J. C. Rodda: On The Problems Of Assessing The World's Water Resources
25
true value is unknown and even a best approximation, such as might be obtained
in the laboratory from a series of repeated measurements, is not often available.
This has led to a preference for the use of the term uncertainty - the interval about
the measurement within which the true value of the quantity can be expected to be
with a stated probability ([46]). There are many different sources of error in
hydrological measurements and Fig. 2 shows how these may arise in the practice
of flow measurement with a weir or a flume. The overall uncertainty of
measurement XQ ([47]) depends on the standard of construction of the gauging
structure, the correct application of the design specifications and a number of
other factors and XQ will also vary with discharge.
Herschy ([48], [49]) discussed the errors of the various methods of flow
measurement drawing on material published by the International Organization for
Standardization (ISO) and also by WMO. He concluded that if ISO standards are
followed, then flow measurements made at a single gauge are expected to have the
following upper limits of uncertainties at the 95% confidence limits:
Single determination of discharge
15 minute average of discharge
Daily mean, monthly mean or annual discharge
7%
5%
5%
It would be most unwise to assume that river flows generally are measured to these
limits: many countries are not able to meet ISO Standards ([50]), nor the
requirements of WMO ([51]). Neither do the measurements of the other variables
satisfy the WMO recommendations for accuracy (Table VII) that have been
adopted by the WMO Commission for Hydrology. Unfortunately, these are
desired accuracies rather than those being achieved at present: indeed the actual
measurement errors must be somewhat larger than those stated. Take precipitation
as an example: errors can range from 5 to 30% for rain and 10 to 80% for snow
depending on gauge type, site, wind speed and other factors.
6
REDUCTION OF ERRORS THROUGH WM 0
ACTIVITIES
WMO has been working in various ways to combat errors in instrument practice
and in the errors which can occur at later stages in data management. One of the
main vehicles for this work has been the series of instrument intercomparisons that
deal with the measurement of the main hydrological variables.
The initial comparison of liquid precipitation measurements selected one
particular reference gauge for the tests undertaken in the 1950s. When the
problems of this gauge were recognized, a different reference was chosen for the
second intercomparison carried out in the 1970s ([52]). For the third
intercomparison, which is of devices for measuring solid precipitation, and which
is concluding at the present time, a third interim reference was employed ([53]).
For these tests reference instruments were installed at a large number of different
sites around the world for comparison with the various national gauges. The
results of the different intercomparisons show how national gauges performed
against the reference.
25
true value is unknown and even a best approximation, such as might be obtained
in the laboratory from a series of repeated measurements, is not often available.
This has led to a preference for the use of the term uncertainty - the interval about
the measurement within which the true value of the quantity can be expected to be
with a stated probability ([46]). There are many different sources of error in
hydrological measurements and Fig. 2 shows how these may arise in the practice
of flow measurement with a weir or a flume. The overall uncertainty of
measurement XQ ([47]) depends on the standard of construction of the gauging
structure, the correct application of the design specifications and a number of
other factors and XQ will also vary with discharge.
Herschy ([48], [49]) discussed the errors of the various methods of flow
measurement drawing on material published by the International Organization for
Standardization (ISO) and also by WMO. He concluded that if ISO standards are
followed, then flow measurements made at a single gauge are expected to have the
following upper limits of uncertainties at the 95% confidence limits:
Single determination of discharge
15 minute average of discharge
Daily mean, monthly mean or annual discharge
7%
5%
5%
It would be most unwise to assume that river flows generally are measured to these
limits: many countries are not able to meet ISO Standards ([50]), nor the
requirements of WMO ([51]). Neither do the measurements of the other variables
satisfy the WMO recommendations for accuracy (Table VII) that have been
adopted by the WMO Commission for Hydrology. Unfortunately, these are
desired accuracies rather than those being achieved at present: indeed the actual
measurement errors must be somewhat larger than those stated. Take precipitation
as an example: errors can range from 5 to 30% for rain and 10 to 80% for snow
depending on gauge type, site, wind speed and other factors.
6
REDUCTION OF ERRORS THROUGH WM 0
ACTIVITIES
WMO has been working in various ways to combat errors in instrument practice
and in the errors which can occur at later stages in data management. One of the
main vehicles for this work has been the series of instrument intercomparisons that
deal with the measurement of the main hydrological variables.
The initial comparison of liquid precipitation measurements selected one
particular reference gauge for the tests undertaken in the 1950s. When the
problems of this gauge were recognized, a different reference was chosen for the
second intercomparison carried out in the 1970s ([52]). For the third
intercomparison, which is of devices for measuring solid precipitation, and which
is concluding at the present time, a third interim reference was employed ([53]).
For these tests reference instruments were installed at a large number of different
sites around the world for comparison with the various national gauges. The
results of the different intercomparisons show how national gauges performed
against the reference.