FOG 
entirely lacking. Probably entirely new techniques will 
be required to resolve this question satisfactorily. 
The formation of fogs over snow cover at very low 
temperatures probably can be adequately described 
from available knowledge. However, it would be de- 
sirable to have confirming data as to whether ice 
erystal fogs form over regions remote from human 
settlements and whether these settlements cause the 
formation of fog by the addition of otherwise absent 
sublimation nuclei or in some other way. 
The vertical gradient of water vapor in the lowest 
thousand feet of the atmosphere is something which 
almost all authorities treat as of considerable impor- 
tance in the formation of fog. Yet quantitative knowl- 
edge of the subject is limited. Some pertinent questions 
concerning vertical moisture gradients are (1) Exactly 
what kind of gradient is required for the formation of 
various densities of fog? (2) How high above the sur- 
face do moisture inversions usually extend during fog 
formations? (8) What rate of flux attends various 
stability conditions, wind flow, ete.? 
It should not be too hard to answer some of these 
questions at least in part by using existing observation 
towers with perhaps simple additions to equipment, 
although the location of these towers with respect to 
sufficient quantities of fog appears doubtful. It would 
seem that such observational data might easily be a 
key not only to broader concepts of the physical process 
of fog formation but to better methods by which to 
forecast it. 
Synoptic Aspects of Fog 
Practically all treatises on fog begin with the author’s 
idea of a classification of fog, usually according to 
causes. Most such classifications are eminently logical 
and serve the purpose for which their author intends 
them adequately and well, and they should be con- 
sidered according to the user’s needs. Willett’s table 
[25], slightly modified by Byers [2, p. 509], is given 
below and serves as an adequate model, despite the 
fact that the original version appeared as long ago as 
1928. 
A. Air-mass fogs. 
1. Advection types. 
a. Types due to the transport of warm air over a 
cold surface. 
(1) Land- and sea-breeze fog. 
(2) Sea fog. 
(8) Tropical-air fog. 
b. Types due to the transport of cold air over a 
warm surface. 
(1) Steam fogs (arctic “sea smoke’’). 
2. Radiation types. 
a. Ground fog. 
b. High-inversion fog. 
3. Advection-radiation fog (radiation over land in 
damp sea air). 
4. Upslope fog (adiabatic-expansion fog). 
B. Frontal fogs. 
1. Prefrontal (warm-front) fog. 
1183 
2. Postfrontal (cold-front) fog. 
3. Front-passage fog. 
In addition to the classes listed by Willett, the following 
types appear in Petterssen’s scheme [19]: (1) isobaric 
fog, (2) isallobaric fog, and (3) fog formed by horizontal 
mixing. All of them (except the last one, under special 
conditions) are designated by the author as rather 
unimportant. 
If a classification such as these is to be used for 
forecasting, then it is not only desirable but necessary 
to eliminate all of the causes listed which are not of 
direct forecasting value. If this simplification is not 
made, the forecaster must continually concern himself 
with rules for the forecasting of various kinds of fog 
whose causes overlap and which cannot be anticipated 
by any clear-cut methods. Furthermore, in the case of 
many types of fog the frequency of formation is so low 
that any attempt to forecast their occurrence results 
in continual errors on the side of forecasting a phe- 
nomenon which does not occur. 
Tf all of the nonessential types of fog (for practical 
purposes) are eliminated or modified, only the following 
classes remain: 
A. Air-mass fogs. 
1. Advection types. 
a. Sea fog. 
2. Radiation types. 
a. Restricted heating fog. 
6. Air-drainage, including marsh fog. 
3. Advection-radiation fogs. 
B. Frontal fogs. 
1. Pre-warmfrontal fog. 
2. Mixing-radiation fog. 
This list is restricted to only six categories of fog, 
compared with the eleven given in the Willett-Byers 
classification; fourteen, if the extra types described by 
Petterssen are added. If the reduction (which is usually 
made in forecast offices whether or not it is officially 
recognized) is justified, it simplifies the forecasting 
problem. Accordingly, a brief consideration of the cate- 
gories modified or eliminated seems desirable. 
Land- and Sea-Breeze Fog. This type of fog is con- 
ceded by most authorities to be essentially the same as 
sea fog. Furthermore its occurrence is distinctly un- 
usual except perhaps in restricted areas of certain sea 
coasts. Its influence is seldom felt more than two or 
three miles inland. For these reasons it does not seem 
that fog of this type merits a separate forecasting 
division except possibly in very special instances which 
are better handled as exceptions. 
Steam Fogs. Although these fogs, which occur over 
open water in very cold temperatures, are extremely 
interesting, they are always shallow and usually do not 
restrict the visibility to very low values. Furthermore, 
they are confined mostly to the edges of open water. 
Byers ascribes them entirely to a difference in vapor 
pressure between the open water and the air above it 
of the order of 5 mb. Again, the basic requirement of 
forecast need is not likely to be present fo: this type of 
fog and it is therefore omitted. 
Upslope Fogs. Almost every textbook on meteorology 
