Climate Model Projections

As with the other resources provided through the GeoPlatform Resilience community, this page is primarily intended for audiences, such as data innovators, who want to use government data to develop tools to help others learn about the impacts of climate change or make decisions in which climate change plays a role. There are a number of important use considerations for the data provided here and therefore recommend reviewing the guidance below before making use of these data.

Climate simulation results can help inform decisions that involve climate change, but using these data is more involved than selecting and downloading data from a web portal. Before using scenarios or climate simulation results, it is important to make sure you have formulated specific questions you want to address, and to investigate whether you can do that using the results of existing analyses, rather than performing your own. The more specific the question(s) you can formulate, the easier it will be to decide whether you need to use and analyze climate model data sets, and if so which ones. This is important, because there is no “best” data set or “best” climate model; which ones you should use will depend on what question(s) you are trying to address, in what geographical region(s), etc.

If you decide that you do need to use and analyze climate simulation results, you should keep several points in mind as you proceed:

  • Each data set of climate simulation results generally contains only selected types of information, which limits the range of questions it can be applied to. For example, many data sets of downscaled climate projections include information about temperature and precipitation only; these cannot be used to address questions involving storm surge or extreme winds, for example. Also, climate data sets may contain only monthly-averaged quantities (e.g. temperature); these may not be useful for many purposes.
  • Climate simulations generally make no attempt to predict the timing of natural climate variability; for example, these data sets may contain useful information about conditions during El Nino years, but they can’t tell you when these years will occur.
  • Many users are interested in information about extreme weather of some sort (heat, precipitation, etc.) These can be difficult to simulate realistically; extreme precipitation is particularly tricky.
  • Simulations of future climate are all based upon assumptions about future greenhouse gas concentrations and other factors that influence climate; this is one reason why these simulations are referred to as “projections” rather than “predictions.”
  • In addition, there are numerous uncertainties in the climate models themselves, due to the challenge of numerically simulating all relevant aspects of the climate system over long timescales of decades to centuries. Therefore, even given the same assumptions about future greenhouse gases, different models will often produce different results, particularly at finer spatial scales and for extremes. (Note that some of these differences also result from random weather variations, and therefore do not represent true differences among model responses to greenhouse gas increases, but nevertheless can lead to different simulation results.) For this reason, it is considered good practice to use output from multiple models to explore a range of scientifically plausible futures – to account for an envelope of future climate risk, rather than a single future pathway.
  • Finally, simulations having finer spatial detail (i.e., “downscaled” climate model projections) do not necessarily have greater accuracy than coarser-resolution simulations; they add contextual detail related to factors such as regional topography and coastlines but may still retain the same basic climatic features simulated at larger scales.

What types of data are available? The links below provide access to a growing body of data, generated by climate models, relevant to understanding potential future climate change. This includes raw climate model output, as well as model output that has been processed by “bias correction” (removal of some known errors) and/or “downscaling” (addition of finer spatial detail). We refer to these types of information collectively as “climate simulation results.” These data have been produced using the leading climate research models, whose outputs have informed important scientific assessments of climate change and its impacts, such as Intergovernmental Panel on Climate Change (IPCC) reports and the National Climate Assessment. They have been collected into several archives and portals for increased ease of access to outputs from multiple models and types of simulations.


In addition, provides scenarios: quantitative and narrative descriptions of plausible future conditions that provide assumptions for analyses of potential impacts and responses to climate change. Scenarios are ways to help understand what future conditions might be, with each scenario an example of what might happen under different assumptions. Scenarios generally blend both model output and other information, such as observed trends. They are not predictions or forecasts, and no probabilities are associated with them. Instead, they provide a range of future conditions to bound uncertainty. The scenarios accessed through include climate change, sea level change, and land use and population change. They are based on peer-reviewed, published sources and were used in the development of the National Climate Assessment, which provides scientific findings about climate change and its impacts on U.S. regions and key socioeconomic sectors.

Updated on May 11, 2021