Environmental Engineering Reference
In-Depth Information
Dealing With Variability and Uncertainty
Variability happens. In spite of rigorous data collection and QA/QC protocols, i eld data
will exhibit variability. Natural systems are inherently variable, and through sampling and
sample handling and analysis, we introduce additional variability. Uncertainty, in turn,
compounds variability. Uncertainty arises because there is no such thing as a truly exact
measurement, and samples cannot be collected continuously, forever. Instead, we periodi-
cally collect samples to represent an environment that is continuously changing over time
and space. And we analyze these periodic samples using methods that have limits in reso-
lution, precision, and accuracy. Much of the following information on variability is based
on Rector 1995.
Statistics is the science of reaching conclusions in the face of variability and uncertainty.
We cannot eliminate uncertainty and variability, but we can use statistics to estimate their
contribution to our observed results and make informed decisions based on the data. The
most frequently used descriptive statistics are those that describe central tendency (i.e.
arithmetic means, geometric means, and medians) and those that describe the distribution
or variability (range, quartiles, and coni dence interval/standard deviation). In general, it
is appropriate to use the average when data sets are normally distributed around a mean
value with no outliers. It is better to use the median if the data is skewed and/or if there
are outliers. The geometric mean is rarely used except for bacteria monitoring.
Other statistical techniques, such as trend analysis, can also be conducted. However, they
may require years of data and/or more advanced statistical techniques. Deciding which
statistical measure to use depends upon the type of data to be summarized.
Table 8.10
rec-
ommends common summaries for different indicators.
Natural systems are inherently
variable, and through sampling
and sample handling and analysis,
we introduce additional
variability.
TABLE 8.10
Suggested Statistical Summaries for General Chemical and Physical Water Parameters
(Table 8.10 is adapted from Data to Information for Coastal Volunteer Water Quality Monitoring Groups in New Hampshire and Maine, by Dates and Schloss (University of Maine
Cooperation Extension and University of New Hampshire/Maine Seas Grant Extension, 1998))
Parameter
Statistical summary
Parameter
Statistical summary
Total suspended solids
Average
Median
Flow-weighted average
Range
Quartiles
Confi dence intervals or standard
deviation
pH
Median or average
Quartiles
Minimum
Median
Quartiles
Minimum
Alkalinity
Temperature (water or air)
Seasonal average
Seasonal median
Maximum
Range
Quartiles
Chlorophyll-a
Seasonal average
Range
Maximum and minimum
Median
Quartiles
Dissolved oxygen (as mg/l)
Seasonal median
Minimum
Quartiles
Confi dence intervals or standard
deviation
Turbidity
Median
Maximum
Quartiles
Flow
Average
Maximum and minimum
Median
Quartiles
(Continued)
Search WWH ::
Custom Search