NIKLAS - Automatical quality control of time series data - ERAD 2010

ERAD 2010 - THE SIXTH EUROPEAN CONFERENCE ON RADAR IN METEOROLOGY AND HYDROLOGY
NIKLAS - Automatical quality control of time
1. Introduction
series data
G.Lempio 1 , C. Podlasly 1 , T. Einfalt 1
1 hydro & meteo GmbH & Co.KG, Breite Str. 6-8, D-23552 Lübeck, Germany
g.lempio@hydrometeo.de
Guido Lempio
The amount of precipitation measured by radar is relatively imprecise, caused by the determination through
empirical ZR-relationships. For hydrological purposes, the accuracy of these measurements is not sufficient. One
option to improve radar data is their adjustment with measurements from rain gauges. For this application it is
extremely important to use only high quality time series (Steiner et al., 1999).
The knowledge about the quality of input data from station measurements for radar adjustment, analysis and
forecast tasks is important in order to assume the plausibility of results derived by models. Small amounts of input
data can be quality controlled manually by experienced observers, but larger amounts of data and real time
applications do not. Automatic algorithms can be used to reduce the manual effort for quality check by identifying
the suspicious cases that have to be examined by human observers and to do a data check for real-time applications
assuring a certain data quality.
NIKLAS has been developed as a module for real-time and non-real-time check of meteorological data, required
from Landesamt für Umwelt, Wasserwirtschaft und Gewerbeaufsicht Rheinland-Pfalz, Germany
(www.wasser.rlp.de), for the Mosel forecast and warning system (www.timisflood.net). It validates time series for the
parameters:
• precipitation
• global radiation
• sunshine duration
• air temperature
• dew point temperature
• relative humidity
• wind speed
• air pressure
The quality check scheme is based on a series of tests. Their choice and order depend on the type of the
meteorological parameters. The methodology for the test algorithms was taken from literature and operational
experiences (Einfalt et. al., 2006).
As result of each test the examined data records are classified into three quality levels – valid, low/limited quality
or bad quality. NIKLAS as a pure quality check tool excludes records with bad quality from the time series and does
not replace them by plausible estimations.
The NIKLAS working package has been applied successfully for real-time (TIMISflood) as well for non-realtime
(ExUS) validation tasks (Quirmbach, et. al. 2009).
2. NIKLAS – tool for automatic quality control
2.1 Control algorithms, for example applied to raingauge data
Quality Control (QC) of raingauge data has been an important topic since the beginning of data collection. Attempts
to formalize this task have been started in several countries (see Einfalt et al. (2000) for an overview). However, the
conclusion has been that the check of point rainfall measurements can only be done by human eye with a reliably
good result (Jörgensen et al. 1998; Maul-Kötter and Einfalt 1998).
Two aspects may prevent such manual procedures: real-time data to be further processed for flood warning, and
large amounts of data to be investigated.

ERAD 2010 - THE SIXTH EUROPEAN CONFERENCE ON RADAR IN METEOROLOGY AND HYDROLOGY
An automatic quality check starting from simple cases before tackling the difficult ones yields surprisingly good
results.
This approach has been proposed by the Bavarian agrometeorological service (Vaitl 1988) and refined by
MeteoSwiss (Musa et al. 2003). Basically, the raingauge quality check starts with items that are easy to check followed
by more complex ones. In practice, this means that firstly, existing gaps in the data are excluded from further
treatment, then features on the data of one station only are investigated, and finally an intercomparison between the
data from several raingauges is performed. The structure from simple check elements at the top, down to more
complex ones, as proposed above, comprises the following steps (Einfalt et al. 2006):
1. Detection of gaps in the data
2. Detection of physically impossible values
3. Detection of constant values
4. Detection of values above set thresholds
5. Detection of improbable zero values
6. Detection of unusually low values (which may be real, though)
7. Detection of unusually high values (which may be real, though)
2.1.1 Gaps in the data (Completeness)
The detection of gaps in the data mainly serves to exclude the gap interval of a station to be used for
comparison to other stations. Furthermore, a gap statistic can be derived which is an indicator of the reliability of the
data from this station. Experience shows that well maintained stations with a good data quality rarely have gaps in
their measurement series.
2.1.2 Physically impossible values (Extreme value check)
Physically impossible data consist of negative rainfall values and very high intensities, e.g., more than 5 mm per
minute in a moderate climate. Such values should be excluded from further evaluation.
As a function of the underlying data and software, very high intensities may be a side effect of digitized paper
charts. Such values can be further used if the time step for further analysis is large enough. For example, a value of 5
mm in one minute can in reality be representative for five or ten minutes when analyzed with additional information
(e.g., the basic paper charts). In such a case, the evaluation on a five or ten minute time grid is the correct further
treatment.
2.1.3 Constant values
Constant values over a certain time are an indicator of unusable data which may be either due to bad digitization
or to missing values in case of digital registration. The “certain time” for digital data with a time step of 1 to 5
minutes is around 15 minutes for intensities above 1 mm/h. For digitized data derived from paper charts, this time
interval is a function of the paper chart resolution and may be as long as 60 minutes for old paper charts.
2.1.4 Values above set thresholds (Extreme value check)
For predefined durations, it is useful to indicate when the measurements are above statistically rare values, e.g.,
higher than an event occurring every five or ten years. Useful durations comprise 5 minutes, 15 minutes, 60 minutes
and 1440 minutes (one day).
Such thresholds were defined in order to identify “interesting” events, i.e., events which should be carefully checked
before accepting the measurements.
2.1.5 Improbable zero values (Spatial consistency)
While all of the above checks are performed on one station only, this check and the following two use the spatiotemporal
rainfall structure as seen by several gauges. Improbable zero values can be detected at one station if all
surrounding stations have significant rainfall and stations are close enough to each other.
2.1.6 Unusually low daily values (Spatial consistency)
A more sophisticated check is the check on too low values where the daily sum of a selected station is compared
to the daily sums of the neighboring stations. If the surrounding stations recorded significantly more rainfall than the
selected one, the measurement of this station has to be considered as doubtful.

ERAD 2010 - THE SIXTH EUROPEAN CONFERENCE ON RADAR IN METEOROLOGY AND HYDROLOGY
2.1.7 Unusually high daily values (Spatial consistency)
A season dependent check is the one on too high values, where the daily sum of a selected station is compared to
the daily sums of the neighboring stations. If the surrounding stations recorded significantly less rainfall than the
selected one, the measurement of this station has to be considered as doubtful if a convective rainfall can be
excluded. This is usually the case in Germany between October and March.
Additional check elements in NIKLAS are:
- Variability
- Inner consistency
2.1.8 Variability
This check compares two successive values from one time series in order to examine unusual differences of
these values.
2.1.9 Inner consistency
The inner consistency refers to the behavior of different parameters to each other at the same place. Values of
different parameters at the same time and the same place must be in a well defined relationship.
2.2 Using NIKLAS for quality control
NIKLAS is a program for the plausibility check of meteorological station measurements of the parameters
precipitation, air temperature, dew point temperature, relative humidity, global radiation, sunshine duration,
atmospheric pressure and wind speed.
2.2.1 Technical details to NIKLAS
The program NIKLAS is being delivered as WINDOWS application. Its required resources are at least
Windows® XP as operating system and a RAM size of at least 512 MB. NIKLAS consists of two executables, first
the command-line oriented computation program and second the graphical user interface (GUI) as support for the
determination of the settings.
The required input for NIKLAS are the time series of the meteorological parameters to be analyzed as well as
the station information file and the configuration file (both in human readable ASCII-format). Output information
comprises files with the results in form of a time series, as a check series, a log file and – if selected – extended test
result information.
All input data files need to have one of the following in Germany used specific formats:HMZ, station file
format, ZRX.
2.2.2 How to work with the NIKLAS-GUI
The command line tool NIKLAS uses the configuration file “niklas.inp” which controls the behavior via control
words and parameter settings. The NIKLAS-GUI is a tool to setup the configuration file in an easy way. There are
six main windows to perform the configuration:
1. Basic Settings: Selection of …
- the GUIs language, - input data format, - output files to be produced, - detailed check information, -
procedures to be performed
2. Extreme value check: Selection of …
- minimum and maximum values for each selected parameter independently for warnings (“Warning value” –
value is flagged) and/or errors (“Extreme value” – value is deleted). Maximum values per 60 minutes, per one
day or per one month can be set for precipitation in order to integrate the cumulative check.
3. Variability: Selection of …
- maximum upward/downward tolerance between successive values of the parameters temperature, relative
humidity, atmospheric pressure and dew point temperature.
- A test according to Vaitl can be selected for wind speed but only available for the routine mode. This test is
only implemented on hourly values for a time interval of a complete day and combines a given value with its
predecessor and successor.
- Results of a failed test are warnings.

ERAD 2010 - THE SIXTH EUROPEAN CONFERENCE ON RADAR IN METEOROLOGY AND HYDROLOGY
FIG. 2-1. NIKLAS-GUI Window "Basic Settings".
4. Constant values: Selection of …
- maximum allowed time interval in hourly values for warnings and extreme values.
- For the relative humidity, a difference is made between values below 95% and values equal or larger than
95%. The first case should have stronger test criteria, but both tests produce errors or warnings when not
passed.
- Also for wind speed, there is a differentiation: values with > 0 m/s are checked with stronger criteria than
values where the wind speed is 0 m/s.
- For the relative humidity and the wind speed, the offline (routine) mode produces a warning, while the online
(operational) mode is producing an error when the test fails.
5. Inner consistency: Selection of …
- tests to perform (Air temperature – dew point temperature comparison, Relative humidity – Precipitation
comparison, Sunshine duration – Precipitation comparison, - Check rain gage with radar data) and their
outcome (Warning or Error).
6. Spatial consistency: Selection of …
- permitted deviation of atmospheric pressure, relative humidity, air temperature, dew point temperature,
global radiation, sunshine duration and precipitation,
- maximum distance and height corridor for analyzed stations,
In order to do the processing, a maximum distance for the analysed stations needs to be defined, and a height
corridor can be set in order to compare stations whose altitude is not too different from each other (“General
settings”). The precipitation test allows to define additional parameters, e.g. the test for daily or monthly
values, the consideration of the synoptical situation, the UMGMAX tests A and B, the test on the duration of
dry weather periods or the test for doubtful zero values through a neighbourhood check.
The upper value is the factor for the maximally allowed value, to be applied to the average of the neighbouring
stations, and the lower value the factor for the permitted minimum value. The minimum value for the check to
be performed is the “minimum average value” which has to be reached either at the tested station or by the
mean of the neighbours.
The routine check only produces warnings when a test is not passed. This is identical for the operational check
except for the zero value test, the consideration of the general weather situation in the spatial consistency test
and the dry weather time period test (duration test) which produce errors.

ERAD 2010 - THE SIXTH EUROPEAN CONFERENCE ON RADAR IN METEOROLOGY AND HYDROLOGY
The zero value test checks whether a station with 0 mm of precipitation is surrounded by stations where there
is rainfall. For this the user can choose whether there need to be all neighbours with rain or only the majority
of the neighbours.
The review interval of the tested station is the time for which this station needs to have measured 0 mm of
precipitation to activate this test (intervals T1 to T3 in Figure 2-2). This value is compared to a comparison
interval (“review interval”) at the neighbouring stations which can be shifted by a defined offset (“buffer
interval” - the figure shows an offset of one hour). Thus the test can accommodate time differences in the
arrival of precipitation at these stations. The test is now performed such that interval T2 of the neighbours is
compared to T1, T2 and T3. The test flags the value of the tested station as doubtful (warning or error) if all
part intervals of the tested station are zero, but for each of the neighbours’ T2 shows a value above the
predefined minimum rainfall. The minimum rainfall is the amount of rainfall above which the neighbouring
stations are considered to be “non-zero”.
FIG. 2-2. Comparison intervals for the zero value test.
3. Automatic quality control of data
3.1 Real time automatic quality control
The LUWG (Landesamt für Umwelt, Wasserwirtschaft und Gewerbeaufsicht Rheinland-Pfalz) is operating the
Mosel flood warning system. In the context of the INTERREG TIMISflood project (www.timisflood.net), NIKLAS
has been developed. Operationally, it is used after data acquisition by WISKI to validate the measurement values.
The result is then handed over to the hydrological simulation model LARSIM.
3.2 Offline automatic quality control
FIG. 3-1. NIKLAS in operational flood warning.
In the project ExUS (Extremwertstatistische Untersuchung von Starkniederschlägen in NRW (1950 - 2008),
statistical analysis of extreme heavy precipitation values in North Rhine Westphalia), analysis of meteorological
station data were performed by NIKLAS.
The number of analysed stations was: continuously measuring 195, daily value stations 418.
The selected thresholds for the NIKLAS configuration were:
- 5 mm amount of precipitation in 1 minute,
- 17.5 mm (*) amount of precipitation in 5 minutes,
- 48 mm (*) amount of precipitation in 60minutes,
- 90 mm (*) amount of precipitation in 1440 minutes,
(*) corresponds to the 100-yearly Precipitation for Essen (Germany)

ERAD 2010 - THE SIXTH EUROPEAN CONFERENCE ON RADAR IN METEOROLOGY AND HYDROLOGY
continous values
daily values
FIG. 3-2. Verification results: blue = no remarks; yellow = remarks by NIKLAS but no data error; red = remarks and data
corrections – numbers are # of stations.
Nearly all continuously measuring and a third of the daily stations got warnings from NIKLAS. A third of these
warnings induced data corrections (set gaps or time shift). A lot of the warnings caused by to high intensities or
amounts occurred during summer and were plausible.
4. Summary
Quality control of data is a task which needs a lot of experience. Although many methods have been developed
to perform these checks automatically, they depend on the climate, the density of the network and the quality of the
data. The latter is very important, since only with good quality data erroneous data can be detected from a spatial
perspective.
The presented software does not use any additional information such as radar or satellite measurements, or
results from numerical weather models as it is done elsewhere (e.g., Musa et al. 2003).
Until now the NIKLAS working package could be applied successfully as a proper tool for real-time
(TIMISflood) as well for non-real-time (ExUS) validation tasks.
References
Einfalt T., Arnbjerg-Nielsen K., Spies S., (2000): Rainfall data measurement and processing for model use in urban
hydrology. 5th International Workshop on Precipitation in Urban Areas. Pontresina, 10.-13. December,
Einfalt T., Jessen M., Quirmbach M., (2006): Can we check raingauge data automatically 7th International Workshop on
Precipitation in Urban Areas, St. Moritz, Switzerland, 7.-10. December, ISBN 3-909386-65-2.
Jörgensen H.K., Rosenörn S., Madsen H., Mikkelsen P.S., (1998): Quality control of rain data used for urban runoff
systems, Water Sci Technol 37 No. 11
Maul-Kötter B., Einfalt T., (1998): Correction and preparation of continuously measured raingauge data: a standard
method in North Rhine-Westphalia. Water Sci Technol, vol. 37, No. 11
Musa M., Grüter E., Abbt M., Häberli C., Häller E., Küng U., Konzelmann T., Dössegger R., (2003): Quality Control
Tools for Meteorological Data in the MeteoSwiss Data Warehouse System. In: Proc. ICAM/MAP 2003., Brig,
Switzerland, 19.-23. May.
Quirmbach M., Papadakis I., Einfalt T., Langstädtler G., Mehlig B., (2009): Analysis of Precipitation data from
climate model calculations for North Rhine-Westphalia. 8 th international workshop on precipitation in urban areas , St.
Moritz, Switzerland, 10.-13. December, ISBN 978-3-909386-27-7.
Steiner M., Smith J. A., Burges S. J., Alonso C. V., and Darden R.W., (1999): Effect of bias adjustment and rain gauge
data quality control on radar rainfall estimation, Water Resources Research, 35 (8), 2487–2503.
Vaitl W., (1988): Beschreibung der Prüfkriterien für die Qualitätskontrolle stündlicher bzw. 10-minütiger Daten von
automatischen agrarmeteorologischen Stationen der Bayerischen Landesanstalt für Bodenkultur und
Pflanzenbau. p 16, München-Freising.