Measurement Errors in the Determination of Physicochemical Water Quality Indicators: Sources, Evalua

Measurement errors in the determination of physicochemical water quality indicators directly affect the reliability of compliance decisions, pollution assessments, and treatment process control. This article systematically examines the sources of systematic, random, and gross errors across the entire water testing workflow, explains how trueness, precision, and measurement uncertainty are evaluated in accordance with ISO, GB, ASTM, and EN standards, and outlines the quality control practices that accredited third-party laboratories apply to keep analytical deviations within acceptable limits.

Technical Overview

Measurement error in water quality testing refers to the difference between a measured value of a physicochemical indicator and its true value. Because the true value of a water sample can never be known exactly, laboratories characterize error through two complementary statistical concepts defined in ISO 5725: trueness, which describes the closeness of agreement between the mean of many replicate results and a reference or certified value, and precision, which describes the scatter among replicate results under repeatability or reproducibility conditions. The combination of both concepts is expressed as accuracy, while the quantitative statement of doubt attached to a single result is expressed as measurement uncertainty.

Errors in the determination of physicochemical indicators are conventionally classified into three categories:

• Systematic errors, which shift all results in the same direction. Typical causes include biased calibration standards, contaminated reagent blanks, unrecognized matrix interferences, improperly standardized titrants, and drift in instrument response.
• Random errors, which fluctuate in magnitude and sign around the mean. They arise from electronic noise, minor variations in temperature and timing, small volumetric dispersions, and the inherent heterogeneity of the sample itself.
• Gross errors, which result from mistakes such as sample mix-ups, transcription errors, carryover between high- and low-level samples, or contamination introduced during collection and pretreatment.

A frequently underestimated fact is that the total error budget of a water quality determination is often dominated by the sampling, preservation, and transport stages rather than by the instrumental analysis itself. Errors introduced before the sample reaches the laboratory cannot be corrected afterward. For this reason, the ISO 5667 series governs sampling program design and sample handling, ISO 3696 specifies the grade of water required for analytical use, and ISO/IEC 17025 requires accredited third-party laboratories to estimate measurement uncertainty, ensure metrological traceability of standards and instruments, and operate documented quality control procedures. Supporting frameworks such as ISO 21748 provide a top-down route for converting repeatability, reproducibility, and bias data into uncertainty estimates, while national method standards such as the GB series, ASTM standards, and EN/ISO method standards specify the permissible deviation criteria for individual parameters.

Testing Samples

The magnitude and structure of determination errors depend strongly on the water matrix being tested. The following sample categories each present characteristic error profiles.

Drinking Water and Treated Water

Drinking water is a relatively clean, stable matrix, so analyte concentrations are often low and the relative contribution of blank contamination and calibration bias becomes significant. Residual chlorine continues to react with nitrogen species and organic indicators after bottling, and losses of carbon dioxide during transport shift pH values. Errors are controlled through immediate on-site measurement of unstable parameters, zero-blank verification, and traceable buffer and standard solutions.

Surface Water

Rivers, lakes, and reservoirs exhibit strong spatial and temporal heterogeneity. Stratification, algal patches, and episodic runoff create natural variability that must be distinguished from analytical error through replicate sampling as required by ISO 5667. Suspended particulates cause sub-sampling bias when the aliquot withdrawn for COD, total phosphorus, or metals analysis is not representative of the whole bottle; homogenization procedures must therefore be strictly followed to avoid systematic low or high bias.

Groundwater

Groundwater sampling is prone to errors caused by inadequate well purging, turbidity introduced by pumping, and geochemical changes during sample contact with air. Loss of dissolved carbon dioxide raises pH, oxygen ingress oxidizes reduced iron and alters redox-sensitive species, and volatile components may be lost during transfer. Consistent flow-rate control and field measurement of stabilizing parameters reduce both random and systematic components.

Municipal and Industrial Wastewater

Effluent samples contain high concentrations of solids, oils, surfactants, and diverse organic compounds that interfere with colorimetric and titrimetric determinations. Chloride interference in dichromate COD digestion, color and turbidity interference in spectrophotometric nutrient methods, and day-to-day load fluctuations make wastewater the most error-intensive routine matrix. Composite sampling, chloride masking, matrix-matched calibration, and standard addition verification are standard countermeasures.

Industrial Process, Boiler, and Cooling Water

High-purity process waters challenge the low end of the measurement range: trace conductivity is affected by atmospheric carbon dioxide absorption, while hardness and silica determinations on scaling samples suffer from non-homogeneous deposits. Conversely, concentrated cooling waters require large dilutions, so volumetric error and dilution water purity become dominant contributors to the uncertainty budget.

Seawater, Brackish Water, and Aquaculture Water

High salinity suppresses analytical recovery in nutrient and metal determinations and produces spectral and matrix interferences in ICP-based and colorimetric methods. Salinity-matched calibration standards, matrix-matched quality controls, and verified dilution integrity are essential to keep systematic error within method tolerances.

Reclaimed and Recycled Water

Reclaimed water combines residual organic matter, nutrients, and disinfection by-product precursors in variable proportions. Because the matrix changes with the treatment train, laboratories must periodically revalidate recovery and bias using spiked samples and certified reference materials rather than relying on fixed historical control limits.

Testing Items

Common physicochemical indicators, their principal determination approaches, and their dominant error characteristics are summarized below. Applicable method families include ISO, EN ISO, ASTM, GB, and JIS standards, for example JIS K 0102 for industrial wastewater testing.

• pH value (ISO 10523, ASTM D1293, GB 6920): electrochemical determination with glass electrodes; repeatability typically within ±0.05 pH units for well-maintained systems; electrode slope must remain within approximately 95–105% of the theoretical Nernstian response, and temperature compensation must be active.
• Electrical conductivity (ISO 7888, ASTM D1125): error is governed by cell-constant calibration and temperature correction accuracy.
• Turbidity (ISO 7027-1, GB 13200): nephelometric determination with formazin-based calibration; stray light and particle settling cause systematic low bias.
• Suspended solids (GB 11901, ASTM D5907): gravimetric filtration; errors arise from filter conditioning, rinse volume, and incomplete transfer of solids.
• Chemical oxygen demand, COD (ISO 15705, GB 11914, ASTM D1252): dichromate or rapid digestion photometric methods; chloride is the classic interference and must be masked or accounted for; duplicate relative deviation criteria are commonly set at or below 10%.
• Biochemical oxygen demand, BOD5 (ISO 5815-1, GB 7488): dilution and incubation method carrying the largest relative error of routine indicators because it depends on seed microbial activity, dilution water quality, and incubation at 20 ± 1 °C.
• Total organic carbon, TOC (EN 1484 and related standards): combustion or wet chemical oxidation; inorganic carbon removal efficiency is a key bias source.
• Ammonia nitrogen (ISO 7150-1, GB 7478, GB 7479): distillation–titration or spectrophotometric determination; interferences from amines, chlorine, and turbidity require blanks and spike recovery checks.
• Nitrate, nitrite, and chloride (ISO 10304-1, GB 11896): ion chromatography or titrimetry; calibration curve linearity and carryover govern precision.
• Total nitrogen (GB 11894, ISO 11905-1): persulfate digestion followed by spectrophotometry; incomplete digestion causes systematic low results.
• Total phosphorus (ISO 6878, GB 11893): ammonium molybdate spectrophotometry; silica and arsenate interferences and digestion completeness are critical.
• Total hardness (ISO 6059, GB 7477): EDTA titrimetry; endpoint recognition subjectivity and titrant standardization dominate the error budget.
• Alkalinity (ISO 9963-1): acid-base titration sensitive to endpoint pH selection and carbon dioxide exchange.
• Dissolved oxygen (ISO 5813 Winkler, ISO 5814 electrochemical, GB 7489, ASTM D888): air bubbles, sample fixation delay, and sensor membrane fouling are principal error sources.
• Metals such as copper, zinc, lead, cadmium, iron, and manganese (ISO 11885, ISO 17294-2, GB 7475): ICP-OES, ICP-MS, or AAS determination; spectral interferences, memory effects, and contamination from containers require acid-washed vessels and internal standard correction.
• Fluoride (GB 7484, ISO 10359-1): ion-selective electrode; ionic strength adjustment and complexing interferences affect recovery.
• Temperature, oxidation–reduction potential, and total dissolved solids: field parameters whose values condition the accuracy of all other determinations and must be logged with every measurement.

Typical acceptance criteria applied during routine quality control include calibration verification within ±10% of the reference value, matrix spike recovery of 90–110% for level determinations, duplicate relative deviation limits set by each method standard, and blank levels below one-tenth of the reporting limit.

Testing Methods

Sampling and Preservation Error Control

Sampling is executed according to the ISO 5667 series and national sampling standards: appropriate container materials are selected for each parameter, samples are filtered or acidified as specified, and holding times and temperatures are strictly observed. Metals samples are typically acidified to below pH 2, nutrient and COD samples are preserved by cooling and acidification as required by the specific method, and unstable parameters such as pH, dissolved oxygen, and temperature are measured on site whenever possible. Chain-of-custody documentation prevents gross errors from sample confusion.

Titrimetric and Gravimetric Determinations

Classical methods such as EDTA hardness, Mohr chloride, Winkler dissolved oxygen, and gravimetric suspended solids rely on calibrated Class A glassware, electronically verified balances, and titrants standardized against primary standards. Endpoint detection is standardized through potentiometric indication where available, and systematic bias is monitored through periodic analysis of certified reference materials.

Spectrophotometric Determinations

Colorimetric methods for COD, ammonia, total nitrogen, and total phosphorus require wavelength verification, matched cuvettes, fresh reagent blanks, and multi-point calibration curves whose correlation and residual distribution are checked before use. Absorbance is kept within the linear range of the Beer–Lambert law, and turbidity or color of the sample is compensated using sample blanks or dual-wavelength correction to suppress matrix-induced systematic error.

Electrochemical Determinations

pH, conductivity, dissolved oxygen, and fluoride electrodes are calibrated before each batch using at least two traceable standards bracketing the expected range. Electrode slope, response time, and offset are recorded and must satisfy the limits stated in the method; temperature probes are verified against a certified thermometer because a one-degree error can shift pH readings by several hundredths of a unit.

Chromatographic and Spectrometric Instrumental Determinations

Ion chromatography and ICP techniques apply internal standards, calibration blanks, continuing calibration verification, and interference check standards. The standard addition method is used to quantify and correct matrix effects in difficult samples such as seawater and industrial effluent, while carryover blanks after high-level injections protect against gross positive bias in trace-level sequences.

Uncertainty Evaluation and Data Processing

Laboratories construct uncertainty budgets following the bottom-up approach of the GUM framework or the top-up approach of ISO 21748, combining components from sampling, calibration standards, volumetric and gravimetric operations, repeatability, and recovery bias. The reported result is expressed as the measured value plus or minus the expanded uncertainty with a coverage factor of two, corresponding to a confidence level of approximately 95%. Rounding and unit conversions are performed only at the final reporting step to avoid accumulated processing error.

Laboratory Quality Control System

Under ISO/IEC 17025, accredited third-party laboratories operate a full quality control regime comprising method blanks, laboratory duplicates, matrix spikes, certified reference materials, control charts with warning and action limits, interlaboratory proficiency testing, instrument maintenance and calibration schedules traceable to national metrology institutes, and documented validation of all non-standard and in-house developed methods. These measures collectively bound systematic and random errors and provide objective evidence of the reliability of reported values.

Applications

• Regulatory compliance monitoring: error and uncertainty statements determine whether a measured concentration lies below or above a statutory limit, and conformity decision rules with guard bands are applied when results fall close to the limit value.
• Drinking water safety assessment: tight control of pH, turbidity, metals, and disinfection by-product indicator errors supports defensible supply-quality conclusions.
• Wastewater treatment process control: reliable COD, BOD5, ammonia, and total phosphorus data allow correct adjustment of aeration, chemical dosing, and sludge retention times; propagation of analytical error into control decisions is explicitly evaluated.
• Environmental impact assessment and pollution investigation: legally defensible datasets require documented sampling representativeness, full uncertainty budgets, and preserved chain of custody.
• Industrial water and utility management: boiler feed water, cooling circuit, and pure water system monitoring depend on low-level determinations in which blank and calibration errors dominate.
• Method development and reference material verification: error studies using replicate design and interlaboratory comparison underpin the establishment of repeatability and reproducibility limits published in GB, ISO, and ASTM method standards.
• Aquaculture and irrigation water management: accurate dissolved oxygen, ammonia, and salinity determinations protect livestock health and crop safety.

FAQ

What is the difference between measurement error and measurement uncertainty?

Measurement error is the idealized difference between a single result and the true value, which is unknowable in practice. Measurement uncertainty is a quantified, statistically derived parameter that characterizes the dispersion of values that could reasonably be attributed to the measurand, combining estimated bias and precision components. Modern laboratory practice, as required by ISO/IEC 17025, reports uncertainty rather than claiming a specific error value.

Which stage of water quality testing contributes the most error?

In most monitoring programs, sampling, preservation, and transport contribute the largest share of total error, frequently exceeding the analytical component. A perfectly executed laboratory determination cannot compensate for a non-representative or improperly preserved sample, which is why ISO 5667 procedures and on-site measurement of unstable parameters are so important.

What typical acceptance criteria are used to judge whether results are acceptable?

Commonly applied criteria include calibration verification within ±10% of the true value of a check standard, blank absorbance or concentration below the method-defined threshold, duplicate relative deviation within the limit stated in the applicable standard, and matrix spike recovery generally within 90–110% at routine levels. Control chart limits, typically set at plus or minus two and three standard deviations, trigger corrective action when exceeded.

Why do COD results in high-chloride samples show large deviations?

Chloride is oxidized by dichromate and consumes the oxidant, producing a positive bias that grows rapidly above approximately one thousand milligrams per liter of chloride. Methods such as GB 11914 and ISO 15705 specify masking agents or correction procedures, and chloride-containing quality control samples are analyzed to verify that the applied correction keeps the systematic error within the stated tolerance.

How can errors in pH determination be minimized?

Use a freshly calibrated electrode with a slope within 95–105% of theoretical, employ at least two buffers that bracket the sample value, verify temperature compensation, rinse electrodes thoroughly to prevent carryover, measure promptly after sampling to avoid carbon dioxide exchange, and stir samples gently without entraining air. Regular electrode maintenance and replacement of aged membranes or junctions eliminate the most common systematic drift.

How is measurement uncertainty evaluated in an accredited laboratory?

Two complementary approaches are used. The bottom-up approach builds an uncertainty budget from individual components such as calibration certificates, volumetric tolerances, and repeatability studies. The top-down approach described in ISO 21748 derives uncertainty from collaborative trial precision data, certified reference material bias studies, and internal quality control records, and both approaches are accepted by ISO/IEC 17025 assessors.

What should be done when two replicate determinations disagree beyond the permissible deviation?

The laboratory identifies whether the cause is instrumental drift, sample heterogeneity, or procedural deviation, checks blank and calibration validity, homogenizes and re-analyzes the sample, and documents the investigation as a nonconforming work record. Only results passing the duplicate deviation criterion of the applicable method standard are released to the client.

Can measurement uncertainty be used directly for compliance decisions?

Yes. When a reported value with its expanded uncertainty overlaps a regulatory limit, a documented decision rule must be applied, such as simple acceptance or acceptance with a guard band that subtracts the expanded uncertainty from the limit. This practice ensures that conformity statements are statistically defensible in regulatory and legal contexts.