Reading the maps · Probable Futures

Geographic coverage

One of the first things you might notice is the lack of data over the Arctic, Antarctica, and all oceans for most Probable Futures maps. CORDEX-CORE does not include output covering the Arctic, Antarctica, and some islands. Additionally, we chose not to display data over oceans to increase efficiency on the “back end” of the maps, and to improve the user experience. In rare cases, we have removed data in individual cells due to anomalies in model output or geographic coverage. In our drought maps, we have chosen to omit data over deserts, as they are dry year-round.

Every two-dimensional representation of our spherical Earth is going to suffer from distortions. There are maps that do a better job of minimizing distortions, such as the Equal Earth Projection. When we present static maps, we often employ Equal Earth. For the interactive maps we use the Mercator Projection. This projection makes the spherical earth a rectangle by forcing the curved longitudinal lines straight up and down. The resulting distortion portrays areas far away from the equator as much larger than they actually are. For example, Greenland and South America appear to be roughly the same size. In fact, South America is about 8 times larger than Greenland. We also have a globe view available in the interactive maps.

Baselines are important to have as a reference point for the past, and to understand the scale and rate of change as the planet warms. A standard baseline time period in climate science and international climate policy is 1850-1900. This range is often called “pre-industrial temperature”. The “warming scenarios” of 0.5, 1, 1.5, 2, 2.5, or 3°C represent global average temperatures relative to the pre-industrial average.
Probable Futures uses CORDEX-CORE data, which is available for 1970–2100. Since 1971–2000 had a global mean surface temperature of 0.5°C above pre-industrial temperatures, that time period is a good baseline for our maps.
Each warming scenario above 0.5°C represents a twenty-one-year time span. There are a couple of reasons to take data from several years on either side of a specific scenario. First, it provides a range of temperatures around each threshold. This range is valuable because even when the climate is “at” a certain average temperature, there will be variations around that average, e.g., warmer and colder years. Second, powerful phenomena that happen regularly but not annually, most notably El Niño and La Niña, tend to occur every 2 to 7 years and can last less than a year or multiple years. By including 10 years on either side of the threshold year, our results will include some El Niño and La Niña events from each of the underlying models.

The data behind Probable Futures maps are represented in 22-kilometer squares called “grid cells.” Most grid cells show three values that help summarize the distribution of model simulations within that cell. The mean value is averaged across the entire distribution. The median is the midpoint of the distribution, meaning half of the outcomes are either above or below this number. The median represents the outcome that we would most expect to see over time at that level of warming. The 5th and 95th percentile values can be thought of as infrequent outcomes that provide a sense of the range of possible outcomes. It is important to emphasize that the range of possible outcomes extends beyond those values. For example, 5% of the values in the simulated years have values lower (higher) than the 5th (95th) percentile value shown. Extreme outcomes that can have enormous impacts lie beyond the 5th and 95th percentile values.

The models calculate average conditions across each 22-square-kilometer area. For example, if one grid cell covers part of a mountain and an adjacent valley, or straddles both land and ocean, the value of that grid cell represents the average of those two different climate conditions.
Think of the specific values in each grid cell not as precise, but as indicating direction and magnitude of change. Natural climate variability will occur in any given year, and outcomes may be higher or lower than the average, including the possibility of extreme events beyond the ranges portrayed on the maps. Extreme events are critical when considering climate change, but climate models are better at simulating averages than extremes.

The key is located in the bottom left-hand corner of each map. It explains the relationship between the values in the data and the colors on the map.
The colors on Probable Futures’s maps are not colors you might expect to represent climate phenomena like heat or precipitation. This choice was intentional. First, we do not want to send subtle signals to the user about what color is “good” or “bad.” Each of us brings connotations and biases when looking at a map and associates different colors with different ideas and feelings. We chose colors that don’t easily correspond to common biases. Second, we chose colors that would work well for many people who have difficulty seeing colors. Third, we sought color combinations that would allow the viewer to see distinctions and change. We hope the color palette will encourage you to look closely and read the supporting resources to better understand the data.
Every map features an info button next to the drop-down box of maps. This button provides context and methodology for the specific dataset behind each map.