Geoscience Reference
In-Depth Information
As discussed in detail below, this means that most of the rainfall and snowfall
data currently available are subject to systematic error.
Rainfall data measured
using gauges have been and still are systematically low, by about 5-10% on average
and the systematic errors in snowfall data are greater than this
. In general terms,
more precipitation is measured when gauges have the top of their funnel mounted
nearer to the ground. This is mainly because being nearer to the ground tends to
reduce the wind speed and hence wind-related gauge errors. However, splash-in of
rainwater to gauges nearer the ground from the surrounding area may also contribute
to a higher catch. In hydrological applications the systematic under-measurement of
measured precipitation is implicitly accommodated (usually without recognition) in
the value of rainfall-runoff coefficients, or in the parameters used in the models that
describe the relationship between precipitation and runoff.
Instrumental errors
Basic instrumental errors can occur even with a device as simple as a rain gauge.
The most obvious of these is the possibility that some of the collected water evapo-
rates before it is measured, this being most likely if measurement is as an average
value over long periods. In practice the now-usual design of gauges, which involves
a funnel feeding the collecting vessel via a thin tube, minimizes this error by main-
taining a more humid environment where the water is stored. Nonetheless, if the
period between measurements is very long, observers may choose to introduce a
known volume of buoyant oil which spreads across the surface of the water in the
container to inhibit evaporation. In periods of light rainfall, evaporation losses
from the layer of water that wets the funnel itself (sometimes known as
wetting
errors
) may be significant in percentage terms. In humid atmospheres, the oppo-
site problem can occur, with water condensed onto cold metal gauges to increase
the apparent rainfall. This latter problem is most likely to occur at high latitude
and coastal sites, but can be minimized by choice of funnel material. In the case of
gauges used for measuring rainfall intensity, timing errors can arise. Delays of up
to 10 minutes may occur with the
tipping bucket
design (discussed later) as the
bucket is filled, particularly for storms that occur after a long dry period.
Site and location errors
The representativeness of rainfall measurements can easily be affected by both
local characteristics within 10 m of the site where the gauge is located and by major
obstructions to the wind flow at some distance upwind of the gauge. The funda-
mental reason for errors is because the proportion of rain measured by a gauge,
the '
catch
, depends on the wind speed and direction immediately above the top of
the gauge. This in turn depends on obstructions nearby and at some distance.
Typically gauges that are mounted above the ground and exposed to wind miss
about 5 to 10% of the incident rain, depending on wind speed and direction of fall
