CLIMATE TRENDS IN ILLINOIS


From The Changing Illinois Environment: Critical Trends,Volume 2: Air Resources, Technical Report of the Critical Trends Assessment Project

INTRODUCTION

This report presents an analysis of climate trends since the late nineteenth century in Illinois. This work was pursued under the Critical Trends Assessment Project (CTAP) of the Illinois Department of Energy and Natural Resources. Weather and climate events have significant impacts on many socioeconomic and environmental aspects of Illinois. Therefore, an analysis of climate trends provides the necessary background for understanding trends in other aspects of the Illinois environment.

Significant climate variability can occur at virtually all time scales. Therefore, a variety of climate data and information from Illinois sites are assembled here in order to investigate the temporal variability of various climate parameters. In addition to studying statewide changes in annual temperature, precipitation, and other climate parameters since the turn of the century, changes by month and by season and in the frequency of extreme events will be investigated over the period of record.

CAUSES OF CLIMATE

Climate is a description of the overall character of daily weather conditions that is manifested from season to season and from year to year. It is the sum of all statistical weather information that describes the average and variability of weather at a given place. Climate does not include information on the weather of a particular day.

We investigated changes in the frequency and magnitude of several climatic parameters. Some of these yield a clear indication of a trend. For example, all stations indicate a common temperature trend over the past 100 or more years, with associated changes in related climatic parameters such as degree-day totals, cloud cover, frequency of frozen precipitation, etc. It is not yet possible to definitively identify the specific causes of these trends. However, it is useful to discuss in a general way some of the possible causes of climate variability to provide a perspective on the analyses presented in this report.

Ultimately, the climate of the earth is driven by energy received from the sun. Although some of that energy is reflected back to space, most of it remains in the atmosphere to provide the fuel for global circulation patterns. The single most important variable affecting climate at a particular point is latitude because this affects the total amount and seasonal distribution of solar radiation reaching the surface at that point. The mid- latitude location of Illinois results in a substantial variation in the amount of solar radiation received between winter and summer. Thus, the seasonal variation in climate is relatively large.

It is known that variations in the orbit of the earth around the sun result in cyclic variations in solar radiation on time scales of greater than 10,000 years. Many scientists believe this to be a major factor in the periodic occurrences and retreat of ice ages. However, on shorter time scales (100 years or less), it is not known with certainty whether significant variations in solar radiation occur. Measurements of solar radiation are not accurate enough nor of long enough duration to identify significant fluctuations.

Another major factor affecting climate is the distribution of continents and oceans. It is known that temperature differences between the ocean surface and the land surface are a major factor in establishing certain circulation patterns in the atmosphere. For instance, during the summertime, the Atlantic Ocean is cooler than the North American continent. This results in a pressure difference (high pressure over the ocean, low pressure over the continent) that forces a southerly flow of air over eastern North America, transporting significant moisture from the Gulf of Mexico into Illinois. This gives us our humid summertime climate with generally abundant rainfall.

It is known that fluctuations in ocean surface temperatures can occur and that these can cause fluctuations in climate that are of significance. The El Ni¤o phenomenon is one example, where a warm pool of water develops aperiodically over the eastern equatorial Pacific. This warm pool of water and the associated disruptions in equatorial wind patterns are known to have effects on climate in many other parts of the globe. It is possible, even probable, that fluctuations in ocean surface temperatures in the Pacific and the Atlantic have effects on climates thousands of miles away and may be responsible for some significant fluctuations that have occurred in the past. However, it is very difficult to identify such ocean temperature changes as the clear cause of any particular climate fluctuation experienced in Illinois.

The radiation balance and energy transfer within the atmosphere are influenced by the composition of the atmosphere. Some constituents of the atmosphere, notably carbon dioxide, methane, chlorofluorocarbons, and water vapor, limit the loss of infrared radiation by the earth to space. These have a net warming effect on surface temperatures. It is well known that the concentration of many of these gases are increasing because of human activities, such as the burning of fossil fuels. If all other causes of climate were to remain constant, we would expect that the temperature at the surface of the earth would increase as the concentration of these gases increased. However, it is still too early to ascribe observed changes in Illinois climate solely to this cause.

Other factors may also play a role in modulating Illinois' climate. For instance, volcanic eruptions can inject large amounts of dust into the earth's stratosphere. This can decrease the amount of solar radiation reaching the earth's surface for a year or more in some cases. It is likely that past volcanic eruptions have caused significant changes in the earth's climate. For instance, the recent interruption of Mt. Pinatubo is believed to have resulted in a temporary decrease in global surface temperature of 1oF. In large urban areas such as Chicago, the change from a natural or agricultural land surface to large percentages of paved areas and buildings can cause a local change in climate. Urban areas tend to be warmer than nearby rural areas.

Since there are multiple factors that can affect Illinois' climate, it is difficult, if not impossible, to cite a specific cause for any particular feature or trend in Illinois' climate.

DATA

Data from two major categories of climate stations are analyzed in this report. The first category includes the National Weather Service (NWS) First Order Sites (FOS). They provide data from at least hourly measurements taken at major airports. The data are derived from continuous observation of cloud cover, visibility, pressure, temperature, dew-point temperature, winds, hydrometeors, and other obstructions to visibility. This very detailed set of observations is archived in digital form for only five sites in Illinois.

The second major category of observing sites are those of the NWS Cooperative Observer Network (CON). This network is composed primarily of individual and institutional volunteers who obtain measurements on a daily basis for the NWS. The NWS provides the measurement equipment, and for the most part, the observers are not paid. The observations include daily maximum and minimum temperatures, daily precipitation, daily snowfall and snow depth, and in a few cases soil temperature and pan evaporation. Although these observations are not as detailed as those for the FOS network (they provide daily rather than hourly observations), the network is much denser and therefore offers a much-used, quality dataset. For this report, 41 CON stations have been selected, each with long-term data extending back to the turn of the century.

For purposes of climate applications, the NWS has divided the state into nine regions, known as "climate divisions." They are shown in figure 1. The National Climatic Data Center has developed a special dataset that consists of temperature and precipitation values averaged for all stations, as available in each of these climate divisions. These values are available monthly from 1895 through the present. Because this is a relatively compact dataset and incorporates information for all available stations, these data were used to assess overall trends in precipitation and temperature.

Data from individual stations were used to assess trends in more detailed aspects of temperature and precipitation that are not available by climate divisions. Although climate division values are available for monthly average temperature and monthly total precipitation, no such data are available on a daily time scale. Therefore, trends in the following climatic elements were assessed from the data from individual stations:

Temperature
* Monthly maximum and minimum temperatures.

* Number of days with daily mean temperature above 50oF. This threshold is examined because the rate of growth and development of corn and soybeans becomes significant when the daily mean temperature is above 50oF.
* Extreme daily maximum temperature. This is the highest single daily value occurring in a month.
* Extreme minimum daily temperature. This is the lowest single daily value occurring in a month.
* Number of days with daily minimum temperature above 70oF. This is a somewhat arbitrary threshold that is meant to represent summer nights of abnormal warmth.
* Number of days with daily minimum temperature below 0oF. This threshold is examined as a representation of particularly "cold" nights.
* Number of days with daily minimum temperature below 32oF. This threshold is examined because the freezing point affects many processes.
* Number of days with daily maximum temperature above 86oF. This threshold is examined because the rate of corn and soybean growth and development typically reach a plateau near this temperature threshold. That is, above this threshold, the rate of crop growth and development no longer increases with rising temperature.
* Number of days with daily maximum temperature above 90oF. This is a somewhat arbitrary threshold, representing particularly warm days of discomfort to many people.
* Number of days with daily maximum temperature above 100oF. This threshold is examined as a representation of the frequency of days of great discomfort to many people.

Precipitation

* Number of days with precipitation.
* Number of days with precipitation of 1 inch or more. This threshold is examined as a somewhat arbitrary representation of rather heavy rain days.
* Total monthly snowfall.
* Number of days with snowfall.

METHODOLOGY - STATISTICAL TECHNIQUES

Trends in Means

In order to detect potential trends in climate, several variables were subjected to a mathematical process to remove the seasonal cycle in the data. This process was performed on climate division values of monthly temperature and monthly precipitation and on station values of monthly minimum and maximum temperatures. Specifically, the data were "standardized" with the following formula:

xij = (xij - xj)/sj

where xij is the value for year i and month j; xj is the average over all years of the values in month j; and sj is the corresponding standard deviation for those values. These converted values are referred to as "standardized" temperature and precipitation. They express the values in terms of the number of standard deviations above or below the mean.

The climate division time series consisted of 1,164 data values for each element (97 years x 12 months). Over-all trends were assessed by subjecting the time series to a linear regression; the statistical significance of the resulting trends was tested using a t-test for significance of slope of the regression (Draper and Smith, 1981). Trends in individual seasons were also assessed. The meteorological definition of seasons was used:

* Winter (December 1 through February 28/29)
* Spring (March 1 through May 31)
* Summer (June 1 through August 31)
* Fall (September 1 through November 30)

Seasonal values were obtained by averaging temperature over the three months and totaling precipitation over the three months. Thus, the resulting seasonal time series each consisted of 97 data points. For the time series of individual stations, only 91 years of data were available.

For variables involving the number of exceedances in temperature and precipitation and for monthly snowfall totals, the monthly values were summed for each year. The resulting data (91 data points) were subjected to the regression analysis for trend indicated above. The extreme monthly maximum temperatures were analyzed by taking the highest value in each year and subjecting each of the 91 values to the regression anal-ysis for trend. The lowest extreme monthly minimum temperature value for each year was also assessed for trend in this manner. In all of the above regressions, the trends were tested for significance at the 10 percent level using a two-sided t-test.

Variance Assessments In addition to testing the time series for trends in the mean, the data were also analyzed for changes in the variability of temperature and precipitation over the period of record. A subjective analysis of the time series suggested that certain periods of time, a decade or more in length, were characterized by significantly less or more variability than the rest of the time series. To assess this, the climate division values of temperature and precipitation were broken into six 15-year segments, beginning in 1901. For each 15-year subset of data, the standard deviation of the time series was calculated. The resulting data were plotted and subjectively analyzed.

Temperature and Precipitation Index An additional objective of this study was to identify certain periods of time characterized by specific climate conditionsin particular, multiyear periods characterized as:

* Warm and wet
* Warm and dry
* Cool and wet
* Cool and dry

To assess these conditions, two combined temperature/precipitation indices were defined and calculated as follows:
* Index 1 = standardized temperature + standardized precipitation
* Index 2 = standardized temperature - standardized precipitation The association between these indices and the above scenarios is:

* Warm and wet (large positive values of Index 1)
* Warm and dry (large positive values of Index 2)
* Cold and wet (large negative values of Index 2)
* Cold and dry (large negative values of Index 1)

Modifications to this rather straightforward categorization scheme were required, as will be explained as part of the section Anomalous Episodes from the Illinois Climate Record. Time series of the index were plotted and examined to identify multiyear periods with the above climate conditions.

TRENDS IN PRECIPITATION Trends in Average Precipitation

Mean annual statewide precipitation, averaged for all available stations since 1840 is shown in figure 2. The data are considered to well represent statewide conditions since about 1890, by which time the annual values were derived from at least 113 stations in the state. The "statewide" totals during the 1880s were based on 18 stations; on six or seven stations from 1852 to 1880; and on only one station (Athens, Sangamon County) from 1840 to 1851. Because of the paucity of data prior to about 1890, some care must be exercised in its interpretation. If representative, they indicate a slight decline in annual precipitation from the mid-1800s to about the turn of the century, with an increase thereafter. However, as shown, the year-to-year variation is the dominant characteristic of statewide precipitation, as with an individual station's record. The statewide mean annual precipitation record based on 150 to 200 sites for each year (figure 3) has varied widely from 25 to 50 inches. (Figure 3 duplicates the more recent portion of figure 2. This time frame facilitates comparison with other figures in this report, most of which cover the period from 1901 to 1991.) Occasionally, consecutive years exhibit similar anomalies, but in the vast majority of cases, one year's precipitation is unrelated to that of the next. The statewide average is about 37 inches, with about 5 inches less in northern counties and about 5 inches more in the south.

No obvious discontinuity in statewide annual precipitation has occurred since 1901. Changnon (1984), in an analysis of data through 1980, also noted that year-to-year variability was the dominant characteristic of the time series. He noted a slight tendency for lower values during the period 1910 to 1940 compared to values both before and after that 30-year period. A close examination of figure 3 does suggest this. In addition, an analysis of trends of individual stations (reported later) also suggests a general upward trend in precipitation during this century. However, this trend is weak compared to the short-term variability. The number of days per year with precipitation (averaged for all long-term stations) exhibits a somewhat stronger upward trend since 1901 (figure 4) than does annual precipitation. The number of days per year with precipitation appears to have increased by about 10 during the 90 years of record, although variability appears to increase as well. The annual number of days per year with precipitation trended upward with time at nearly all the 41 long-term stations; only Carbondale (Jackson County) and Walnut (Bureau County) digressed. Thus, the trend has been rather ubiquitous throughout the state. However, several of these upward trends were rather weak and not statistically significant. The number of days per year with at least 1 inch of precipitation (with a present recurrence frequency of about once every other month in Illinois) is shown in figure 5. As with the above two precipitation measures, identification of trends is difficult. Nonetheless, a slight decline appears during the first few decades with an increase during recent decades. The days per year with a least 1 inch of precipitation at individual stations in the state all exhibit upward trends, except for White Hall (in Greene County). Changnon (1984) found a general increase in the number of days with rains of 2 inches or more during the warm season. In addition, Huff and Angel (1989) found that the frequency of very heavy rainstorms had increased from the early part of this century to the latter part.

To summarize temporal trends in precipitation over the state, magnitude (figure 2) appears to decline from 1840 to about the 1930s, increasing thereafter. The frequency of precipitation events appears to have increased slightly in the 90 years since 1901, paralleling the trend in annual precipitation. Much more significant is the substantial year- to-year variation. Trends significant at the 10 percent level, were identified for 9 to 21 of the 41 sites for the five precipitation parameters. Trends for individual stations are listed in appendix table A.1, and a summary is given in table 1. Interestingly, nearly all these trends were positive, except for the number of days with measurable snowfall. Sites in northern and east-central Illinois were somewhat more likely to have experienced significant trends than sites in other parts of the state.

Trends in Snowfall

Annual snowfall for the state since 1901 (figure 6) shows some rather clear characteristics: 1. Decline from 1901 to the 1930s. 2. Increase from the 1930s to the 1980s. 3. Decline in the last decade. 4. Continuity from year-to-year over episodes of three to six years. For example, note similar snowfall amounts from 1903-1906, 1919-1923, 1939-1945, 1952-1959, 1961-1965, and 1987-1991.

Because the year-to-year variation is so large, five-year running averages of statewide annual snowfall were calculated (figure 7). This figure exhibits a slight increase with time, but much more apparent is the fluctuating nature, as with annual precipitation. Fluctuations in annual snowfall exhibit secular episodes of about 13 years, varying from 7 to 17 years since 1901.

In general, years with greater than average snowfall also experienced a greater than average number of days with snowfall (figure 8). A decline in the number of days with snowfall is apparent from the turn of the century to the early 1930s (coincident with the warming shown in figure 11), with a possible increase thereafter. More prominent than trends are the several years with relatively high frequencies of days with snowfall, notably, 1914, 1924, 1951, 1960, 1977, 1978, and 1980.Of 16 Illinois locations with statistically significant trends in annual snowfall (table A.1, p.72), five (four in east-central Illinois) exhibited declines, whereas 11 indicated increases in annual snowfall from 1901 to 1991 (table 2). Such differences in trend are not likely within such a relatively small area, emphasizing a problem with snowfall observations, i.e., they are extremely sensitive to the exposure of the site, as well as the method used to determine snowfall (precipitation in a gage, average of several depth measurements, or density of snow cores). Trends in Variability

Trends in year-to-year variability in precipitation were examined by separating the precipitation data into 15-year segments, beginning with the period 1901-1915. For each 15- year segment, the interannual standard deviation of precipitation was calculated for each season (figure 9).

For the summer, a notable feature is the low variability during the period 1961-1975, as opposed to much higher variability from 1901 to 1945 and 1976 to 1990. Figure 10 shows summer precipitation for this century. The low variability during 1961-1975 is the result of an absence of severe summer droughts during the 1960s and 1970s. More frequent droughts during the period 1976-1990 have made variability similar to that during the earlier part of the century.

A period of relatively low variability also occurred during the fall in the 1940s, 1950s, and 1960s. By contrast, the period 1916-1945 was characterized by the highest variability. For the winter season, the first half of the century exhibited very low variability compared to the second half. The greatest variability was exhibited from 1946 to 1960, virtually opposite to the trends for summer. Finally, during the spring, high variability was experienced between 1916 and 1945, with low values before and after, essentially the same trends as those of fall.

TRENDS IN TEMPERATURE Trends in Average Temperature

Changnon (1984) presented an analysis of trends in statewide average annual temperatures for Illinois for the period 1940 to 1980. His analysis is extended here to 1991 and presented in figure 11. The number of observing sites prior to 1890 is limited, as they were for statewide precipitation. Temperatures in Illinois follow a similar trend to those of the nation, i.e., rather stable temperatures during the first few decades (with a state-wide mean of about 50oF). A warming trend follows until about 1930 (to about 53oF), after which cooling reduces the mean to about 52oF by 1980, with warming suggested thereafter. Cool temperatures during the first few decades reflect an expression of what has been called the "Little Ice Age" (Lamb, 1966). It is important to recognize that temperatures cooled substantially from the 1930s to 1980 or after (a similar trend for the Northern Hemisphere), five decades during which carbon dioxide was increasing!

Mean annual Illinois maximum and minimum temperatures since 1901 (figures 12 and 13, respectively) both exhibit similar trends, i.e., warming of 2 to 3oF from 1901 to the 1930s, followed by an equivalent cooling through about 1980, and the suggestion of warming again during the last few years. These trends are not surprising, since they are found in temperature records from the entire United States, the Northern Hemisphere, and the world, although the magnitudes of change decrease with ever-increasing areas of interest (e.g., Wigley and Barnett, 1990).

The average maxima and minima exhibit extremes during the same years, i.e., those much warmer or colder than the long- term average, e.g., 1917, 1921, 1924, 1931, 1978, 1979, etc., throughout the period of record. There is no evidence of periodicities.The interseasonal fluctuations of average statewide summer maxima (figure 14) and those of average statewide winter maxima (figure 15), are quite different from each other. This is not unexpected, since temperature anomalies seldom continue longer than just a few months, and, meteorologically, one would not expect similar temperature characteristics to prevail from one season to another. Similar comments apply to statewide winter and summer minima (not shown), i.e., the interseasonal fluctuations need not parallel each other, and they do not.The mean number of days (averaged over all stations) with temperatures above 100oF (figure 16) are indeed few, particularly during the most recent decades, which are exhibiting cooling. Several years, namely 1901, 1913, 1930, 1934, and particularly 1936, ex-hibited high frequencies of such days, all during the first four decades of the record. Years with a high frequency of days with temperature greater than 100oF tended to occur in episodes composed of a few sequential years, e.g., 1913-1914, 1930-1936, 1939-1941, and 1952-1954, all of which represented major droughts!

The statewide average number of days per year with maximum temperatures above 86oF (figure 17) exhibits trends similar to those shown in figure 16, although the absolute frequencies are greater. The 1930s experienced more days with higher temperatures than any other decade on record.Figures 18 and 19, showing the number of days per year with maximum temperatures below 32o and below 0oF, respectively, exhibit inverse trends to those found in figures 16 and 17. More frequent cooler maxima have occurred since about 1960 than before. This trend suggests that the cooler 1960s, 1970s, and early 1980s comprised more frequent cool days, as opposed to cooler extremes only. Note the relatively high frequency of cold maxima in 1912, 1924, and 1936 in both figures 18 and 19. The relatively high frequency of maxima below 32oF in the late 1960s, 1970s, and early 1980s is also seen in the frequency of maxima below 0oF (figure 19).

Figure 20 presents the statewide average number of days per year when the minimum temperature exceeded 70oF. Not surprisingly, the trend follows that of average temperature and maximum and minimum temperatures, with increasing frequency from 1901 until the 1930s, a decline thereafter, with a possible increase again during the last few years. It should be noted that the high frequencies of 1913, 1931, 1934, and 1936 have not been equaled in Illinois since.

Figure 21 shows the statewide average number of days per year when the minimum temperature was below 32oF. A declining trend may be suggested from 1901 to perhaps the mid-1940s, with either stable frequencies or perhaps a slight increase thereafter.

Figure 22 presents the statewide average number of days per year when the minimum temperature was below 0oF. Such frequencies were decidedly minimal during the 1940s and 1950s, which were warm decades. The highest frequency of such days occurred in 1936, 1963, 1977-1979, and 1985.

Change in Parameters Related to Temperature

Mean Statewide Annual Heating Degree-Days. Heating degree days (HDDs) are derived from the accumulated positive differences of 65oF minus the mean daily temperature. This rather simple calculation results in a value that closely parallels heating fuel needs.Average annual HDDs, shown in figure 23, demonstrate a crude inverse correlation to annual temperature. As one would expect, the fewest HDDs are found in the 1930s and 1940s. HDD totals during the warmest decades are about 25 percent fewer than during the colder years. Five-year running averages of this parameter (presented in deviations from the period average in figure 24) rather clearly demonstrate fewer HDDs in Illinois from the 1920s through the 1950s, the warmest decades of the century-long Illinois temperature record.

Mean Statewide Annual Cooling Degree-Days. Cooling degree-days (CDDs) represent a measure of electricity required for cooling buildings during the warm season. They represent accumulated positive differences of daily mean temperature minus 65oF.Annual CDDs (figure 25) also show the 1930s and adjacent decades as warmer than those earlier or later, as did the HDD record. The total CDD five-year running averages (figure 26) show more CDDs during the 1930s and early 1940s, but they also show peaks during the first few years of record, in the early teens and twenties, in the 1930s and early 1940s particularly, and the mid-1950s.The five-year running averages of heating and cooling degree days (figures 24 and 26, respectively) demonstrate a rather clear inverse relationship to each other, except for the late 1970s and 1980s.

Mean Statewide Corn Growing Degree-Days. Corn growing degree-days (GDDs) are derived from the accumulated positive differences of daily mean temperature minus 50oF (when maxima exceed 86oF, the maximum is set to 86oF; when minima are less than 50oF, the minimum is set to 50oF). Trends of temperature, mentioned earlier, are also apparent in this measure (figure 27) of growth. Annual statewide corn GDDs since 1878 have been lower than those of 1992 in only six other years, 1882, 1883, 1915, 1917, 1924, and 1967, and then only marginally! The relatively few GDDs in 1992 virtually devastated the northernmost 100 miles of the Corn Belt. As a result, corn in that area was of lower quality and did not fully mature everywhere within that region. Interestingly, 1992 witnessed a record total corn crop in the United States by a substantial margin, largely because of reduced heat and moisture stress in the heart of the Corn Belt.

Trends in the Growing Season. Trends in the length of the growing season (figure 28), the date of the last spring freeze (figure 29), and the first fall freeze (figure 30) correspond only weakly to the trends indicated in average annual temperature. This is due to the fact that averages reflect the overall anomaly of the period included in the average. Individual events, e.g., the date of first or last frost, are often poorly correlated with monthly or seasonal temperature averages, which are composed of many days surrounding the frost event, because the frost event is short-lived (only hours in duration). Moreover, the event is also dependent upon other specific local conditions occurring during the time of the frost, e.g., low cloud cover, wind speeds, humidity, etc. The dominant characteristic of the growing season time series is the year- to-year variability. Changnon (1984) had noted a slight increase in the length of the growing season caused primarily by a trend towards earlier spring freezes. The results of that study, based on data from nine stations, are also seen in this analysis of 41 stations. Table 3 (p.40) summarizes a statistical analysis of the trends from these 41 stations for two freeze thresholds (32oF and 28oF). The most obvious feature of these results is that all stations with significant trends in the date of the last spring freeze exhibit a tendency for earlier spring freezes. There is also a tendency towards later first fall freezes. The combination of these two trends leads to a tendency toward longer growing seasons.

Linear Trends in Temperature Records at Illinois Sites Linear trends in the temperature parameters are listed for all 41 long-term stations in appendix A. Average temperature parameters, minimum temperature parameters, and maximum temperature parameters are listed in tables A.2-A.4, respectively, beginning on p.74). These trends are summarized in table 4. For most temperature parameters, statistically significant trends (at the 10 percent level) are found for less than half of the stations. However, in many cases, the direction of the trend is similar for many stations. For instance, 27 stations exhibited significant downward trends for the highest temperature of the year. This reflects the high summer temperatures experienced during the 1930s and the relative absence of extreme high temperatures during the last three to four decades. This same feature is also evident in half or more of the sites in the number of days with maximum temperatures above 86oF, 90oF, and 100oF. This feature is in general agreement with the general downward trend in average temperature from the 1930s into the 1970s (figure 11). Although the average temperature record and the number of "hot" days in the last decade suggest a reversal of this trend, this change has not been of long enough duration or magnitude to overcome the large downward trend in the earlier period. Somewhat surprisingly, the number of days of extreme cold, identified as the number of days with maximum temperatures below 0oF (figure 19), has exhibited an upward trend at many stations. This was caused by a significant number of extremely cold arctic air outbreaks during the last 15 years. These outbreaks have even occurred with some frequency during the 1980s, a period of generally very warm winters. This also reflects the strong influence of the very cold winters in Illinois during an extended period in the late 1970s and early 1980s.

Trends in selected parameters (seasonal average, minimum, and maximum temperature and seasonal precipitation) were also examined by individual seasons. The results of this analysis are given in tables A.5-A.8 (beginning on p.82) for the autumn, winter, spring, and summer seasons, respectively, and summarized in table 5. Very few stations show statistically significant seasonal trends in total precipitation. However, a number of stations show downward trends in temperature for autumn. Also, seven to eight stations show downward trends in winter temperature. Trends in spring are weakly positive.

Trends in Variability

An analysis of the trends in temperature variability are shown in figure 31 by season. For the summer season, the period 1961-1975 was characterized by the lowest variability. Figure 32 shows summer temperatures for this century. The low variability during 1961-1975 was the result of the absence of severe summer heat waves. By contrast, variability was much higher from 1976 through 1990, caused by more frequent summer heat waves. This latter 15-year period was quite similar to the first half of the century. For the fall season, low variability was experienced during 1901-1915 and 1945-1990, while higher variability characterized the years 1931-1945. For winter, 1976 through 1990 exhibited the greatest variability. This was the result of the very cold winters during the late 1970s, contrasted with the mild winters of the middle and late 1980s. The spring season was also characterized by high variability during the period 1976-1990, as was winter. Interestingly, trends of seasonal temperature variability in summer and fall parallel those of precipitation (figure 9), whereas those of winter and spring do not appear to be related. TRENDS IN CLOUD COVERChangnon (1984) presented the number of days per five years with cloudy skies (any cloud type or combination covering 70 percent or more of the sky) at Springfield, St. Louis, and Evansville, IN (figure 33) and Peoria, Moline, and Chicago (figure 34), from 1901 through 1980. All six sites showed a rather steady increase in cloudy days for the period of record, from about 90 days per year for the first few decades of the century, to about 160 days per year after about 1940 and continuing to 1980. Additional data from 1981 through 1990 maintain the higher frequencies (figure 33), except those of Evansville, IN (figure 34). Chang-non (1984) questioned whether the increases might be due to greater frequencies of jet contrails, perhaps instead of or in addition to natural clouds. The question is as yet unanswered.Although Petersen (1990) reported that cloud observing techniques were changed in June 1951 by the National Weather Service, the cloud cover record does not exhibit any discontinuity during the period shown. Prior to June 1951, fractional amounts of only two possible cloud layers were observed and therefore recorded, plus the height of the lowest scattered layer. Following 1951, sky cover and cloud height were reported. Whether this procedural change may have biased the record has not been evaluated.

Changnon et al. (1980) investigated temporal changes in cloud cover over Illinois from the turn of the century to 1977 (not shown). They demonstrated that the number of cloudy days increased by about 50 percent from the early years (about 110 days per year in southern Illinois, and 140 days per year in the north) to about 160 days per year in the south, and 180 such days in the north.Petersen (1990) further calculated seasonal solar radiation since 1948 at several locations about the Midwest, based upon surface pressure, dew-point temperature, cloud height, and fractional sky cover. The results relevant to Illinois are presented in table 6. Positive trends in solar radiation over much the same areas were noted at most sites and during all seasons except spring and fall. However, the increase in cloud cover occurred primarily during the first half of the century, a period not covered in the Petersen study.As summarized by Folland et al. (1990), the upward trend in cloudiness has been observed in many other parts of the globe. They point out that some of the increase observed in the 1940s may be due to a change in observing practice; obscuring effects of smoke, haze, dust, and fog were included from the 1940s onward. However, the increases thereafter are believed to be real.

TRENDS IN SEVERE WEATHER Trends in Freezing Precipitation since the Turn of the Century

Changnon (1984) presented the number of days per year with observations of glaze (freezing precipitation) for six NWS stations (Peoria, Moline, Chicago, Spring-field, St. Louis, and Evansville, IN) from 1901 through 1980. At six sites, annual frequencies virtually doubled or more, beginning rather abruptly during the early 1940s. The higher frequencies continued until 1980, the end of the record. The absolute frequencies varied from site to site, with the greatest at Springfield, Peoria, and St. Louis. Data were collected from 1981 through 1990 to bring Changnon's figures to the present (figures 35 and 36). Although the "extreme" frequencies reported in the 1970s at Springfield and Peoria have not continued, they remain within the range of values reported after the increase noted in the 1940s or 1950s.

Procedures for observing or reporting freezing precipitation have not changed significantly during the 90-year period. As a result, no explanation is forthcoming for the doubling (or more) of frequencies of glaze beginning in the early 1940s. A question remains whether these abrupt changes are real or an artifact of a procedural change in observations. Discussions with the NWS did not uncover any procedural changes coincident with the upward trend.

Temporal Change in Thunderstorm Frequency Thunderstorms form in unstable air, the larger the area of instability, the greater the area of thunderstorms.

Further, greater instability leads to greater thunderstorm severity, i.e., strong, gusty winds, and possible hail and tornadoes. Changnon (1984) investigated the frequency of thunderstorms within and near Illinois from the turn of the century to 1980. We have updated those data through 1990, and present five-year totals of thunderstorm days for Chicago O'Hare and Peoria (figure 37), and Springfield, Moline, and St. Louis, MO, from 1901 through 1990 (figure 38), the pentad totals plotted at the first of the five years. First, there are no clear temporal trends. Thunderstorm frequencies were relatively low during the late 1940s and early 1950s at both Chicago and Peoria, the very years when thunderstorms at St. Louis were most frequent! Although thunderstorms are somewhat related to temperature, i.e., warmer temperatures lead to more thunderstorms, they also depend on instability and available moisture. Also, thunderstorms are relatively small-scale phenomena, i.e., the area of any one event is usually a linear feature (associated with a cold front), confined to a one- to three- state area. Since mean frontal locations shift northward and southward with warming and cooling, respectively, the area favoring thunderstorms tends to be located north of usual during warm episodes, and south of usual during cold episodes.

Trends in Tornado Frequency

Illinois lies at the northeastern limit of "Tornado Alley," the group of states that exhibits the greatest tornado frequency on average, generally Texas through Illinois. Tornado frequencies change rather dramatically from year to year. Since 1955, the period during which tornado observations are thought to be essentially complete, Illinois has experienced 28 tornadoes per year on average. But as many as 107 were recorded in 1974, and as few as seven and eight were recorded in 1964 and 1968, respectively (Wendland and Guinan, 1988).

The annual frequency of tornadoes in Illinois is not well known prior to 1955, a period when the NWS "received" as opposed to "sought out" data on tornado occurrences within the United States. Annual Illinois tornadoes since 1881 are presented in figure 39. The data from 1845 to 1954 are known to be incomplete, having been compiled from newspapers and U.S. Weather Bureau publications, as available and summarized by Wendland and Hoffman (1993). However, it is interesting that some annual tornado frequencies prior to 1955, such as 1883, 1886, and 1890, were as great or greater than some years since 1955, the time during which tornado records in Illinois are thought to be complete.

The annual record of the number of days with tornadoes in the state since 1960 (figure 40) exhibits several features: 1. Annual frequencies hover about a mean of about 11 tornado days per year from 1962 to 1971. 2. About 35 tornado days occurred in five consecutive years, 1973 through 1977. 3. About 25 tornado days occurred per year thereafter.

As to areas of the state that might seem to "favor" tornado activity, raw tornado frequencies clearly peak in those counties with greater population densities (Wend-land and Guinan, 1988). The authors believe, however, that this relationship results merely from the fact that more people are likely to see and report isolated, small-scale events as tornadoes, and does not represent an actual urban enhancement of tornado frequency.

Trends in the Frequency of Severe Snowstorms

Severe snowstorms in Illinois are defined as those where 6 inches or more snow or freezing precipitation of any intensity falls over at least 10 percent of the area of the state within a 48-hour period. These storms exert tremendous hardship on those within the area of impact in terms of disruption to commerce, communication, comfort, and economics. As opposed to the historical tornado record, that for severe winter storms is believed to be rather well defined since the turn of the century because each such storm generally impacts an area of some 15,000 square miles and would therefore be recorded by observations from 50 or more NWS Coop sites, even in areas of less population density than represented today (Changnon, 1969, 1978). Newspapers have proven to be a reliable source for a relatively complete record of severe winter storms, whereas tornadoes are not necessarily so reported, due to their relatively small area of impact (on the order of a few tens of square miles).

From the turn of the century to the early 1960s, severe winter storms have occurred about five times per winter (figure 41). After the early 1960s, the frequency has declined to about three such storms per winter. Winters with greatest severe storm frequencies include 1911-1912 with 12 storms, and 10 each in 1925-1926, 1943-1944, and 1950-1951. There were no severe winter storms in ten of the last 30 winters, including 1961-1962, 1962-1963, 1965-1966, 1967- 1968, and 1968-1969, 1970-1971, 1980-1981, 1981-1982, and 1982-1983, and 1991-1992.The record of severe winter storm frequencies in Illinois somewhat parallels that of temperature, suggesting increased winter storm frequency with increased mean temperatures. This relationship may be due to a number of factors, including greater moisture supplies during warmer winters, a more northerly storm track during the middle of winter, etc. However, this study did not examine specific factors for this relationship.

Trends in Illinois Droughts Drought can be measured by several techniques. In this instance we restrict our discussion to meteorological drought, i.e., a shortfall of precipitation. We present the percentage of the state experiencing 50 percent or less of the 30-year average precipitation during July and August from 1901 through 1991. The second uses the same threshold but for the full growing season, April through August, also from 1901 through 1991.

Five of the July-August droughts (figure 42) impacted at least 50 percent of the area of Illinois during the period 1901-1992, notably 1991 (impacting 79 percent of the state), 1930 (70 percent), 1936 (51 percent), 1971 (66 percent), and 1983 (55 percent). The recent 1988 drought impacted only about 30 percent of the state by this criterion.

According to the more stringent criterion, 50 percent or less than average precipitation from April through August (figure 43), a somewhat different distribution emerges, with extraordinary droughts in 1914 and 1936 (impacting about 60 percent of the state) and 1988 (54 percent). Neither 1914 nor 1988 were particularly note-worthy according to the different criterion, 50 percent or less precipitation during July and August. However, those years (particularly 1988) were characterized by extraordinary dryness in late spring and early summer.

Temporal Changes in Visibility

There has been and continues to be concern as to whether the atmosphere is becoming more turbid with time, perhaps due to industrialization, increased plowing and cultivation frequencies etc. Such a change could lead to reduced insolation, and related reduction in surface temperature, atmospheric clarity, and visibility. Although measurements of atmospheric turbidity are not generally available, we use several proxy records, including visibility and frequency of days with smoke or haze.

Wendland and Bryson (1970) discussed a simple relationship between mean temperature, carbon dioxide (CO2) concentration, atmospheric turbidity and sun-spots from 1880 to 1960. Using dust concentrations obtained in high snowfields of the Caucasus (Davitaia, 1965) which they suggest reflect integrated hemispheric loadings, they demonstrated that global mean temperature changes during those eight decades could be statistically explained by those three parameters. Specifically, the warming from 1880 through the mid 1900s was primarily related to increases in CO2 concentration, whereas the cooling thereafter was primarily related to the dramatic increase in dust load- ings, noted during the 1940s and 1950s. Atmospheric clarity can be defined and measured several different ways. The concept of visibility, however, is straightforward, the measurement technique has changed little over time, and hence visibility measurements of decades ago are generally comparable to those of today. Vinzani & Lamb (1985) showed declining visibility at NWS sites in and around Illinois since 1950. We analyzed hourly observations of visibility since 1948 at Chicago-O'Hare, Moline, and Peoria. Frequencies of occurrences of visibilities in various categories were accumulated for each year. Following Vinzani and Lamb (1985), we estimated the values that were exceeded 60 percent and 90 percent of the hours. Trends in these values are shown in figures 44 and 45. The trends in the 60 percent and 90 percent exceedance values are upward at Chicago, indicating improving visibility. Decreasing trends in the 90 percent value are seen at Moline and Rockford. Decreasing trends in the 60 percent value are seen at Springfield. The increasing visibility at Chicago, in contrast to the negative or insignificant trends at the other sites, may be due to more stringent pollution controls on industry.

Using yet another approach, Changnon (1987) showed that the number of smoke and/or haze days at six NWS First Order Stations in and around Illinois showed a dramatic increase from a few tens of days per year prior to the 1930s, to 100 or more days per year thereafter through 1980! We have updated the frequency of smoke and/or haze data through 1990 (figures 46 and 47), and find that the increased frequencies which abruptly began in the 1930s essentially continue to the present, though the highest frequencies at all sites were noted in the 1930s and 1940s. The values of the last few pentads continue at several times greater than pre-1930 frequencies, but about half the maximum-ever values.

ANOMALOUS EPISODES FROM THE ILLINOIS CLIMATE RECORD Global climate models (GCMs) suggest a future warm-ing in Illinois and redistribution of precipitation during the year under greenhouse-induced effects of additional trace gases into the atmosphere. What would the impacts of such climate changes be for Illinois? The model outputs provide only average conditions, although state resources vulnerable to climate are often influenced by interannual or multiyear anomalies or by several conditions not defined by GCMs. For example, the range of certain species of insects or vegetation in Illinois depends on such conditions as the number of frost-free days, the timing of precipitation events, and changes in diurnal temperature ranges, conditions not available from GCMs.The variability of a potential future climate may also be documented by studying the climates of the past. Identifying extreme seasons or years from the past that lie within the average conditions specified by the GCMs allows one to reconstruct variabilities and use past analogs to estimate future climates.

Over the past century, notable months, seasons, and years stand out as having been particularly warm, cold, wet, or dry. Indeed, it is often thecombination of temperature and precipitation extremes that stand out and impact Illinois resources. For example, the Dust Bowl years of the 1930s were prominent for their extreme drought and warm temperatures as were the 1988 drought and 1993 floods.

To understand how climate has varied over the past century, it is useful to isolate those episodes classified as 1) warm and dry, 2) cold and wet, 3) warm and wet, and 4) cold and dry. This characterization is also useful in grappling with the impacts of potential future climates in Illinois. To identify combined temperature and precipitation extremes from the Illinois climate record, individual mean monthly temperature and precipitation values were recorded for each year between 1895 and 1991 and normalized relative to the 97- year average. The index for temperature is calculated as: It = (Ti - Tavg) / Tstd

where: It = normalized monthly temperature Ti = average temperature for month in question Tavg = 97-year monthly mean temperature Tstd = 97-year monthly temperature standard deviation

For precipitation, the calculation is:

Ip = (Pi - Pavg) / Pstd

where: Ip = normalized monthly precipitation Pi = total precipitation for month in question Pavg = 97-year monthly mean precipitation Pstd = 97-year monthly precipitation standard deviation

A normalized monthly value near zero indicates a tem- perature or precipitation value similar to that of the 97- year average. Values above zero represent positive deviations (warmer or wetter than the average). Values below zero represent a negative deviation (cooler or drier than the average). The magnitude of the deviation is expressed in standard deviation units, i.e., +1.0 indi-cates that the anomaly is one standard deviation greater than the period average. A seasonal normalized value was calculated for each season. Warm and dry, and cold and wet seasons were distinguished by subtracting the normalized seasonal precipitation from normalized temperature: large positive numbers represent the former, and large negative numbers represent the latter. Warm and wet, and cold and dry seasons were distinguished by adding the normalized seasonal precipitation and temperature values: large positive results represent warm and wet seasons, whereas large negative numbers represent cold and dry seasons.

When adding the normalized values, results near zero indicate that either temperature or precipitation is greater than the mean by about the same amount that the remaining parameter is less than the mean. When subtracting normalized precipitation from normalized temperature, results near zero indicate near-equal anomalies for each parameter.

An inherent problem occurs when interpreting the result of either adding or subtracting normalized values of temperature and precipitation. One relatively large normalized value will dominate the result of either process if the other normalized value is a very small number, i.e., very close to the long-term mean. When results from either addition or subtraction were near zero, they had to be individually studied to determine the significance of the number. In this study, small normalized values, those between -0.25 and +0.25, were set to zero. (Therefore some years in the following figures exhibit no difference from the long-term mean.) Normalized precipitation and temperature values were added and subtracted to identify warm and dry, cold and wet, warm and wet, and cold and dry statewide conditions for each of four seasons between 1895 and 1991.

Statewide Summer Index

The combined indices for summer are plotted in figures 48 and 49. They show a high frequency (over three times as likely) of warm and dry, and cold and wet summers, as opposed to summers that were either warm and wet, or cold and dry. In addition, the magnitude of the latter index is much less than that for warm and dry, and cold and wet summers. As expected in a continental climate, heat waves are more likely to be associated with moisture deficiencies, and cool summers are more likely associated with excess moisture. The combined extremes of warm and dry, and cold and wet summers are more pronounced than warm and wet, and cold and dry summers. A second feature is the persistent nature of the occurrence of warm and dry, and cold and wet summers; i.e., summers of a given character tend to occur in succession. For example, the period between 1899 and 1926 is characterized by a variety of combined temperature and precipitation extremes. The period between 1930 and 1944 is almost exclusively warm and dry, whereas the period from 1961 to 1981 is almost exclusively cold and wet. The period between 1950 and 1959 is characterized by a return to a variety of extremes, as is the period between 1981 and 1991. All but one of the cold and dry summers occurred prior to 1920. Individual summers that stand out as warm and dry are 1901, 1913, 1914, 1930, 1933, 1936, 1983, 1988, and 1991; whereas cold and wet are 1902, 1907, 1915, 1924, 1958, and 1981; warm and wet are 1980 and 1987; and cold and dry are 1920 and 1976. For the summers with identified extremes, seasonal temperature and precipitation values and deviations from the 97-year average are presented in table 7.

Statewide Winter Index

The combined indices for the winter months are plotted in figures 50 and 51 as a function of time. In contrast to the summer months, warm and dry, and cold and wet winters are less frequent than warm and wet, and cold and dry conditions over the past 97 years. Cold and dry winters were three times more likely than cold and wet winters. The occurrence of cold and wet winters is restricted to the years after 1962, and three out of the four occurred after 1978. Individual winters that stand out as warm and dry are 1921, 1931, 1953, 1954, and 1987; whereas 1979 and 1985 were cold and wet; 1949, 1950, and 1983 were warm and wet; and 1963, 1977, and 1978 were cold and dry. For the win-ters with identified extremes, seasonal temperature and precipitation values and deviations from the 97-year average are presented in table 8.

Statewide Spring Index

The combined index for the spring months are plotted in figures 52 and 53 as a function of time. In contrast to both the summer and winter seasons, springtime combined temperature and precipitation extremes are more equally distributed among the four possible combinations, reflecting the relatively large day-to-day temperature differences during transition seasons. However, the warm and dry springs were for the most part restricted to the years between the 1920s and 1930s and the late 1980s.

The spring climate record also shows active and inactive intervals. The period between the 1920s and 1930s experienced numerous deviations from average condi-tions (usually warm and dry, or cold and dry), whereas the late 1940s to the mid-1970s was a somewhat benign period with only minor variations from the 96-year average. Individual springs that stand out as warm and dry are 1930, 1934, 1936, 1986, 1987, and 1988; whereas cold and wet are 1983 and 1984; warm and wet are 1921, 1922, 1945, and 1991; and cold and dry are 1901, 1926, and 1971. For the springs with identified extremes, seasonal temperature and precipitation values and deviations from the 97-year average are presented in table 9.

Statewide Autumn Index The combined indices for the autumn months are plotted in figures 54 and 55 as a function of time. Warm and dry autumns have occurred more often than the other three possible combinations, with the largest cluster of occurrences confined to the period between the late 1950s and mid-1960s. A significant warm and dry autumn has not occurred since 1971. The warm and wet autumns are clustered in the period between the early 1920s and the late 1930s. Individual autumns that stand out as warm and dry are 1938, 1939, 1953, 1963, and 1971; whereas cold and wet are 1911 and 1926; warm and wet are 1927, 1931, and 1941; and cold and dry are 1917 and 1976. For the autumns with identified extremes, seasonal temperature and precipitation values and deviations from the 97-year average are presented in table 10.

Statewide Annual Index While warm and dry, cold and wet, warm and wet, and cold and dry seasons have been identified from Illinois' 97-year climate record, the succession of the seasons and their climate has the ultimate impact. For example, the impact of a warm and dry summer would likely be greater if the previous spring and winter were also warm and dry in relation to their respective averages. Thus, efforts to group the climates of consecutive seasons are useful. For example, an examination of calendar years (a grouping of four seasons) identified 1914, 1930, 1934, 1936, 1953, and 1987 as warm and dry; the years 1926, 1950, 1951, and 1972 as cold and wet; the years 1973 and 1983 as warm and wet; and the years 1895, 1917, 1920, and 1976 as cold and dry. The results of several GCMs suggest that the trend of the Illinois climate of the future will be toward warm and dry summers and warm and wet winters. Analog years with similar summer and winter trends are 1930, 1933, and 1983.

Regional Index

The state of Illinois measures approximately 400 miles north to south, and the climates of northern, central, and southern Illinois are not homogeneous. For example, a cold and wet summer in northern Illinois may occur in concert with a warm and dry summer in the southern third of the state. The National Weather Service divides Illinois into nine climate regions (see figure 1, p.9). For this comparison, the northern sector of the state was represented by region 1, the central district by region 5, and the south by region 9. Warm and dry, cold and wet, warm and wet, and cold and dry seasons were identified in a fashion similar to the statewide calculations. Comparison of the combined index for each of the three regions and the statewide calculation shows general agreement across the region for each of the four seasons. A number of identified years overlap (especially those identified as extreme) and the trends over time are similar to the statewide trends. However, some year- to-year differences occur when an identified season for a particular region is not similarly identified as a statewide occurrence and vice versa. As illustration, warm and dry, cold and wet, warm and wet, and cold and dry summer seasons are identified for each of three regions (figures 56 and 57).

CONCLUSIONS

Although long-term trends can be identified in various climate parameters, the dominant characteristic of Illinois's climate is the presence of large variability on time scales of a few years or less. That is, the magnitude of long-term trends is in general considerably less than the changes that can occur from one year or a few years to the next. With that being said, however, some identifiable longer-term features can be noted. The most persistent and extreme summertime high temperatures occurred primarily during the 1930s, the Dust Bowl era. Indeed, average temperatures in Illinois clearly increased by 4 to 5oF from the mid- to late 1800s to the 1930s, and then cooled by about half that amount to the present. All of the temperature parameters assessed herein support those trends, including average temperatures and frequency of days with extreme temperatures, whether warm or cold. These trends are found in records of the continent, the hemisphere, and the world, although the magnitudes of change decrease with increasing areas of integration.

During the peak of Illinois temperatures during the 1930s, the frequency and intensity of hot days was unprecedented in Illinois and much of the upper Midwest. The many severe impacts of the heat and dryness of that period are well known. Since then, the frequency and intensity of summertime heat has been substantially less, particularly during the 1960s and 1970s. However, some resurgence of extreme summertime heat was noted during the 1980s, with accompanying impacts on crop yields. The extent of this trend toward higher frequencies of hot and dry summers cannot be predicted at this time.

In general, the frequency of extreme cold events increased from 1930 to the present. This upward trend reached its peak in the late 1970s and early 1980s, with an unprecedented string of extremely cold winters. They had major negative impacts on the Illinois economy and environment, affecting transportation, health, energy consumption, etc. During this cooling episode, precipitation may have increased marginally, but the major change in precipitation was the increase in annual snowfall, peaking in the late 1970s. Also notable was the abrupt decline in the annual number of severe winter storms in the state, dropping from an average of five per year prior to about 1980, to about three during recent years. The 1992-1993 winter was perceived by many to be a severe winter, but it was only near-average when compared to the situation of the last two or three decades.

Generally benign, moderate summers occurred during the 1960s and 1970s. As noted above, the 1980s were characterized by more frequent hot and dry but quite variable summers, similar to the earlier half of the century. It should be noted that midwestern agriculture experienced rapid technological advances during this relatively benign period. In addition, virtually all farmers active today have spent their entire careers during that benign period. The severity of the 1988 growing-season drought in Illinois had only been equaled in two other years since the turn of the century. The most recent Illinois drought of equal intensity occurred in 1936, long before virtually all present-day farmers were in the business.

The impact of climate variability in the state is exemplified by the cool summer of 1992. Even though overall corn yields were the highest ever in the United States, the northernmost 100 miles of the Corn Belt accumulated so few GDDs (sixth fewest since 1878) that the corn crop did not mature at its usual rate in that area. Indeed, it was not yet ripe at the time of the first frost in southern Minnesota and Wisconsin, which was within about one week of the long-term average.

The degree of cold of the 1992 summer was also demonstrated by the low total CDDs for the state. Total CDDs in 1992 were the seventh lowest since 1878. This was a boon for the consumer and a bane to the electric utility. The tornado frequency record in Illinois exhibits no trends, rather an average of some 25 tornadoes per year, but varying from 6 to 107 per year! Of interest was the relatively high frequency recorded during the early 1880s (essentially equal to that of today), even though the tornado record for that time is known to be incomplete. Cloud cover and the frequency of ice/glaze storms have both increased in the state. These changes do not appear to be due to changes in observing procedure, nor are they related to other parameters that might assist in their explanation. It is clear that climate has not been stable in Illinois during the last 150 years, nor should it have been so anticipated. Many of the changes, although abrupt, were of relatively small magnitude. Yet those relatively small-scalechanges levied a substantial toll on the inhabitants of the state through discomfort; lost income; increased costs; and impediments to commerce and agricultural production, in spite of major strides in hybridization and field management practices. Since climate has always changed, it will likely continue to do so. The past gives little hint as to the future direction and magnitude of these changes, but it does suggest limits within which climate may be constrained over the immediate future. The current increase to atmospheric carbon dioxide by the burning of fossil fuels, however, is suspected to change some climate parameters at a faster rate than during the past centuries, thereby possibly rendering past climate a poor tool for evaluating future varie period between the early 1920s and the late 1930 changed, it will continue to do so. The past gives little hint as to the future direction and magnitude of these changes, but it does suggest limits within which climate may be constrained over the immediate future. The current increase to atmospheric carbon dioxide by the burning of fossil fuels, however, is suspected to change some climate parameters at a faster rate than during the past centuries, thereby possibly rendering past climate a poor tool for evaluating future varibility.

ACKNOWLEDGMENTS

We acknowledge the many hours of programming by Julie Dian, who produced the scores of statistical calculations and figures initially produced for this study. Robin Shealy performed many of the statistical calculations. James Angel assisted with the figures, and Jean Dennison managed the manuscript.

REFERENCES

Changnon, S.A. 1969. Climatology of Severe Winter Storms in Illinois. Illinois State Water Survey Bulletin 53, Champaign.

Changnon, S.A. 1978. Record Severe Winter Storms in Illinois, 1977-78. Illinois State Water Survey Report of Investigation 88, Champaign.

Changnon, S.A., R.G. Semonin, and W.M. Wendland. 1980. Effect of Contrail Cirrus on Surface Weather Conditions in the Midwest - Phase I. Final report to NSF grant ATM 78- 09568.

Changnon, S.A. 1984. Climate Fluctuations in Illinois: 1901- 1980. Illinois State Water Survey Bulletin 68, Champaign.

Changnon, S.A. 1987. Historical Atmospheric Trans-mission Changes and Changes in Midwestern Air Pollution. Bulletin American Meteorological Society, 68(5):477-480.

Davitaia, F.F. 1965. Possible Influence of the Atmospheric Dust on the Regression of Glaciers and the Warming of Climate. Trans. Soviet Acad. Sci., Geogr. Ser. (2):3-22.

Draper, N.R., and H. Smith. 1981. Applied Regression Analysis. John Wiley & Sons, Inc., New York.

Folland, C.K., T. Karl, and K.Ya. Vinnikov. 1990. Observed Climate Variations and Change. In J.T.

Houghton, G.J. Jenkins, and J.J. Ephraums (eds.), Climate Change: The IPCC Scientific Assessment, Cambridge University Press, Cambridge, pp.195-238.

Huff, F.A., and J.R. Angel. 1989. Frequency Distribution and Hydroclimatic Characteristics of Heavy Rain-storms in Illinois. Illinois State Water Survey Bulletin 70, Champaign.

Lamb, H.H. 1966. The Changing Climate. Methune & Co.

Petersen, M.S. 1990. Implementation of a Semi-Physical Model for Examining Solar Radiation in the Midwest. Midwestern Climate Center, Illinois State Water Sur-vey Miscellaneous Publication 123, Champaign.

Vinzani, P. and P.J. Lamb. 1985. Temporal and Spatial Visibility Variations in the Illinois Vicinity during 1940- 1980. Journal of Climate and Applied Meteorology, 25:435- 451.

Wendland, W.M. 1990. A History of the Weather Observations in Illinois. Trans., Ill. Academy of Sci-ence, 83 (1 and 2):43-56.

Wendland, W.M., and R.A. Bryson. 1970. Atmospheric Dustiness, Man, and Climatic Change. Biological Conservation, 2(2):125-128.

Wendland, W.M., and P. Guinan. 1988. A Tornado and Severe Windstorm Climatology for Illinois: 1955-1986. Trans., Ill. Academy of Science, 81 (1 and 2): 131-146.

Wendland, W.M., and H. Hoffman. 1993. Illinois Tornadoes Prior to 1916. Trans., Ill. Academy of Science (in press).

Wigley, T.M.I. and T.P. Barnett. 1990. Detection of the Greenhouse Effect in the Observations. In T.J. Houghton, G.J. Jenkins, and J.J. Ephraums (eds.), Climate Change, the IPCC Scientific Assessment, Cambridge University Press, Cambridge, pp. 239-255.

Continue to Air Quality Trends in Illinois