![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Assessing the impact of measurement frequency on accuracy and uncertainty of water quality data |
VerfasserIn |
Björn Helm, Stefanie Schiffner, Peter Krebs |
Konferenz |
EGU General Assembly 2014
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 16 (2014) |
Datensatznummer |
250100354
|
Publikation (Nr.) |
EGU/EGU2014-16295.pdf |
|
|
|
Zusammenfassung |
Physico-chemical water quality is a major objective for the evaluation of the ecological state
of a river water body. Physical and chemical water properties are measured to assess the river
state, identify prevalent pressures and develop mitigating measures. Regularly water quality is
assessed based on weekly to quarterly grab samples. The increasing availability of
online-sensor data measured at a high frequency allows for an enhanced understanding of
emission and transport dynamics, as well as the identification of typical and critical
states.
In this study we present a systematic approach to assess the impact of measurement
frequency on the accuracy and uncertainty of derived aggregate indicators of environmental
quality. High frequency measured (10 min-1 and 15 min-1) data on water temperature, pH,
turbidity, electric conductivity and concentrations of dissolved oxygen nitrate, ammonia and
phosphate are assessed in resampling experiments. The data is collected at 14 sites in
eastern and northern Germany representing catchments between 40 km2 and 140 000
km2 of varying properties. Resampling is performed to create series of hourly to
quarterly frequency, including special restrictions like sampling at working hours or
discharge compensation. Statistical properties and their confidence intervals are
determined in a bootstrapping procedure and evaluated along a gradient of sampling
frequency.
For all variables the range of the aggregate indicators increases largely in the bootstrapping
realizations with decreasing sampling frequency. Mean values of electric conductivity, pH
and water temperature obtained with monthly frequency differ in average less than five
percent from the original data. Mean dissolved oxygen, nitrate and phosphate had in most
stations less than 15 % bias. Ammonia and turbidity are most sensitive to the increase of
sampling frequency with up to 30 % in average and 250 % maximum bias at monthly
sampling frequency. A systematic bias is recognized in oxygen and temperature with
sampling at working hours. Turbidity has a systematic negative bias with decreased sampling
frequency. Most matter constituents show a site specific bias that increases with decreasing
sampling frequency. Discharge compensation largely decreases uncertainty and bias of the
results for almost all constituents and stations. For upper quantiles and maximum
concentration the described tendencies in bias are almost always more pronounced. Small and
steep catchments are most sensitive to bias of the aggregate indicators with frequency
decrease.
The expanding use of high frequency sensors for water quality monitoring enables an
enhanced understanding of water quality dynamics. Turbidity and ammonia as
particle related variables are especially sensitive to low measurement frequency. The
observation that especially small catchments require a high monitoring frequency
poses a challenge for future research and the development of monitoring schemes. |
|
|
|
|
|