SR/SSD 98-19

Technical Attachment

Brian Kyle and Carolyn Levert
NWS Office, Houston/Galveston, Texas

1. Introduction

The Houston/Galveston National Weather Service Office assumed aviation forecast responsibility for southeast Texas on January 1, 1996. A new challenge was presented to the NWSO forecasters by this transition -- developing the skill to specifically and accurately forecast categories of low cloud heights and visibility.

Airplane accidents due to low ceilings and visibility claim more lives per year than any other cause, including engine trouble. A pilot with a visual-only rating flying into IFR conditions can expect to live for an average of 28 seconds (Hoffman 1996). Pilots not acquainted with "flying blind" have a tendency to believe their own sensations of motion rather than their instruments (Hoffman 1996). Low clouds and fog are also a major factor in airport delays. In 1993, 74 percent of all delays were caused by weather, with the majority of the delays caused by low ceilings and visibility. For a major hub such as Houston Intercontinental (IAH), these delays cost the airlines close to $24 million per year (NWSTC 1996).

The purpose of this paper is to analyze sounding data to investigate possible correlations between low-level (surface to 700 mb) relative humidity and wind, and low ceilings and fog. Such correlations could then be used to improve forecasts for the aviation community.

2. Data and Procedure

This study examines a six-month period from October 1992 to March 1993. Historically, early fall through spring is the period with the greatest concentration of fog events in southeast Texas. Since Houston is not an upper air station, sounding data from Corpus Christi and Lake Charles were used to estimate conditions for the Houston area. This sounding data was extracted from the NCDC CD-ROM "Radiosonde Data of North America: 1946-1995, Version 1.0." Both 0000 UTC and 1200 UTC soundings were used to find relative humidity and wind data for the surface to 700 mb.

Each sounding was separated into four layers: surface, 1000-925 mb, 925-850 mb, and 850-700 mb. Average humidities, wind directions, and wind speeds were found for each layer. Then, corresponding layers from Corpus Christi and Lake Charles were averaged to find an estimate of the upper-air conditions at Houston.

The daily surface observations at Houston (IAH) were used to find days on which fog occurred, and the conditions under which it occurred. Days with LIFR and IFR fog were examined in this study because these are the conditions which have the greatest impact on the aviation community. The table below shows the days during each of the months studied on which various fog conditions (or no fog) occurred. Of the days studied, 41 had LIFR conditions, 32 had IFR conditions, and 78 had no fog at all.

LIFR Conditions IFR Conditions No Fog
October 1992 6, 10, 17, 21, 22, 27-29 5, 7 1-4, 8, 11-15, 19, 25, 26, 30
November 1992 10, 11, 17, 18 1, 19, 20, 24, 30 2-8, 12-14, 16, 22, 23, 25-28
December 1992 1, 2, 4, 9, 18-23, 27-31 3, 5, 6, 14, 25, 26 10-13, 16, 17, 24
January 1993 8, 9, 10, 17-19 11, 12, 16, 20, 21, 29 5, 13-15, 22, 25-27, 30
February 1993 9, 19 5, 10, 11, 24, 25 1-3, 8, 12-18, 21-23, 26-28
March 1993 1, 2, 21, 22, 25, 29 11, 16, 18, 20, 23, 24, 28, 30 3-9, 13-15, 17, 26, 27, 31

Table 1. Dates used

Category Visibility Ceiling
LIFR <1 mi. <500ft.
IFR 1 mi.<vis<3 mi. 500 ft.<cig<1000 ft.
MVFR 3 mi.<vis<5 mi. 1000 ft.<cig<3000 ft.
VFR <5 mi. <3000 ft.

Table 2. Aviation categories

For each month, all days with LIFR conditions were averaged to find relative humidities, wind directions and speeds at the different layers for both 0000 UTC and 1200 UTC. This was also done for IFR conditions and days with no fog. Then the entire six-month period was averaged in the same manner.

3. Observations and Analysis

After studying 73 days on which fog was reported during the six-month study period (October 1992-March 1993), we conclude that differences in wind direction, wind speed, and relative humidity play important roles in determining whether or not LIFR and IFR conditions occur.

Surface to 850 mb mean relative humidity was in the 58-63 percent range for no fog cases, 74-82 percent for LIFR conditions, and 72-81 percent for IFR conditions. Humidities of at least 70 percent in the bottom three layers studied were present in both the 0000 UTC and 1200 UTC soundings for the fog cases. LIFR events tended to occur at higher temperatures and seemed less dependent on whether there was recent rainfall, in comparison with IFR events.

Low level wind speeds, especially in the 925 mb to 850 mb layer, were much stronger on the days with fog than without fog. Interestingly, worse conditions (LIFR) occurred with higher speeds in this layer in the 1200 UTC sounding. On average, fog with imbedded LIFR conditions lasted four hours longer than fog with imbedded IFR conditions. Most notably, there was east and southeast low level flow on days with fog, and northeast and northwest flow on days without it.

The amount of moisture from the surface to 850 mb seems to be a critical factor in determining fog formation. This is in contrast to the previously recommended (NWSTC 1994) 850 to 700 mb layer used to identify moist layers and warm/cold advection. Aviation forecasters along the upper Texas coast might begin to anticipate fog formation (IFR or LIFR conditions) when the 0000 and 1200 UTC soundings indicate relative humidities greater than 70 percent up to 850 mb, especially with southeasterly wind components. Comparing IFR and LIFR conditions, on average, moisture in the lowest two levels (surface to 925 mb) - although not dramatically different - does seem to indicate that LIFR conditions occurred with slightly greater relative humidity up to 925 mb, both at 0000 and 1200 UTC.

During fog events, soundings indicated a southeasterly and southerly flow throughout the lower levels of the atmosphere always veering with height. This indicates warm air advection along with moisture transport from the Gulf of Mexico. Surprisingly, low level wind speeds were greater in cases with fog than those without. Wind speeds greater than 5 kt from the surface to 925 mb generally supported low ceilings and fog, while the reverse was true for speeds less than 5 kt. This most likely indicates fog along the upper Texas coast is most frequently due to advective rather than radiative processes. This makes sense, especially during the six months studied, because of the warm and moist air from the Gulf moving over cooler land. Advection fog deepens as wind speed increases up to about 15 kt (USDOT 1975). Wind much stronger than 15 kt lifts the fog into a layer of low stratus or stratocumulus.

Radiation fog, on the other hand, is associated with wind speeds around 5 kt or less and relatively high humidities. It is not uncommon for this type of fog to occur after rain has soaked the ground. The majority of IFR conditions may be associated with radiation fog as it is more dependent on recent rainfall and lower wind speeds.

4. Summary

Low ceilings and fog should be forecast if the mean relative humidity from the surface to 850 mb is greater than about 70 percent, and east to southeast winds greater than about 5 kt at the surface veer and increase in speed with height. There were subtle differences in IFR/LIFR conditions, most likely dependent on the amount of moisture available at the lowest levels.

Any individual month covered in this study probably does not include a sufficient number of events to allow definitive statements, compared to data compiled and averaged over several years. Additional studies on fog at other locations in southeast Texas would be useful in aiding forecasters with understanding elements necessary for fog formation. This information could be used to provide the aviation community with better forecasts.


The authors would like to thank Steve Allen (NWSO Houston SOO) for his assistance in preparing and managing the sounding data. Also, the input from Bill Read (NWSO Houston MIC) and Dan Smith (SRH) was a great help and is deeply appreciated.


Hoffman, Ted, 1996: Forecasters Development Course Lecture. NWS Training Center.

NWS Training Center, 1994: Aviation Forecasting. NWS Training Center, 54 pp.

NWS Training Center, 1996: Terminal Aerodrome Forecasts (TAF). NWS Training Center, 49 pp.

U.S. Department of Commerce, 1996: Radiosonde Data of North America 1946-1995. National Climatic Data Center, Version 1.0.

U.S. Department of Transportation, 1975: Aviation Weather. U.S. Government Printing Office, 219 pp.