This study focuses on dense fog cases that develop in association with low clouds and sometimes precipitation. A climatology of weather conditions associated with dense fog at Peoria, Illinois, for October–March 1970–94 indicated that fog forming in the presence of low clouds is common, in 57% of all events. For events associated with low pressure systems, low clouds precede dense fog in 84% of cases. Therefore, continental fogs often do not form under the clear-sky conditions that have received the most attention in the literature. Surface cooling is usually observed prior to fog on clear nights. With low cloud bases, warming or no change in temperature is frequent. Thus, fog often forms under conditions that are not well understood, increasing the difficulty of forecasting fog. The possible mechanisms for fog development under low cloud-base conditions were explored for an event when dense fog covered much of Illinois on 7 November 2006. Weather Surveillance Radar-1988 Doppler (WSR-88D) and rawinsonde observations indicated that evaporating precipitation aloft was important in moistening the lower atmosphere. Dense fog occurred about 6 h following light precipitation at the surface. The surface was nearly saturated following precipitation, but relative cooling was needed to overcome weak warm air advection and supersaturate the lower atmosphere. Surface (2 m) temperatures were near or slightly cooler than ground temperatures in most of the region, suggesting surface sensible heat fluxes were not important in this relative cooling. Sounding data indicated drying of the atmosphere above 800 hPa. Infrared satellite imagery indicated deep clouds associated with a low pressure system moved east of Illinois by early morning, leaving only low clouds. It is hypothesized that radiational cooling of the low cloud layer was instrumental in promoting the early morning dense fog.
In the Midwest fog can be a significant safety hazard, greatly impacting ground (e.g., Goodwin 2002; Westcott 2007) and air (e.g., Keith and Leyton 2007) transportation. Continental fogs are generally thought to occur when air is cooled to the point of saturation by radiation under clear-sky conditions, or when moist air is advected over a cold surface, such as snow, resulting in cooling of the overlying air to saturation (Roach 1995). In fact, much of the current understanding of continental fogs is based on field studies of radiation fogs under clear-sky conditions (e.g., Roach et al. 1976; Meyer and Lala 1990; Jiusto and Lala 1980; Mason 1982; Turton and Brown 1987). Based on observations taken in the October–March cold season from 1970 to 1994, Westcott (2005) found that precipitation is often present at the onset of dense fog in the Midwest. Tardif and Rasmussen (2007) likewise found for New York City, and the nearby metropolitan areas, that fog often occurs with precipitation or low cloud bases. Fogs forming in association with or between low pressure systems or nearby fronts were often observed by early researchers (e.g., George 1951; Byers 1959) and more recently by Westcott (2005), Croft and Burton (2006), and Tardif and Rasmussen (2008). While forecasters are aware that continental fogs can be associated with low clouds and precipitation (George 1940; Byers 1959; Croft et al. 1997), few case studies examining processes leading to the development of fogs over land (e.g., Tardif 2006) have been reported.
A climatology of fog occurrence is presented for a site in the continental Midwest. Furthermore, a case study is presented to illustrate some of the physical processes that may be important for the development of dense fog under low cloud conditions, or following the passage of a precipitation system.
A climatology of surface conditions, before and at the time of dense fog formation, was developed for all dense fog events observed at Peoria, Illinois, using hourly surface airways data for the period 1951–96 (Westcott 2005, 2007). These hourly observations consist of horizontal visibility, as well as standard surface observations including cloud-base height and prevailing weather. The areal extent of each fog event was based on surface airways data collected at National Weather Service (NWS) first-order stations in the Midwest region. Dense fog events were defined as having at least one hour of visibility ≤400 m (¼ mi) when fog was reported, similar to the definition of fog events by Meyer and Lala (1990). Peoria, located in central Illinois, was found by Westcott (2007) to be representative of the Midwest. The Champaign, Illinois, area is about 110-km southeast of Peoria, and the location of some of the surface observation sites used in the current case study is similar to Peoria in that topographic features and urban influences do not greatly affect the surface observations.
The majority (>80%) of widespread, long-duration events in the Midwest occur during the October–March cold season period. These events are the basis of the present climatology (Westcott 2007). During the 1970–94 cold seasons, 302 dense fog events were identified. Each of these events was classified by synoptic type based on 3-h surface weather maps. Ten synoptic types were considered in Westcott (2005). The most notable feature was that precipitation often occurred within one hour of the onset of dense fog events associated with low pressure systems and approaching or nearby fronts. Because of the similarity in the distribution of surface conditions and prevailing weather at dense fog onset and for simplicity of analysis, the 10 original classifications were combined into 5 synoptic types for this study. The synoptic classification was made with respect to Peoria, at the time of dense fog onset (Table 1). Examples of the 5 synoptic types are presented in Fig. 1, based on images from the U.S. Daily Weather Map Series. [Available online at http://docs.lib.noaa.gov/rescue/dwm/data_rescue_daily_weather_maps.html archived by the National Oceanic and Atmospheric Administration (NOAA) Central Library Data Imaging Project for dates prior to 2003, and by the National Centers for Environmental Prediction (NCEP; available online at www.hpc.ncep.noaa.gov/dailywxmap/pdffiles.html) for later dates.] Of the 302 dense fog events, 23% occurred when weather in Illinois was dominated by a low pressure system centered in Illinois or one of the surrounding states at the time dense fog formed, as was the case for the event on 7 November 2006. Other synoptic categories included events associated with approaching or nearby fronts (35%), with high pressure system (26%) events that occurred following the passage of a front (7%), or were between synoptic features (termed transition events, 9%).
The original climatology was expanded upon for this study by the inclusion of upper-air data from Peoria. The current study focuses on available rawinsonde pairs for those events when the 0000 UTC (1800 LT) sounding was launched before dense fog onset, and when the 1200 UTC (0600 LT) sounding launch occurred at fog onset (21%), during dense fog (73%), or within 2 h of the end of dense fog (6%). There were 120 sounding pairs that characterized about 40% of the dense fog event sample. These were employed to characterize the depth of the moist layer near the time of dense fog events, and determine the possibility of cold or warm air advection at low levels.
Precipitation data for the 302 dense fog events at Peoria were obtained from 3 sources: 1) daily precipitation totals at Peoria, ending at midnight; 2) the NOAA Daily Map Series daily precipitation analyses for the surrounding region, ending at 1200 UTC (0600 LT); 3) hourly observations of prevailing weather (precipitation type) at Peoria. The midnight observation at Peoria and the Daily Weather Map Series data provide an indication of availability of moisture for fog development at Peoria and in the general region. Peoria is not one of the sites analyzed on the Daily Weather Maps, so there is not perfect agreement between the maps and precipitation at Peoria. In all cases in which precipitation was observed near the time of dense fog onset and in most cases (>90%) in which precipitation was observed to occur within the previous 24 h at Peoria, an area of precipitation was observed on the Daily Weather maps. In cases in which precipitation was not specifically observed at Peoria, it usually was observed within about 100 km. This suggests that there may be cases where precipitation was in the immediate area, but not reaching the ground.
b. Case study
Operational observations taken by NOAA facilities and local field experiment observations in central Illinois were examined to obtain as complete a description as possible of the 6–7 November 2006 dense fog event (Fig. 2). Surface observations from NWS Automated Surface Observing System (ASOS) and Federal Aviation Administration (FAA) Automated Weather Observing System (AWOS) locations in Illinois and Indiana, taken at time intervals of 20 min to 1 h, were obtained from NOAA’s Midwestern Regional Climate Center (available online at http://mrcc.isws.illinois.edu) at the Illinois State Water Survey. These sites record visibility, cloud-base height, and precipitation, as well as standard meteorological variables. Hourly-averaged data from the Illinois Climate Network (ICN) gave valuable additional information on standard meteorological variables, precipitation, net radiation, ground temperatures, and 10 cm (4 in.) soil temperature and moisture at 19 sites throughout Illinois. In addition, data taken at the University of Illinois Willard Airport AWOS (KCMI) about 5 km south of Champaign gave detailed information on the evolution of the fog. An AmeriFlux site (Billesbach et al. 2004), located about 4 km south–southwest of KCMI, provided information on surface sensible heat fluxes, and 2-m air and ground temperatures; a duplicate set of most instrumentation was located at a second tower about 4 km south of KCMI. The two AmeriFlux towers were separated by 2 km. While visibility was not measured by AmeriFlux, the temperature, humidity, wind measurements, and temporal trends were quite similar to those at KCMI and at the ICN sites at Bondville, Illinois, and Champaign.
For fog development in cases of low clouds and precipitation, it is anticipated that variations in atmospheric conditions throughout the lower troposphere might be important. High-resolution (6 s) rawinsonde observations taken near the center of the area of dense fog development were obtained from the NOAA Weather Service Forecast Office in Lincoln, Illinois (KILX). These observations were taken at 0000 and 1200 UTC (1800 and 0600 LT) on 6–7 November 2007 and examined using the Universal Rawindsonde Observation Program (RAOB) software (Environmental Research Services 2006). Weather Surveillance Radar-1988 Doppler (WSR-88D) observations taken at Lincoln, were examined to understand the potential impacts of precipitation on fog development in central Illinois. Archived level II data were obtained from the National Climatic Data Center (NCDC; available online at http://hurricane.ncdc.noaa.gov/pls/plhas/has.dsselect) and viewed using the software program GRLevel2Analyst (Gibson Ridge Software 2006). Satellite observations gave critical information on the spatiotemporal evolution of cloud systems in the region of fog development. Visible and infrared imagery taken by Geostationary Operational Environmental Satellite (GOES-8) were examined. This imagery was obtained from the National Aeronautics and Space Administration (NASA) Langley Cloud and Radiation Research Group Web page (available online at http://www-angler.larc.nasa.gov/).
In a number of ways, fogs forming under differing synoptic conditions in Illinois are quite similar (Table 2). Dense fog is usually a nighttime event. It often occurs with low wind speeds and when surface flow is from the south or east. Further, it often occurs when snow is not present within 100 km of Peoria. There also are some obvious differences. Winds are usually lighter when high pressure dominates Illinois (median 2.6 m s−1), or when the weather is not clearly dominated by a major surface feature (median 2.1 m s−1). In comparison with events associated with high pressure systems, when a low pressure system is nearby fog is more likely during daytime hours and with higher (although still light) surface winds (median 3.6 m s−1). However, within any particular synoptic class, fog forms under a wide variety of surface conditions.
Clouds are known to impact fog development by delaying its onset or in leading to the rapid termination of a fog event (e.g., Saunders 1960). Examination of cloud-base heights (Table 3), however, suggests that fog in the Midwest often occurs under cloudy conditions. Employing the cloud-base height threshold of 1 km used by Tardif and Rasmussen (2007), low cloud-base events are common, particularly when low pressure centers (84%) dominate Illinois weather. Low cloud bases also are common when fronts are approaching or nearby (67%). The majority of all dense fog events (57%) were associated with low cloud bases in the 6 h prior to the onset of dense fog. The impact of cloudiness on changes at the surface and in the moist layer is examined in the next two sections.
a. Cloud-base height with surface temperature and moisture changes
Typically under clear-sky conditions during nighttime hours, divergence of radiation fluxes cools the surface and lowest region of the atmosphere. In cases where the sky was clear, or where cloud bases were higher than 1 km in the 6 h prior to dense fog onset, made up only 34% of the dense fog events (Table 3) and were usually associated with high pressure or followed the passage of a warm or cold front, or a front without precipitation. Figure 3 reveals that temperature falls of 0.6°C or more in 6 h can be expected when cloud bases are greater than 1 km. Dewpoint temperatures were also observed to decrease, but not as frequently as did temperature. Relative humidity in the 6 h prior to dense fog onset often increased by more than 5% when cloud bases were higher than 1 km. This suggests that radiational cooling occurred, allowing the relative humidity to increase perhaps without the addition of moisture. By 3 h prior to dense fog onset, however, while cooling still frequently occurred, the atmosphere was nearly saturated, and so only small changes in relative humidity were seen. It should be noted that particularly at nearly saturated conditions, relative humidity changes of 5% may be within sensor accuracy, and thus variations at these high values are likely within sensor noise. When cloud bases are variable in the 6 h prior to dense fog onset, a weaker but similar pattern of temperature and relative humidity change is observed.
For low cloud-base dense fog events, small or positive changes in temperature, dewpoint temperature, and relative humidity are common in the 6 h (Fig. 3a) and 3 h prior to dense fog onset (Fig. 3b). In the majority of cases a warming of ≥0.6°C or no temperature change occurred over the 3 and 6 h prior to dense fog onset. If precipitation was present at or near dense fog onset, warming was most common. Note that hourly temperature was recorded in whole degrees Fahrenheit (0.6°C), so more subtle changes in temperature may have gone undetected. An examination of dewpoint temperature by cloud-base height indicated that under low cloud conditions with and without precipitation at fog onset, the dewpoint temperature often increased by ≥0.6°C in the pre-fog hours. This occurred with more frequency than did increasing temperature cases, suggesting the addition of moisture into the surface layer.
For low cloud cases with precipitation at dense fog onset, the atmosphere often was nearly saturated, with only 15% of the cases exhibiting increasing humidity by more than 5% in the 3 h prior to dense fog onset, and 30% increasing more than 5% in the 6 h prior to dense fog onset. Similar results were presented in Tardif and Rasmussen (2008) for precipitation fog events. Within 3 h of dense fog onset, in the majority of cases, the atmosphere was saturated or nearly saturated no matter the cloud-base height.
b. Vertical structure of the moist layer from sounding data
Sounding pairs were examined for 120 events when dense fog onset occurred after the 0000 UTC (1800 LT) sounding, with the 1200 UTC (0600 LT) sounding generally representative of the fog event (94% of the 1200 UTC soundings occurred when dense fog was present). The data for this subset of dense fog events showed some similarity in vertical structure among the events. It was found that an inversion was nearly always present in the early morning, no matter the synoptic type or cloud base presence or height. Of all soundings, the inversion generally (86%) lowered and/or strengthened from what was found 12 h previously. The inversion layer was typically (86%) within or corresponded to the moist layer. At 1200 UTC (0600 LT), 50% of inversion layers were less than 172 m AGL (970 hPa), 90% <500 m AGL (935 hPa), and 95% <1000 m AGL (850 hPa).
At 1200 UTC, the depth of the moist layer (T − Td ≤ 1°C) differed by synoptic type and by cloud-base height (Table 4). The events with no or high clouds had the shallowest moist layers. These were dominated by high pressure and post frontal cases, but many of the frontal cases likewise showed shallow moist layers. When low clouds were present, the moist layer was generally deeper. While it might be expected the moist layer would be deepest when precipitation was observed near the time of dense fog onset, this was not always the case (Table 4). This suggests that observed precipitation sometimes may have resulted from settling within the fog layer. The depth of the moist layers found here for no/high cloud events and for low cloud events, are similar to those for radiation and coastal advection fog events, respectively, reported by both Jiusto (1981) and Croft et al. (1997).
Other differences could be observed between synoptic type and cloud-base categories regarding changes in conditions overnight and in conditions present at 1200 UTC. The 0000 and 1200 UTC sounding pairs were compared to determine the percentage of cases when cooling or moistening predominated in the moist layer during the intervening 12-h period. The results were similar to those found examining surface observations. Cooling or no change in temperature predominated in the no and high cloud cases (100%), and in the variable cloud-base cases (>90%). Cooling or no change in temperature, however, was also often present in the low cloud cases when precipitation was not occurring at dense fog onset (∼70%), but was less frequent when precipitation was present (38%). When precipitation was occurring at onset, temperatures often increased (62%). Overnight, dewpoint temperature was often observed to increase or remain constant in cases where precipitation was occurring at dense fog onset (∼90%), and when lows or fronts were present (∼75%). This overnight increase, or constant dewpoint temperature, suggests that warm moist advection and/or the occurrence of precipitation preconditioned the atmosphere for fog formation. In addition, it should be noted that wind profiles exhibiting a veering (suggesting low-level warm air advection) or no obvious shear pattern predominated; no matter the cloud-base height, the synoptic type, or the height of the moist layer at 1200 UTC (0600 LT). Backing occurred infrequently and was not associated with any particular synoptic type, moist layer height, or synoptic type.
It was suggested by Croft et al. (1997) that radiational cooling was important in both radiation and coastal advection fog cases. They found T − Td differences of ∼20°C above the layer of dense fog in their study cases (excluding cases where precipitation occurred at dense fog onset), and a rapid warming and drying above the fog layer (i.e., “goal post” structure). In the Peoria fog sample, approximately 60% of the soundings showed a difference of 20°C or more above the near-surface moist layer at 1200 UTC. A T − Td difference of 20°C or more was found for 75% of no cloud or high cloud-base events and for 70% of nonprecipitating low cloud-base cases. Only 35% of precipitating low cloud-base cases showed a 20°C temperature difference. When cooling rates at the surface and at 950 hPa were examined over the 12-h period with regard to the presence of a drying region, no discernable difference could be found within synoptic or cloud-base categories. This may not be surprising, as over the 12-h period between soundings, temperature change rates could result from changes in air mass, from storm outflows, and advection, as well as the rate of radiation flux divergence, or other processes. Nearly all of the 1200 UTC soundings (95%) had a T − Td difference of at least 12°C above the fog layer.
Precipitation fogs were described by George (1940), Byers (1959), and recently by Tardif and Rasmussen (2008). Light precipitation often occurs at the onset of dense fog events (Westcott 2005; Tardif and Rasmussen 2007, 2008). This precipitation may be from deep clouds, or the result of settling from shallow clouds. While the available datasets do not allow for determination of precipitation fog processes, the current study quantifies the frequency of occurrence of regional precipitation and thus, the potential for precipitation to precondition the atmosphere for later fog development (Westcott 2004; Westcott 2005; and for the New York City region by Tardif and Rasmussen 2008).
The Peoria climatology indicated that some 71% of all dense fog events occurred where precipitation had been observed within the previous 24 h (Table 5). Precipitation generally occurred during the prior 24 h for low pressure (∼95%) and frontal system (81%) synoptic categories, and more than half of low pressure or frontal system events had precipitation at the time of dense fog onset. In most cases (76%) 24-h precipitation totals were small (<0.63 cm, <0.25 in.), and in only four cases were totals more than 2.54 cm (1 in.). When precipitation was observed at or near dense fog onset, it was almost always light (≤0.25 cm, 95%), and usually unfrozen (86%).
The frequent presence of regional precipitation suggests that it may aid in preconditioning the low levels of the atmosphere for fog development, and thus would have an important impact on forecasts of fog occurrence and coverage.
4. Case study
To illustrate how some of the events with low clouds and prior precipitation may give rise to dense fog, as well as limitations in operational observations for understanding and predicting fog formation, the widespread dense fog event of 6–7 November 2006 is explored. In this event, dense fog developed over much of eastern Illinois. This event exhibited many of the features found to be common in the climatological analyses, such as small temperature changes prior to fog onset, a warm–dry layer above the low clouds, and precipitation prior to dense fog development.
a. Synoptic-scale processes
During the 18-h time period preceding the development of widespread dense fog, the atmosphere was preconditioned by the vertical redistribution of moisture and horizontal advection into the region by processes associated with an approaching cyclone. A weakening area of surface low pressure moved from Oklahoma to northern Kentucky from 6 to 7 November 2006, transporting a deep layer of moisture northward and northwestward into the relatively dry Illinois region. As surface winds turned from southwesterly to southeasterly at lower altitudes, precipitation developed to the south and east of the study region. As the cyclone departed, marked drying was evident at higher levels. Rawinsonde observations that were taken at Lincoln, (Fig. 4) show changes in the atmospheric column that occurred as a result of these processes.
During the nighttime hours of 6–7 November, an important redistribution of moisture in the atmosphere took place. Large increases in atmospheric moisture from 1200 UTC 6 November to 0000 UTC 7 November (0600–1800 LT on 6 November) were found between about 3- and 10-km height (Fig. 4). Somewhat smaller increases were observed below about 1.5 km. Despite the increase in moisture at higher altitudes, air below about 850 hPa was unsaturated at 0000 UTC (1800 LT).
Observations from the Lincoln, WSR-88D site (KILX, Fig. 5), indicated that precipitation moved northward into the region at altitudes between about 3 and 10 km. Widespread precipitation was observed in this layer over much of central Illinois, which evaporated as it fell into the subsaturated air below. Precipitation reached the surface about 40–60-km south of Peoria and briefly in Champaign. Evaporation of precipitation, as well as continued weak moisture advection from the southeast, allowed for important increases in relative humidity. During the period leading to the onset of precipitation, ceilometer observations at many central Illinois ASOS/AWOS sites indicated a lowering cloud base that was accompanied by falling surface air temperatures and rising dew point temperatures and relative humidities. An example of this is presented for KCMI (Fig. 6). As the precipitation and deep cloudiness moved northeastward and away from the region, considerable drying of the atmosphere above ∼2 km took place (Fig. 4).
b. Evolution of weather conditions and dense fog development
Fog formed in the early morning of 7 November 2006 in an area of overcast skies and with an absence of cooling at the surface. Figure 7 shows the area where visibilities of 400 m (¼ mi) or less were observed during the early morning hours of 7 November, overlaid on an IR satellite image from 0445 UTC (2245 LT). This shows that the area of eventual dense fog development was located in the wake of, and was closely oriented with, a departing widespread area of deep clouds associated with the low pressure system. Fog began to develop in south–central Illinois between 0800 and 0900 UTC (0200 and 0300 LT) and developed to the northeast throughout Illinois between 1100 and 1400 UTC (0500 and 0800 LT). Most dense fog dissipated between 1400 and 1530 UTC (0800 and 0930 LT), except at sites close to Lake Michigan. Sunrise was at 1230 UTC (0630 LT) on this day.
During the evening of 6 November, little or no precipitation was observed at the surface in the northern and western portions of the state, while continuous light precipitation was observed in the southern third of Illinois. After about 0600 UTC (0000 LT), precipitation became much lighter and more scattered. It was observed at several locations in east–central Illinois until about 1000 UTC (0400 LT). Virtually no precipitation was indicated by the WSR-88D in the central Illinois fog region after 1000 UTC (0400 LT). While precipitation occurred overnight at many of the sites that later reported fog, no precipitation was reported within 4 h of the fog formation or during the fog event at these sites.
In the hours after the movement of precipitation across the region and before initial development of widespread dense fog, surface relative humidities in excess of 90% and low cloud bases were observed over a very large region, in Illinois and surrounding states. Despite this, fog only developed in a rather narrow band stretching much of the length of the eastern and central regions of Illinois (Fig. 7). Examination of the evolution of surface and higher-level atmospheric conditions both in and out of the area of dense fog formation gives important clues as to the processes that may have led to fog.
c. Possible processes related to the development of dense fog
In order for fog to develop in the region after 1200 UTC (0600 LT), atmospheric moisture must have remained at near-saturated levels, and a mechanism was needed to cause the air to become supersaturated. Examination of the surface temperature field and wind velocities reveals that modest warm air advection was present over most of Illinois with the exception of a small area in southwestern Illinois. Surface dewpoint and wind fields indicate that most regions had weak positive moisture advection, except for a few areas with negative moisture advection in far southern and far northern Illinois.
In the absence of any other processes (i.e., vertical mixing, condensation on the surface, etc.), the balance between moisture advection and changes in the amount of moisture needed for saturation would determine temporal trends in the surface relative humidity field. Figure 8 shows the difference in horizontal moisture and saturation moisture fluxes (determined by temperature advection) at 1200 UTC (0600 LT), near the time of widespread dense fog development. Regions with negative difference values, where less moisture was advected than needed to keep the air saturated, are seen in much of northern and western Illinois and in a small portion of southeastern Illinois. A comparison between the remaining areas of positive differences shown here, and the overall area of dense fog development shown in Fig. 7, show reasonably good correspondence despite the small advection values. It should be noted that calculation of small differences between moisture and temperature advection with available operational data will have significant errors. Such errors could result in local changes in the sign of the differences. However, the overall pattern of horizontal winds, temperature, and moisture fields suggest that the presence of a narrow band of positive values in the region is a reasonable conclusion.
While moisture advection can give rise to saturated conditions, another process, such as mixing between two saturated air masses of different temperatures or cooling, must be present to allow fog to form. In this case, weak shear was evident in the Lincoln, sounding near the top of the moist layer. Weak and variable winds within the moist layer, however, suggest that shear-driven turbulent mixing was unlikely to have been an important process on this date. Theoretically, horizontal mixing of air masses could also give rise to regions of supersaturation. However, no mechanism has been proposed to allow for a region of horizontal mixing on the size scale of Illinois.
Supersaturated conditions also could be generated through cooling of the near-surface air relative to advective warming by the presence of external heat sinks. Based on horizontal sensible heat fluxes (used in development of Fig. 8), surface temperatures would be expected to increase by 0.1°–0.5°C throughout most of the region with dense fog formation. However, temperatures remained constant or decreased at all sites in the dense fog region between 0400 and 0700 LT. In central Illinois, where low clouds were observed, the decreases were 0.1°–0.5°C. In far northwestern Illinois, where no deep clouds had been present and where horizontal moisture advection was negligible, temperature decreases were on the order of 1.5°–2.0°C where fog developed.
One possible source for cooling in central Illinois is from the ground surface. Throughout Illinois, observations of ground temperatures are obtained hourly by the ICN, allowing for an estimate of whether regional surface fluxes were negative. At 1200 UTC (0600 LT), observed ground temperatures were the same as or warmer than surface (2 m) air temperatures for nearly all ICN sites within the region where dense fog formed. Observations at the AmeriFlux sites in the vicinity of Champaign, Illinois, were somewhat inconclusive. The two sets of instrumentation indicated weak downward heat fluxes at 10 m AGL (<5 W m−2 within 6 h of fog formation). However, they differed on whether the air was warmer than the surface (but both with <1°C difference in air and ground temperatures). The heat flux 10 cm below the ground surface was directed upward (∼5 W m−2) at the one site taking these observations. Given the tendency for air temperatures to be near or colder than the ground temperatures in most of the region of fog development, and small and inconsistent estimates of surface-air heat exchanges at the AmeriFlux sites, it seems unlikely that cooling from the surface played an important role in fog development in this case.
An alternative source of cooling is radiation divergence between cloud top and the surface. As seen in both the satellite image (Fig. 7) and soundings at Lincoln, (Fig. 4), deep clouds departed the region during the early morning hours of 7 November, leaving behind a shallow cloud layer of approximately 2-km depth. This gave rise to between 2 (far eastern and southern Illinois), 6–8 (east–central Illinois), and 12 (northwestern Illinois) nighttime hours without deep clouds in the region of dense fog formation. Since the cloud layer was at the lowest measureable altitude for 4 h before dense fog formation, and was therefore apparently linked to the surface, it might be speculated that convective motions driven by cloud-top radiative processes could distribute the cooling throughout the layer as suggested by previous studies.
As a fog layer forms, the level of maximum radiation flux divergence is known to migrate to the top of the fog layer (e.g., Jiusto and Lala 1980; Brown 1987; Fitzjarrald and Lala 1989). Similarly, a maximum in net radiation flux divergence has been found at the top of stratiform cloud layers (e.g., Caughey et al. 1982; Slingo et al. 1982; Frish et al. 1995; Nakanishi 2000). Curry (1986) demonstrated that negative net radiation fluxes can penetrate downward, well into shallow stable cloud layers. While radiational cooling is likely important in fog development and maintenance, with an existing cloud layer, it often is not observable at the surface. On 7 November, the net radiation flux measured at the AmeriFlux site was negligible in the 3 h prior to and at dense fog onset. The cloud layer and weak warm air advection likely masked any radiational contribution toward cooling at the surface.
Drying of the atmosphere above about 800 hPa (Fig. 4), with the departure of deep clouds in satellite imagery (Fig. 7), would allow for radiational cooling at the top of a fog (e.g., Jiusto and Lala 1980; Brown 1987; Fitzjarrald and Lala 1989; Croft et al. 1997). With vertical mixing, radiational cooling at fog or cloud top may generate supersaturated conditions (Jiusto and Lala 1980). In stratus clouds, buoyancy fluctuations may occur when the mixed layer beneath is destabilized as radiationally cooled air from above sinks and is replaced by warmer air from below (e.g., Caughey et al. 1982; Frish et al. 1995; Nakanishi 2000). Fluctuations in wind speed, direction, and momentum flux noted in the surface data (Fig. 6) suggest that this process may be acting. However, the exact cause of the fluctuation in winds could not be determined.
Without additional measurements above the surface, including information on the cloud structure and depth, the exact mechanism leading to dense fog when clouds are present is ambiguous. As noted by Curry (1986) and others, the heat budget of low-level clouds is very complex. The profile of all sources and sinks of heat is required to understand the balance of radiative cooling, latent heat exchanges, entrainment, and surface heat fluxes.
Regardless of the specific mechanism, it is hypothesized that cooling of the layer between the top of the low clouds and the surface, in combination with a region of favorable moisture advection, allowed the near-surface air to cool at a rate slower than expected because of warm air advection, but sufficient to allow a wide region of supersaturation and dense fog to develop. Qualitatively, this is consistent with the close correspondence between the southwest–northeast orientation of the fog region and the western edge of the departing deep clouds.
Cold-season fogs in the Midwest are often associated with low pressure systems or fronts. Based on a climatology of dense fog events in Peoria, Illinois, the majority of these continental fog events form when precipitation has occurred in the region and sometimes when it is still occurring, and when cloud bases in the 6 h prior to dense fog formation are less than 1000 m. When fog forms in the presence of a low cloud base, an unchanging or increasing surface air temperature is not uncommon. The processes by which saturation and supersaturation occur in these cases are unclear. Despite the frequency of these prefog conditions, little quantitative work has been reported on low cloud-base continental fog events. The variability found in the frequency of surface changes, precipitation occurrence, and synoptic classifications suggests that low cloud-base fogs form under a variety of conditions. Clearly, further case studies would be invaluable in determining the processes by which supersaturation occurs in these cases.
The 6–7 November 2006 case study observations fit well within the Peoria climatology of dense fog events associated with low cloud bases and with low pressure systems. For this event, precipitation and continued moisture advection associated with a nearby cyclone played a critical role in lowering of the cloud base. Precipitation developed above about 3 km over a wide region of central Illinois, but based on ceilometer, radar, and surface observations, evaporated before reaching the surface for several hours. The preconditioning helped saturate the lower atmosphere prior to fog formation.
After precipitation ended, the surface temperature remained nearly constant, and the relative humidity was saturated or nearly saturated. AmeriFlux observations taken near Champaign, Illinois, suggest that surface sensible heat fluxes were small, and results from various other methods of estimating the direction of heat fluxes were inconsistent. Therefore, processes at the surface likely played a minimal role in fog development in this case. With small differences between the ground and the 2-m air temperature and the low wind speeds, this would not be considered an advection fog event in any case. The drying and clearing of the upper-level cloudiness following precipitation and prior to fog formation, however, suggests that radiational cooling at cloud top played an important role leading to supersaturation and the development of fog.
Thus, it is hypothesized that the primary mechanism for cooling of the surface layer relative to sensible heat advection was radiative cooling of the cloud and subcloud layer. However, verification of this hypothesis is not possible with the datasets available on this day. With upper-air observations available only at 0000 and 1200 UTC (1800 and 0600 LT), timing of the changes in the cloud, temperature, and moisture structure is unknown. A radiative transfer model could shed further light on processes involved in fog formation, and would be most useful if supported by intervening nighttime soundings, a profiling radiometer, above-cloud aircraft observations, and/or a cloud radar to document the changes of the temperature and moisture fields, and the presence and depth of clouds aloft.
The authors thank Robert Scott at the Illinois State Water Survey for providing ICN observations, and Carl Bernacchi for providing the AmeriFlux observations. We also thank the personnel at the National Weather Service Forecast Office at Lincoln, Illinois, and in particular James Auten for useful discussions and Dan Kelly for providing high-resolution rawinsonde data. We also appreciate the advice of James Angel, Michael Palecki, and three anonymous reviewers, which led to important improvements in this manuscript. This research was partially supported by NOAA Cooperative Agreement NA67RJ0146. Any opinions, findings, conclusions, or recommendations are those of the authors and do not necessarily reflect the views of the Illinois State Water Survey or NOAA.
Corresponding author address: Nancy E. Westcott, Illinois State Water Survey, 2204 Griffith Dr., Champaign, IL 61801. Email: email@example.com