To better understand dense fog events in the midwestern United States, a fog climatology was developed that examines the surface weather conditions at dense fog onset and during dense fog events, in relationship to fog duration. Surface airways hourly observations for the period 1948–96 were examined, focusing primarily on Peoria, Illinois, during the cold season (October–March). Temperature, winds, and visibility at dense fog onset did not prove to be useful in differentiating between short- (1–2 h) and long- (>5 h) duration dense fog events. However, it was found that once dense fog forms, it is more likely to persist if the horizontal visibility is 200 m (1/8 mi) or less and the ceiling height lowers to 30 m (100 ft) or less. Further, dense fog events at Peoria tend to last longer if they are widespread, that is, when many other midwestern surface airways hourly stations also report dense fog. When dense fog develops early in the night to the hours just after midnight, it is more likely to persist than when it develops later in the night or during the day. This was found to be the case for many other midwestern stations as well. Fog events forming earlier in the night may last longer because of the absence of solar insolation upon the fog layer during the night. As longer-duration fogs often become more opaque and more widespread than short-duration events, more time may be required to dissipate fog once the sun has risen. Dense fog onset time and the physical dimensions of the fog events appear to be the best predictors of fog duration considering all types of fog in the Midwest.
Widespread dense fog can have large socioeconomic impacts through disruption of commerce and jeopardizing personal safety. Ground transportation can be severely affected. Low surface visibility can slow or delay ground transportation throughout the year and result in accidents. Data obtained from the Illinois Department of Transportation indicate that between 1975 and 1995 some 4000 collisions occurred annually under foggy conditions in Illinois, excluding the city of Chicago, resulting in an average of 30 deaths per year and millions of dollars in damage (K. McGee 2001, personal communication). Accident reports from 1992 to 1995 indicate that, on average, at least two fog-related traffic accidents occur somewhere in Illinois, excluding Chicago, on 288 days per year.
Fog has not been examined in detail in the midwestern United States since the early to middle 1900s (e.g., Stone 1936; George 1951; Byers 1959). One reason for this may be that fog is not easily studied by remote sensing measurements such as weather radar and satellite. Fog occurs at too low an elevation to be studied by conventional radars. It is difficult to distinguish fog from low-lying stratus by satellite, although multiple wavelength techniques may allow more routine study of widespread fog events (Ellrod 1995; Bendix et al. 2005). Further, fog is highly variable in both time and space; at any one continental location in a given year, dense fog events may occur many times or not at all (e.g., Ratzer 1998). In addition, fog is a boundary layer feature, but also a phenomenon set up by synoptic-scale conditions, and one that sometimes is adversely affected by overlying cloudiness. Consequently, fog is not simple to model or to forecast on a routine basis, particularly for regions such as the Midwest where fog develops under a variety of conditions and where both radiation and advection processes may be important to the formation process. Fog forecasts are often observationally based (e.g., Herzegh et al. 2004) or model based but enhanced by observations (Vislocky and Fritsch 1997; Leyton and Fritsch 2004; Bergot et al. 2005). A physically based operational forecast model for the variety of conditions under which inland fogs may develop remains elusive.
To better understand dense fog events in the midwestern United States, a fog climatology was developed that examines the surface weather conditions at the time of dense fog onset and during dense fog events, in relationship to fog duration. The climatology is based on Peoria, Illinois (PIA), a site that is centrally located in the Midwest region and that is minimally affected by topographic and urban effects. This study is not confined to any one fog type. Instead conditions associated with long-duration events were contrasted with short-duration events to determine if there are identifiable conditions that may aide in forecasting/nowcasting the probable duration category of all types of inland dense fog events.
2. Data and methods
A half century of surface airways hourly surface observations have been collected at National Weather Service first-order stations since fog was examined by Stone (1936), George (1951), and Byers (1959) in the midwestern United States. These hourly observations consist of horizontal visibility, as well as other standard surface variables including cloud height and prevailing weather, for the period of study, 1948–96. For Peoria, during the period 1965–69, these data were recorded every 3 h instead of hourly. No adjustments were made for the 3-h data.
Horizontal visibility was used to examine the frequency and spatial extent of dense fog days over the Midwest and to identify dense fog events at Peoria (Greater Peoria Regional Airport), located near the center of the study region (Fig. 1). The airport is situated on level tableland that is surrounded by well-drained and gently rolling terrain, well away from (1.6 km) the rim of the Illinois River valley and well above (∼70 m) the riverbed (National Climatic Data Center 2005). The airport is on the sparsely populated southwestern edge of the city of Peoria. Thus, urban and topographic influences likely have minimal affect on the weather at the site. A number of parameters were examined to further address whether Peoria might be representative of the region: the annual frequency of dense fog events, the proportion of events occurring in the cold versus the warm season, the relative proportion of long- versus short-duration events, and the average wind speed during dense fog events. Based upon this analysis, Peoria was found to be representative of many sites in the Midwest, most located in Iowa, Missouri, Illinois, and Indiana.
Dense fog is defined by horizontal visibilities of ≤400 m (0.25 mi) accompanied by a reported fog observation. A dense fog day is identified when at least 1 h of dense fog occurs at a station within a calendar day. Following the method used by Meyer and Lala (1989), a dense fog event is separated from other events by at least 6 h with no fog reported [horizontal visibility exceeds 3.2 km (2 mi)] or 12 h with visibilities equaling or exceeding 1.6 km (1 mi). Though unusual, several fog events can occur on a single day, and it is not uncommon for fog events to extend over the midnight hour.
Dense fog events were grouped into long-, medium-, and short-duration categories to discern conditions leading to duration differences. Objective criteria for classifying fog events by duration were not found in the literature. For practical and subjective reasons, fog events were classified by duration as follows: short events are those lasting only 1–2 h, medium-duration events are those lasting 3–5 h, and long-lasting events are those lasting 6 or more hours. When only 3-hourly observations were available, a short-duration event will have one observation, a medium-duration event two observations, and a long-duration event three or more observations.
Data on fog-related traffic accidents were obtained from the Illinois Department of Transportation (K. McGee 2001, personal communication). Summary information including the number of fog-related accidents, all weather-related accidents, and road surface conditions for Illinois, excluding the Chicago area, were provided for the years 1975–95. Individual accident reports were provided for 4 yr, 1992–95. Individual reports included the time, date, and type of accident; the county; the road surface conditions; and observed weather conditions.
3. Spatial distribution of dense fog days
A number of fog climatologies have described the spatial pattern of fog days for the United States either annually (Stone 1936; Court and Gerston 1966; Pearce 1969; Hardwick 1973) or monthly (Hardwick 1973). Other early studies described the seasonal attributes of fog in various regions of the United States and their causes, but did not quantify the frequency of dense fog (George 1940; Byers 1959; George 1960). The spatial distribution of dense fog days based on the 1948–96 data is presented in Fig. 1. In general, the frequency of dense fog days is similar in both frequency and spatial distribution to the annual average of all fog days observed by Court and Gerston (1966) for the period 1896–1960, but is much larger than the dense fog day frequency observed by Stone (1936), and by Court and Gerston (1966). The more recent data are improved in three ways (Court and Gerston 1966). They are more frequent, with 8 (1965–69) to 24 rather than 2 to 5 observations per day. They are taken primarily at airport locations, reducing the impact of urban heating, and finally, the severity of fog is recorded as horizontal visibility rather than or in addition to descriptive terms.
During 1948–96, dense fog was reported by at least one of the 57 stations in the midwestern United States, on 11 641 days, or 65% of all possible days. The median number of dense fog days per station per year was 17. Three stations along the southern edge of Lake Superior averaged more than 40 days of dense fog per year, and stations along Lake Michigan and in northern Iowa, southern Wisconsin, and northwestern Illinois averaged between 20 and 30 days of dense fog annually. Note that large differences can occur at sites in close proximity. For example, 22 and 7 days per year (Fig. 1) are indicated at the two Kansas City stations near the western border of Missouri. The Kansas City International Airport site reports, on average, some 3 times more fog than the Kansas City Downtown Airport site. Note also the reduced number of fog days near Chicago and Detroit. This strongly suggests that the often-observed elevated minimum temperatures in urban areas may reduce the occurrence of fog (e.g., Landsberg 1981; LaDochy 2005). There is also a suggestion of an increase in dense fog frequency in western Pennsylvania. The region under study, however, does not extend far enough to the east to adequately reflect the prevalence of fog in the Appalachian region found by Stone (1936), Court and Gerston (1966), Pearce (1969), and Hardwick (1973).
4. Fog as a cold season feature
Illinois traffic accident data suggest that fog events can be hazardous throughout the year, but that the most serious impacts are found in the cold season. On many days at least two fog-related traffic accidents were reported (not necessarily dense fog), but most fog-related accidents occurred on a few days. During a 4-yr period, 1992–95, at least 25 accidents were reported on 104 days, and more than 100 accidents were recorded on only 22 days. For Illinois, 84% of days with at least 25 traffic accidents involving fog occurred during the cold season, and 100% of days with at least 100 accidents involving fog occurred from October through March. More than 15 stations reported dense fog and lasted more than 6 h on all days when 100 or more fog-related traffic accidents occurred in Illinois.
The monthly distribution of fog days is presented in Fig. 2. Dense fog days occurring at only a few stations (one to five) occur throughout the year. As the number of stations experiencing fog on a given day increases, the frequency of occurrence increases in the colder months and decreases in the warmer months. Of the 384 days in the Midwest region with more than 15 stations recording dense fog, only 15 occurred outside the October–March cold season. Quite clearly, the most widespread events are cool season features. Byers (1959) found that radiation fogs occurred year round and that fogs associated with fronts were cold season features. George (1940) and Stephens (1943) also found a cold season peak in fog frequency, but their sample also included cases of low stratus. Ratzer (1998) and Friedlein (2004) also found fog occurring throughout the year, but fog was primarily a cold season feature in Chicago, Illinois.
For Peoria, during the 49 yr of data, there were 792 dense fog events, and 71% of these events and 84% of the 3360 dense fog hours occurred during the October–March cold season. Peoria events with 10 or more other stations reporting fog also occurred primarily during the cold season (91%), with the most widespread event encompassing 32 stations. Longer-duration events, those with durations of more than 5 h, also were found to occur primarily during the cold season (87%), with the longest dense fog event at Peoria lasting 47 h. Again, dense and widespread fogs occur more frequently during the cold season months of October–March. The meteorological conditions associated with 561 Peoria cold season dense fog events are presented in the following section.
5. Cold season dense fog events at Peoria
a. Onset and ending times
It is well known that fog is primarily a nighttime feature with the peak frequency in the early morning hours surrounding sunrise (e.g., Meyer and Lala 1989). Examination of dense fog events at Peoria, however, indicate a difference in onset and end times for the short and long-lasting events. The peak onset time of the long-duration events is around midnight and progressively later for the medium- and short-duration events (Fig. 3a). The peak onset for short-duration events is near sunrise (0545–0715 LST between October and March). The peak end time for all categories is centered near sunrise, with that of the short-duration events at 0700 LST and long- and medium-duration events at 0800 LST (Fig. 3b). The longer-duration fogs not only form earlier, but also take somewhat longer to dissipate after sunrise than the shorter-duration fogs. The difference in the median end times for the long- and short-duration events is significant at the 0.05 level employing the Mann–Whitney U test (e.g., Burt and Barber 1996).
The onset and end times were also examined for 43 other surface airways stations for which hourly data are available to the author and have continuous observations for the period 1951–96. The analysis excluded the six most western stations in Fig. 1, and stations with missing years of data. The onset and end times were composited by averaging the percent frequency of occurrence at each hour for each station. Evidence of long-duration fog events starting earlier than the medium- and short-duration events was found at all but two stations. The timing of fog onset varied between stations, however. For long-duration dense fog events, one or more peaks in onset times occurred within the hours of 2200 and 0400 LST for most stations. For short-duration fog events, one or more peaks in onset and end times for most stations occurred within the hours of 0600–0900 LST.
b. Spatial extent of dense fog events
The spatial dimensions of dense fog events occurring at Peoria were estimated by considering the number of surface airways stations in the Midwest with dense fog, when dense fog was observed at Peoria. Another spatial estimate considered the number of stations adjacent to (within about 200 km of) Peoria with dense fog. The stations and years that were employed in the previous section were employed in this analysis with one difference. Both stations located in Chicago, Illinois (Chicago O’Hare International Airport and Chicago Midway Airport) were considered, but were counted as one station (when both stations reported dense fog) during the time period where their operations overlapped (1959–79). The two Kansas City stations were combined in a similar fashion. Thus, for the spatial dimension analysis, there were 43 possible stations in the Midwest, and 6 possible “adjacent” stations within about 200 km of Peoria. The number of stations reporting fog and the number of adjacent stations with fog are daily values, and the fog stations within a given event are often but not necessarily contiguous.
The three duration classes of fog events are clearly different in horizontal extent (Figs. 4a and 4b). The median number of stations with dense fog is 15 and 8 for long- and short-duration events, respectively (Table 1). For long-duration events, often four to seven of the stations within 200 km of Peoria reported fog, while often only zero to three stations reported fog within 200 km of Peoria during short-duration events. With the deployment of the Automated Surface Observing Sites in the late 1990s, the number of stations with horizontal visibility has approximately doubled and a better estimate of spatial extent would be possible with these more recent data.
c. Visibility and ceiling
Horizontal visibility and ceiling height observations were examined to estimate the intensity of dense fog events. The horizontal visibility at the time of onset was similar but somewhat lower for long-duration events (Table 1). This similarity was likely the result of the definition of dense fog as any fog with a visibility of 400 m (1/4 mi) or less. However, during long-duration fog events, the horizontal visibility often decreased. The minimum visibility was 100 m (1/16 mi) or less for 73% of the long-duration events and was 200 m (1/8 mi) or greater for 80% of the short-duration events (Fig. 5a). The difference in the median minimum event visibility for the long- and short-duration fog events is significant at the 0.05 level.
Ceiling height is defined as the vertical visibility into obscuring phenomena, not classified as thin, that covers 6/10 of the sky, or the height of the lowest sky cover that is more than 1/2 opaque (Steurer and Bodosky 1998). Ceiling height might be expected to be higher for less dense fog events where visually one can see further vertically into the atmosphere. The ceiling height at the time of onset was similar for all duration categories (Table 1). The minimum ceiling observed over the duration of the event, however, was lower for the long-duration fog events. About 70% of the long-duration events had ceilings of 0 m during the event, whereas that was the case for only about 18% of the short-duration events (Fig. 5b). The difference in median minimum event ceiling for the long- and short-duration fog events is significant at the 0.05 level. Thus, the long-duration dense fog events became more opaque both horizontally and vertically with time.
d. Wind and temperature at onset and during dense fog events
Low wind speeds have long been known to favor fog development. Meyer and Lala (1989), for example, found that 69% of their dense radiation fog events had an average speed of ≤1 m s−1 (2 kt) and 87% had an average speed of ≤2.1 m s−1 (4 kt). The median wind speed at the time of dense fog onset, 3.1 m s−1 (6 kt), was similar for the three fog event duration categories (Table 2), but higher than the nearly calm conditions often reported for radiation fog events (e.g., Meyer and Lala 1989; Roach, et al. 1976). Considering all cold season dense fog events at PIA, only 15% had speeds of less than 2.5 m s−1 (5 kt) at onset. During long- and medium-duration events, peak speeds often were higher than during short-duration events (Fig. 6). The fact that higher wind speeds than those of typical radiation fog events were found for all three duration categories of Peoria events suggests that advection may be occurring during many of the midwestern dense fog events and may help maintain the longer-duration events. As fog is predominantly a nighttime phenomenon, it is likely that radiation processes are important even in events where advection is occurring.
The predominant wind direction at fog onset was from the southwest, south, or southeast. A secondary maximum in wind direction was from the northeast. No difference was found in the distribution of wind direction for the long- and short-duration dense fog events (not shown). Fog also can form over a wide range of temperatures, with 90% of all events occurring when the temperature at the time of dense fog onset was between −7° and 13°C. No difference in median temperature at the time of fog onset was observed for the long- and short-duration events (Table 2).
Wind speed and temperature varied more during the long-duration events than during the shorter-duration events, with both higher maximum and lower minimum values while visibility remained 400 m (1/4 mi) or less. Sometimes, the temperature was observed to increase even during the nighttime hours. In part, these observations may reflect the diurnal change in atmospheric conditions, may occur simply because with more observations, there is a higher probability of higher or lower values, or may indicate warm-air advection. These observations are somewhat reminiscent of a case study by Jiusto and Lala (1980), who found the following after the onset of dense fog: stronger winds and vertical mixing, a temperature inversion transformation to warmer temperatures near the surface as radiational cooling shifted from the surface to cloud top, and higher droplet concentrations. No data regarding the vertical structure of the dense fog environment are available for the Peoria events that might explain the mechanisms behind fog maintenance in the Midwest.
An important aspect of this study was to determine if there are significant differences in meteorological conditions that can be discerned at dense fog formation, or during dense fog events that would help predict whether a dense fog event would be short or long in duration. The time of fog onset was found to differ for long- and short-duration events. The longer-lasting events attained lower horizontal visibilities, ceilings, and wider ranges of wind speed and temperature during the event than did the shorter-duration events. Dense fog also was reported at more stations during the longer-duration events. The longer-lasting events, events associated with high pressure, and even those associated with a low or front were observed to form earlier in the nighttime hours and to persist an hour later in the early morning than the short-duration events.
The proportion of dense fog events forming at each given hour that is of short, medium, or long duration was computed for the entire sample of cold season dense fog events at Peoria (Fig. 7). For dense fog events that form during the hours of 1900–0200 LST, 50% or more are long-duration events at any given hour. For dense fog events that form between 0700 and 1300 LST, about 50% or more are short-duration events. For events forming at 0500 LST or later in the morning, 25% or less are long-duration events at any given hour. Less than 20% of the events that form between 1800 and 0300 LST are short-duration events. This proportion of dense fog events was also examined for all of the 44 stations (including Peoria). A similar pattern was found when examining all 44 stations (Fig. 8). In both figures, long-duration events peaked in the nighttime hours, suggesting the importance of radiational cooling in all types of fog. A number of long-duration fog events were observed to form during the afternoon hours suggesting that advection also is important in this region.
Longer-lasting fogs events tend to persist longer after sunrise than do the short-duration events. Pilie et al. (1975) found that fog could persist after sunrise due to the evaporation of dew. Mason (1982) and Nakanishi (2000) showed that even as insolation increases after sunrise, radiational cooling continues, and in the absence of advection, fog begins to dissipate only when surface heating becomes dominant. Meyer and Lala (1989) found that the majority of radiation fogs dissipated within 3 h after sunrise largely because of continued mixing through local heating resulting from increasing insolation and the entrainment of drier air. As longer-duration fogs often become more opaque than short-duration events, more time may be required to dissipate fog once the sun has risen. Widespread fogs also might be more difficult to dissipate if fog erodes at the edges as suggested by Gurka (1978). Regardless of the exact physical processes involved in fog dissipation, the onset time of dense fog and the areal extent of dense fog appear to be important predictors of fog duration.
7. Summary and conclusions
Meteorological conditions at dense fog onset and during dense fog events were examined for the cold months of October–March 1948–96 at Peoria, Illinois. Peoria was specifically chosen for investigation because of its central location in the Midwest and its suitable site conditions. The events were examined to determine whether conditions at dense fog onset and during the fog event could predict dense fog duration. It was found that once dense fog forms, it is more likely to persist if the horizontal visibility is 200 m (1/8 mi) or less and the ceiling height is 30 m (100 ft) or less. Further, the more widespread events tend to last longer than dense fog events that encompass only a few stations. Last, when dense fog develops during the late evening to postmidnight period, it is more likely to persist than when it develops later in the night or during the day. Longer-lasting events tend to be more opaque and widespread than the short-lived events and dissipate an hour or two later than the shorter-lived events. Fog events forming earlier in the night may last longer simply because of the absence of the destructive influence of solar insolation on the fog layer. Also as longer-duration fogs often become more opaque than the short-duration events, more time may be required to dissipate fog once the sun has risen. Dense fog onset time and the physical dimensions of the fog events appear to be the best predictors of fog duration for all types of fog common in the Midwest.
Data on fog-related traffic accidents were obtained from Karen McGee, of the Illinois Department of Transportation. The author thanks Scott Isard, Stanley Changnon, David Kristovich, Kenneth Kunkel, and Colin Thorn for their inspiration in this research. The author also thanks the two anonymous reviewers for their helpful insights. Any opinions, findings, and conclusions are those of the author and do not necessarily reflect the views of the Illinois State Water Survey.
Corresponding author address: Nancy E. Westcott, Illinois State Water Survey, 2204 Griffith Dr., Champaign, IL 61820. Email: firstname.lastname@example.org