Back to top or back to home page or back to Whats New or back to Onalaska Ecology Links
Quality Assurance/Quality Control (QA/QC) are the measures that you take to ensure your data are accurate and useful. Even the best water quality data will have errors, and it is the goal of the QA/QC program to measure and minimize these errors. Data quality is described by its accuracy, precision, completeness, representativeness, and comparability.
Accuracy is the difference between your measured value and the "true" value. It is one of the most difficult QA/QC parameters to measure, since you usually don't know what the true value is. The most common way to estimate accuracy is to test your methods with a sample that has a known chemical concentration (this kind of sample is known as a reference sample ). Accuracy is then the difference between the measured value and the true value, expressed as a percentage of the true value:
Accuracy = 100. x (Measured Value - True Value)/True Value
If a method's accuracy is 10%, you would expect your measurements to be within 10% of the true value.
Precision describes the repeatability of your methods. A method is precise if you get the same result everytime you analyze similar samples, and imprecise if you get widely-differing results. Precision is measured by analyzing two duplicate water samples that are taken at the same location and time. It is expressed as the Relative Percent Difference (RPD) between the chemical concentrations measured from the two duplicates:
RPD = 100. x (Duplicate 1 - Duplicate 2)/(Average of the two duplicates)
where Average = (Duplicate 1 + Duplicate 2)/2
Completeness measures how well you finished all of the sampling that you originally planned to do. Completeness is expressed as the percentage of samples you measured relative to the number that were planned. For instance, if you were supposed to take 10 samples at a location and were only able to take 9, your data completeness would be 90 percent. Few monitoring programs can achieve 100% completeness; bad weather, equipment problems, and budget problems all result in some loss of data.
Representativeness describes how well your sample represents the environmental condition you are trying to measure. It is controlled primarily by how you choose your sampling locations and timing. For instance, a sample collected just after an oil spill would not be representative of typical conditions in the river. A sample collected downstream of a sewage treatment plant would not be representative of background (or natural) water quality.
Comparability describes how well your data can be compared with other data. To maintain comparability, the methods you use to collect and analyze samples should remain consistent - you should not switch methods in the middle of a study without good reason. If you are trying to compare your results to data from an earlier study, you should sample at the same locations and at the same times of year.
Go to: What Is Quality Assurance/quality Control?
- | - Sources Of Error
- | - Quality Control Checks
- | - Quality Assurance/quality Control Plans
- | - Exercises
or Back to top or back to Water Quality Index Page or back to Whats New
A typical sample goes through a lot of steps before it becomes part of your water quality data set, and there is potential for error at each of these steps. The major sources of error are measurement error, sample handling error, and natural variability.
Measurement errors result because none of the methods (field kits, calibrated instruments, or laboratory analysis) provide perfect water quality measurements. Measurement error can be reduced by instrument calibration, proper training, and equipment maintenance, but it never goes away entirely.
A special kind of measurement error comes from the method's detection limit. As chemical concentrations approach zero it becomes more and more difficult to get accurate measurements. The point where the method is no longer able to detect a chemical is called the method detection limit. For instance, the GREEN field kit cannot measure nitrate concentrations below 5 mg/l; this method's detection limit is therefore 5 mg/l. The important idea here is that you never report a value of zero concentration, since all you really know is that the concentration is less the detection limit. Instead of writing zero as the result, you write the detection limit with a less-than symbol (for example, nitrate concentration = <5 mg/l).
Sample handling errors come from the ways in which you collect and handle your samples. Samples may be contaminated from your hands, or because air is trapped in the sample bottle when you close it. Improper storage and transportation of the sample are other sources of handling error. This kind of error is minimized by closely following proper handling procedures.
Natural variability is often the biggest source of imprecision, and is unfortunately largely out of your control. When you are measuring water quality in a river, you are really only sampling the small piece of the river that you are able to fit into your sample bottle. During the few seconds it takes you to fill your 1-liter sample bottle, literally thousands of cubic feet of water have flowed past you. Every parcel of this water will have different water quality characteristics than what you measure in your sample. Natural variability is a basic feature of a river, and cannot be controlled. The best approach is therefore to quantify this variability by taking as many samples as you can afford.
Go to: What Is Quality Assurance/quality Control?
- | - Sources Of Error
- | - Quality Control Checks
- | - Quality Assurance/quality Control Plans
- | - Exercises
or Back to top or back to Water Quality Index Page or back to Whats New
Every sampling program should have a set of tests and checks that measure data quality. Common checks include duplicate samples, blanks, reference samples, and performance audits.
Duplicate samples are simply two identical samples collected and handled in the same way. They measure the precision of your methods. Field duplicates are two samples collected in the field from the same location at the same time; these measure the precision of your entire procedure (sampling, storage and handling, and laboratory analysis). Laboratory duplicates are two samples split from a single sample once it has arrived at the laboratory. These test the precision of the laboratory methods only.
Blanks are samples containing pure, uncontaminated water. Blanks contain none of the chemical you are trying to measure, and are used to identify contamination that might occur in the field or laboratory. If your laboratory measures a positive chemical concentration in a blank, you then know that there is a source of contamination somewhere in your procedure. A field blank is a blank sample that is placed in a sample bottle at the field site, and is handled the same as a normal sample. It identifies contamination that might occur in your entire procedure (from field sampling to laboratory analysis). A laboratory blank is prepared at the laboratory, and tests for laboratory contamination only.
Reference samples are prepared by an independent laboratory, and contain a known concentration of the chemical you are measuring. They are similar to the standards used in instrument calibration, except that the actual concentration is kept secret from your laboratory people. Reference samples measure the accuracy of your laboratory procedures.
A performance audit is an independent review of your sampling and laboratory methods, conducted by someone who is familiar with your project but is not a part of your day-to-day project team. The idea is to have your work reviewed by a qualified person who has no stake in the outcome of your project. The performance auditor will check to see how well you are following your sampling and QA/QC plans.
Go to: What Is Quality Assurance/quality Control?
- | - Sources Of Error
- | - Quality Control Checks
- | - Quality Assurance/quality Control Plans
- | - Exercises
or Back to top or back to Water Quality Index Page or back to Whats New
Every water quality sampling program should have a detailed Quality Assurance/Quality Control Plan that describes:
Data quality objectives are your goals for accuracy, precision, completeness, representativeness, and comparability. The chain of custody identifies who is responsible for collecting, transporting, and analyzing each sample. This helps to ensure that your project team knows their responsibilities, and makes it easier to identify the sources of problems. Data reduction, validation, and reporting describes how you will store and report your results. All data should be filed in an organized database, and should be checked to make sure that numbers are entered correctly.
Go to: What Is Quality Assurance/quality Control?
- | - Sources Of Error
- | - Quality Control Checks
- | - Quality Assurance/quality Control Plans
- | - Exercises
or Back to top or back to Water Quality Index Page or back to Whats New
1. The accuracy goal for nitrate in your sampling plan is 20 percent. An independent laboratory has given you a reference sample with a known nitrate concentration of 30 mg/l. Your laboratory analyzes the sample and finds a nitrate concentration of 40 mg/l. What is the accuracy for your laboratory's nitrate analysis? Does it meet your sampling plan goal?
2. Duplicate samples were analyzed for dissolved oxygen. The results were 7.5 mg/l and 7.0 mg/l. What is the Relative Percent Difference between these duplicates? Does it meet a precision goal of 10%?
3. A field blank was run through a fecal coliform sampling and laboratory analysis procedure. The result was a measured fecal coliform value of 50 #/100ml. A laboratory blank was then run through the laboratory procedure only, and had a measured fecal coliform count of 60 #/100ml. Do you think the contamination occurred in the field or in the laboratory?
4. The project manager for a water quality monitoring project has decided to do the performance audit herself. Is this appropriate, or should she choose someone else?
This water quality course material created by Rob Schanz. Send comments to Rob Schanz
This page created and maintained by Chehalis River Council
Send comments or questions to the: Chehalis River Council