Illinois Water Supply Planning


    Home    Question 7    Further Resources
Climate Forecasting and ModelingBack
All of us are familiar with the daily weather forecasts produced by the National Weather Service, and by broadcast and private meteorologists. We often make or modify our plans based on these forecasts. Beyond a few days in advance, it is not possible to forecast weather conditions that will occur at a specific time and place. However, it is often possible to forecast general conditions, for instance, warmer and drier than normal.

For the past 10–20 years, atmospheric science researchers have expended considerable effort to develop techniques that can provide probabilistic outlooks for general conditions months to a year or more in advance. There is, of course, considerable variability in the specific conditions that actually occur in any given month, season, or year. These variations can have important impacts. Advance warning of such variations potentially can have substantial socioeconomic benefits.

Climate forecasts by reputable scientists do not specify exact weather conditions on any specific day. There are scientific reasons to believe that this is impossible more than about two weeks in advance. Rather, climate forecasts identify general conditions that will occur, whether the period will be warmer or cooler or wetter or drier than normal. A number of techniques have been developed in two broad categories: statistical and physical. Some techniques are purely statistical, but most physically based techniques also have statistical aspects.

Statistical techniques rely on historical climate data to establish relationships between different time periods. A simple example is an analysis that identified the 35 warmest winters from 1895–2001 temperature data for Illinois. Temperatures following those winters were above normal in 18 summers, near normal in 10 summers, but below normal in only 7 summers. A very simple statistical climate forecast can be developed for summer temperatures based on winter temperatures: if winter temperatures are above normal, the odds for a warm summer are increased.

“Optimal climate normals” are a simple statistical method based on long-term trends in the historical climate record. There is a tendency for anomalous conditions to be more frequent over 5- to 15-year periods. If such a regime is occurring, some predictive skills are required to project that such conditions will persist. For example, summers during the 1930s were usually hot and dry. If a forecast had been made during such a period, this technique would imply that the forecast should specify a higher probability of hot, dry conditions compared to cool, wet ones. Other statistical techniques have been developed to identify cycles in climate data and use such cycles as the basis for forecasts in the hope that these cycles are real and not just a statistical fluke.

Statistical techniques have several limitations. They do not incorporate information about the causes of variations. In many cases, there is no predictable relationship on which to base a forecast. For example, following the 35 warmest falls, 11 winters were drier than normal, 12 winters were wetter than normal, and 12 winters were near normal. Thus, warm fall temperatures provide no predictive information about precipitation the following winter, and statistical forecasts have limited usefulness. That being said, the causes of many variations also are not well understood, and physical techniques can provide no predictive information. Statistical techniques at least incorporate probabilistic information about how the atmosphere behaved in the past.

By contrast, physical forecast techniques rely on known causes of climate variations. A prominent example is the El Niño phenomenon, a periodic disruption of ocean and wind currents in the equatorial Pacific Ocean. The weather is affected not only in this region, but in other parts of the world, including the United States. El Niño events occur about once every 4–7 years and last for about one year.

Scientific research over the past 20 years has led to breakthroughs in understanding the El Niño phenomenon. As a result, it is now possible to anticipate by several months the beginning and evolution of El Niño events. Research also has increased knowledge about how these events affect the climate of Illinois. For example, during El Niños, particularly strong ones, winters often are characterized by warmer than normal conditions with little snow, and summers tend to be wetter and cooler than normal.

Considerable research is underway to identify other causes of climate variations. It is likely that physically based climate forecasts gradually will become more skillful. A primary focus of this research is on the relationship of atmospheric circulation patterns to the condition of the land and ocean surface. It is known that the distribution of sea surface temperatures affects atmospheric circulation patterns. El Niño is perhaps the most prominent example. However, anomalies of sea surface temperatures in other parts of the oceans also have effects. As yet, these anomalies have not led to the dramatic improvements in predictive skill obtained with El Niño events. But it is likely that some future improvements will result. The condition of the land surface, particularly the extent of snow cover and the amount of soil moisture, also affects climate. For example, there is evidence that deficient soil moisture in the Midwest during early summer often leads to dry, hot conditions later in summer because of decreased evaporation.

Since 1995, the National Weather Service has produced “Climate Outlooks” or climate forecasts up to a year in advance. These forecasts, updated monthly, are produced using three primary statistical and physically based techniques. These include:

  • El Niño/La Niña. Models have been developed to predict the evolution of El Niño and its opposite phase, the La Niña. They use statistically based relationships from past events to determine likely surface weather conditions in the United States.
  • Canonical Correlation Analysis. This statistical technique is based on the patterns of circulation and climate conditions that have occurred over the most recent four seasons. Future evolution of climate conditions is predicted based on similar past conditions in the historical record.
  • Optimal Climate Normals. This simple statistical method is based on long-term trends in the historical climate record.

WSP FAQ Home - Glossary - Search

Illinois State Water Survey

2204 Griffith Dr
Champaign, IL 61820-7463

Terms of use. Email the Web Administrator with questions or comments.

© 2017 University of Illinois Board of Trustees. All rights reserved.
For permissions information, contact the Illinois State Water Survey.
Site Map