Systematic Error or Bias

➢ is a systematic difference, either positive or

negative, between the measured value and
the true value.

SYSTEMATIC ERRORS OCCUR:

1. Due to malfunctioning of instrument.
2. Due to difficult measuring conditions, such
as
1. Unsteady flow
2. Meandering and Bad location of
observation stations
3. Lack of knowledge of observer
✓ If the systematic error has a known value,
this should be accounted for appropriately
and error due to this source should be
considered zero.
Regarding precipitation, WMO 1987 (World Meteorological Organization)
listed the ff. errors for which adjustment needs to be made to get a
near accurate estimate of precipitation from a measured precipitation
report:

a) Error due to the systematic wind field deformation above the gauge
orifice
b) Error due to the wetting loss on the internal walls of the collector
c) Error due to evaporation from the container (generally in hot climates)
d) Error due to the wetting loss in the container when it is emptied
e) Error due to blowing and drifting snow
f) Error due to splashing in and out of water, and
g) Random observational and instrumental errors
• The net error due to blowing and drifting snow and due to splash in and
out of water can be either negative or positive, while net systematic
errors due to the wind field and other factors are negative.
• Since for liquid precipitation the errors listed at (e) and (f) above are near
zero, the general model for adjusting the data from most gauges takes
the form
Pk = K (Pg + ∆P1 + ∆P2 + ∆P3)
where:
Pk = adjusted precipitation amount
k = adjustment factor for the effects of wind field deformation
Pg = the measured amount of precipitation in the gauge
∆P1 = adjustment for the wetting loss in the internal wells of the
collector
∆P2 = adjustment for wetting loss in the container after emptying
∆P3 = adjustment for evaporation from the container
➢ vary in an unpredictable manner, both
in magnitude and sign, when repeated
measurements of the same variable are
made under the same conditions.

➢ cannot be eliminated, but their impacts

can be reduced by repeated
measurements of the variable.

➢ These are equally distributed about the

mean or ‘true’ value.
THE OTHERS RANDOM ERRORS WHICH COULD BE DUE TO OBSERVER INCLUDE:
i. misreading and transposing digits,
ii. misrecording because of faulty memory,
iii. recording the data at the wrong place on the recording sheet,
iv. making readings at improper interval,
v. incorrect dating of the report,
vi. incorrectly reading or communicating the data to a reporting
center, etc.

EVEN WITHOUT HUMAN INTERVENTION CHANCES OF ERRONEOUS READING IN

CASE OF PRECIPITATION MAY BE POSSIBLE BECAUSE OF:
i. evaporation from gauge,
ii. overflowing gauge,
iii. mechanical or electrical mal-functions.
➢ These arise due to human mistakes or instrument
malfunction or some abnormal external cause.
➢ Reported data appear to be clearly in error.
✓ Sometimes the errors become obvious,
For example:
• Wrong placement of decimal and the
data can be easily corrected in such
cases.
✓ In other cases, the concerned
measurements may have to be discarded.
For example:
• An animal may drink water from the
evaporation pan and introduce errors in
the data.
✓ Sometimes, such errors may be readily
detected but it may not be easy to correct
them.
 ( . .)
SOME TYPICAL SOURCES OF ERROR WERE GIVEN BY (WMO 2008):
1. Datum or Zero Error
• originates from the incorrect determination of the reference point of an instrument
2. Reading (or observation) Error
• results from the incorrect reading of the value indicated by the measuring instrument.
This error is normally attributed to neglect or incompetence of the observer.
3. Interpolation Error
• is due to inexact evaluation of the position of the index with reference to the two
adjoining scale marks between which the index is located.
4. Error due to wrong assumption or neglect of one or more variables needed to determine
the measured value.
5. Hysteresis
• the phenomenon in which the value of a physical property lags behind changes in
the effect causing it
6. Insensitivity Error
arises when the instrument cannot sense the small change in the variable being
measured
 ( . .)
7. Non-linearity Error
• is that part of error whereby a change of indication or response departs from
proportionality to the corresponding change of the value of the measured quantity
over a defined range
8. Drift Error
• is due to the property of the instrument in which its measurement properties change
with time under defined conditions of use
9. Instability Error
• results from the inability of an instrument to maintain certain specified metrological
properties constant
10. Out-of-range Error
• is due to the use of an instrument beyond its effective measuring range, lower than
the minimum or higher than the maximum value of the quantity, for which the
instrument/installation has been constructed, adjusted, or set
11. Out-of-accuracy Class Error
• is due to the improper use of an instrument when the minimum error is more than the
tolerance for the measurement.
➢ Many hydrological variables are not directly measured but are
estimated form the measured values of several variables.

WE CAN ESTIMATE THE UNCERTAINTY E Q BY APPLYING THE GAUSS ERROR

TRANSFER THEOREM (WMO 2008):
2 𝜕𝑄 2 𝜕𝑄 2 𝜕𝑄 2
𝑒𝑝 = 𝑒𝑥 + 𝑒𝑦 + 𝑒𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧
where:
𝜕𝑄 𝜕𝑄 𝜕𝑄
, and are the partial differentials of the function expressing
𝜕𝑥 𝜕𝑦 𝜕𝑧
explicitly the relationship of the dependent variable with the independent
variables.
 ( . )
▪ Errors may also arise in data entry, during computations and (hopefully very rarely),
from the mistaken ‘correction’ of ‘right’ data.
▪ Data validation is the means by which data are checked to ensure that the corrected
values are the best possible representation of the true values of the variable.
▪ procedure includes primary and secondary data validation
▪ Validation of hydrologic data must never be considered as a purely statistical or
mathematical exercise. Staff involved in it must have a background in hydrology and
must understand the field practices.
▪ To understand the source of errors, one must understand the method of
measurement or observation in the field and the typical errors of given instruments
and techniques.
BASICALLY, DATA VALIDATION IS CARRIED OUT:
▪ to correct errors in the observed values where possible,
▪ to assess the reliability of data even though it may not be possible to correct errors,
and
▪ to identify the source of errors to ensure that these are not repeated in future.
 The input variables in an analysis may be directly measured (e.g., rainfall)
or they may be derived using a relationship with one or more variables (e.g.,
discharge that has been obtained from a rating curve). In the latter case the
error in the variable (discharge) depends both on field measurements and the
error in the relationship. An error may also be introduced if the relationship is no
longer valid or the values are extrapolated outside the applicable range.

VALIDATION INVOLVES DIFFERENT TYPES OF COMPARISONS OF DATA AND

INCLUDES THE FOLLOWING:
Single series comparison
1. between individual observations and pre-set physical limits;
2. between sequential observations to detect unacceptable rates of change
and deviations from acceptable behavior (most readily identified
graphically); and
3. between two measurements of a variable at a single station, e.g., daily
rainfall from a daily gauge and an accumulated total from a recording
gauge.
Multiple Stations/Data
1. between two or more measurements at nearby stations, e.g. flow at two
sites along a river; and
2. between measurements of different but related variables, e.g., rainfall
and river flow.

( . .)
• Validation of data is best done soon after observation and at the
observation station because secondary or related information to support
validation is readily available. However, data validation at observation sites
may not be always possible due to logistics and the lack of trained
personnel.
Validation of hydrological data can be grouped in three major categories:
a) primaryvalidation,
b) secondary validation, and
c) hydrological validation

( . . )
Primary data validation is done to highlight and, if possible,
correct those data which are not within the expected range.
Primary validation involves comparisons within a single data
series or between observations and pre-set limits and/or
statistical range of a variable or with the expected behavior of
the generating process.
( . . )

After primary validation secondary validation of data is

taken up to for expected spatial behavior of the variable as
inferred from neighboring observation stations. It is assumed
that the variable under consideration has spatial correlation
within small distances. This assumption must be supported by
the underlying behavior of the process under examination.