What are climate models?
A climate model is a collection of equations that represent the dynamical nature of the atmosphere, oceans, land surface, and cryosphere (ice), as well as variations such as seasonal changes, alterations in the composition of gasses in the atmosphere, and shifting ocean and atmosphere currents. Scientists create these models by compiling large amounts of information on what we know about the changing climate.
Climate models are run on historical periods of 60-200 years to verify that they simulate the Earth’s past climate to a reasonable degree. Models can then be run for future periods to project what might happen to temperature, precipitation, sea level, and other parameters due to the potential impacts from greenhouse gas emissions, land use change, and population increases. Since the rate of change in variables such as population increases, technological advances, and the development and implementation of non-carbon-producing energy sources are unknown, the models are run using a collection of future scenarios, ranging from a low to high degree of change in these variables.
Read more about climate models here. You can also explore some of the factors that are used in climate models with the World Resource Institute’s interactive Climate Data Explorer.
What do they do?
Climate models can help us to understand what could happen in the world 50, 100, or 200 years. For example, culverts have a typical life span of fifty years. Looking at future rainfall or expected sea level rise over the next 50 years can help to select the correct size culvert now. Organizations like NASA use global climate models combined with advanced imaging to create data on projected sea level rise, using it to predict which coastal areas might be inundated in the future. Climate scientists will often create maps from these model results to show trends in temperature around the world or for a specific area.
What is an emissions scenario?
An emissions scenario, or representative concentration pathway (RCP), is an estimate of the amount of change that might happen, generally measured by changes in the concentration of carbon or greenhouse gasses in the atmosphere. We cannot know exactly what humans will decide to do in the future—whether our carbon output will increase, decrease, or remain constant. There could be a large-scale shift to clean energy in the next ten years or emissions could continue to rise. For this reason, models are usually run for at least three emissions scenarios: high, medium, and low. The “low” scenario projects future climate even if we stopped adding carbon to the atmosphere immediately. Because some changes take a decades or longer to affect the climate, we are still feeling the effects of actions from 50 years ago. The “high” scenario reflects what might happen if we do not make substantial emissions reductions.
Which models to use
There is no such thing as the “best” model. Models are not designed to simulate accurate day-to-day change, but instead show a general trend of change. This is one reason it is recommended to average over two or more decades when analyzing future results. Using an ensemble of multiple models will provide more reliable results. It is also helpful to compare models to see the similarities and differences in their projections for the future.
In order to obtain quick results and a general idea of what might happen in the future, a middle-of-the-road model can be helpful. This is typically done only with a pilot project. While the one model approach will not give you the most accurate picture, it should be able to at least provide a trend (e.g., upward or downward, seasonal, gradual or abrupt change, increased variation, etc.). This approach is not complete enough for making predictions or recommendations, but rather shows if more research is merited.
You can learn more about model uncertainty by watching the ICNet webinars.
Downscaling is a method to convert the large-scale information that is created by coarse global climate models (GCMs) to higher resolution information for the region or location of interest. There are two types of downscaling: statistical and dynamical. Statistical downscaling finds statistical relationships between coarse climate model output and localized observations in order to produce a high resolution simulated output. This output is much closer to reality than GCMs, which cannot represent regional topographic features. Dynamical downscaling uses the same processes as GCMs, but at much finer scale. Dynamical downscaling models, called regional climate models (RCMs), are able to resolve physical processes of the general circulation of the atmosphere and surface processes, using the output from the GCM as the driving conditions in the finer scale RCM. Global climate models need to be downscaled in order to look at information on a local scale. Many models are available that are already downscaled.