to have very different temperature patterns, wind patterns, precipitation patterns. We know this, we can measure them. We have weather station observations, we have satellite observations that can measure variations in say heat and precipitation, sometimes down to the scale of hundreds of meters. We also have other dynamic models that we can run to simulate regional events and particularly extreme events. Things like tropical cyclones or the response of flooding to changes in precipitation.
At Jupiter, we combine all these tools to go from those regional trends down to local climate variability and local climate trends, leveraging those observations and those other dynamic simulations. This process of building that mapping from the global to the local scale is called climate downscaling and it’s an essential thing to do if you want to be able to use climate projection numbers to inform your own local risk. If you’re using raw global climate model output, you’re not going to be capturing local extremes and the numbers are probably not going to match what you would actually observe on the ground. So climate downscaling is essential for accurate local climate analysis.