978 
Irrespective of how many stations there are or will be 
and how many elements are observed at each, we will 
never have complete and homogeneous coverage of the 
whole surface of the earth. Even in the densest net- 
Taste I. Some Curmatic ELements Requirine (A) 
ADDITIONAL OBSERVATIONS OR (B) BETTER OBSERVATIONS 
(A) Insufficient coverage with existing equipment (extent). 
Solar radiation: Total intensity, spectral distribution, vari- 
ous exposure angles, illumination, vertical 
variation in intensity, cloud and surface 
albedos. 
Weight of ice deposits, weight of snow ac- 
cumulations, dew, horizontal patterns of 
intensity (simultaneous), extent of snow 
cover, depth of snow, character of snow 
cover, vertical variations in precipitation 
intensity. 
Soil temperatures, depth of penetration of 
frost. 
Water temperatures. 
Frequency of lightning discharges to ground. 
Cooling power. 
Suspended particles (dust, condensation nu- 
clei). 
Soil moisture. 
Soil temperatures (vertical distribution). 
Soil and surface temperatures (horizontal 
distribution). 
Extent and character of sea, lake, and river 
ice. 
(B) Inadequate techniques or instruments. 
Evaporation. 
Corrosive substances in air. 
Composition of rain water. 
Drop sizes. 
Radiation losses. 
Soil conditions (state of ground). 
Slant visibility. 
Vertical component of wind. 
Cloud thickness and layering. 
Rainfall over oceanic areas. 
Evaporation from water and soil surfaces; rates of transpira- 
tion. 
Foliage temperatures. 
Cloud density; areal extent of cloud cover as a function of 
altitude of observer above or below cloud deck; cloud 
temperatures. 
Vertical temperature, moisture, and wind structure near 
ground. 
Rate of heat transfer between soil or water surface and atmos- 
phere (instantaneous rates). 
Precipitation: 
Other: 
works there remain blind spots. One of the important 
questions is therefore, How can we interpolate between 
stations? The synoptic reconstruction of an observation 
which is lacking, or the bridging of an observational 
gap, is an important technique. Given a series of good 
synoptic charts, the meteorological analyst can make a 
guess about the probable values for a given locality. 
This is not much different from making a forecast, 
except that the whole sequence of events is synoptically 
known. However, the reliability of such guesses has 
seldom been put to a controlled test. Hence the limita- 
tions and errors of the method are usually not available 
to the climatologist. Such information can be obtained 
by applying objective tests to representative analyses. 
The question of how to interpolate between stations 
also enters into the problem of the representativeness of 
individual climatic stations for various elements. It 
can, for example, be assumed that a mean pressure 
value and a frequency distribution of pressures for a 
CLIMATOLOGY 
station are valid for distances of many miles. The 
synoptic method can also be trusted to yield readily- 
corrections to these values over distances of more than 
a hundred miles. It is doubtful if such extrapolations 
can be made for other elements. Temperature adjust- 
ments have occasionally been attempted by making 
corrections for topography, but it is doubtful if any 
reliable extrapolations can be made for temperature, 
precipitation, and surface winds except, perhaps, over 
the simplest terrain forms such as the ocean. The rea- 
sons for this are (1) that present methods of measuring 
these elements constitute very poor sampling, and (2) 
that our present knowledge of terrain influence on 
temperature, wind, and precipitation is inadequate. 
A daily task in applied climatology is the substitution 
of an observed element for one that is desired but has 
remained unobserved. Such substitution can often le- 
gitimately be made because of known theoretical rela- 
tions or correlations which have been established from 
statistics. For example, it is certainly possible and 
permissible to obtain wind vectors in the free atmos- 
phere from a set of upper-level pressure charts. To use 
shelter temperatures for an estimate of frost at the soil 
surface is less reliable, but still within the capabilities 
of the experienced meteorologist. In many areas it is 
possible to translate mean cloudiness data into fre- 
quencies of clear and cloudy days [53]. On the other 
hand, very little faith can be placed in an interpretation 
of precipitation data in terms of limitations to flying 
conditions (ceiling and visibilities), although that has 
been tried on occasion. In this field there is a great, 
need for establishing correlations among various me- 
teorological elements at one point and of spatial cor- 
relations between one and the same element. 
In many practical cases the climatologist is con- 
fronted with the question: How long a period of records 
is needed to give an answer to a problem within an 
appropriate safety factor? There is no universal rule. 
In the past there has been a tendency to require 
decades of records. Many times this proves to be un- 
necessary in practice. Often suitable statistical tech- 
niques, evolved from the theory of sampling, can give 
quite satisfactory results on the basis of a small uni- 
verse. The question reduces to this: When does a fre- 
quency distribution become essentially stable, so that 
adding another year of observations would not add 
significantly to the result? 
Experience has shown that this time limit varies 
from element to element, season to season, and region 
to region. Preliminary investigations have shown that 
the orders of magnitude given in Table II may serve 
as a guide. 
The data given in Table II are very tentative. They 
are based on a few pilot tests [5]. The studies leading to 
this type of information should be expanded to verify 
the statements and to cover additional elements. 
We frequently encounter the following situation: A 
short record is available from a locality which is of 
interest. A long-record station is also located in the 
same region. The problem is to reduce the short to the 
long record. This question is dealt with in many stand- 
