Review of WQI and Some Future Directions

Water Qual Expo Health (2011) 3:11–24DOI 10.1007/s12403-011-0040-0

A Review of Genesis and Evolution of Water Quality Index (WQI)and Some Future Directions

Ashok Lumb · T.C. Sharma · Jean-François Bibeault

Received: 31 December 2010 / Revised: 28 February 2011 / Accepted: 28 February 2011 / Published online: 25 March 2011© Springer Science+Business Media B.V. 2011

Abstract The concept of indexing water with a numeri-cal value to express its quality, based on physical, chemicaland biological measurements, was developed in 1965 by USbased National Sanitation Foundation (NSF). In NSFWQI,the selection of parameters is based on Delphi method andthese models were formulated in additive and multiplicativeforms. The models were implemented across various statesin the US while being continually refined. One refined formis known as Oregon Water Quality Index (OWQI). The in-dex enjoys the advantage of being free from the arbitrationin weighting the parameters and employs the concept ofharmonic averaging. Another model of WQI from Europe(Spain) is that of Bascaron (Bol. Inf. Medio Ambient. 9:30–51, 1979), which is based on the normalization of the con-

This paper was presented at the 62nd Annual Conference of CanadianWater Resources Association, held in Quebec City, QC, Canada fromJune 9–12, 2009 and was also presented at the InternationalConference on Hydrology and Watershed Management(ICHWAM-2010), held at Hyderabad, India from February 3–6, 2010.

A. Lumb (�)Ecological Monitoring and Assessment Network CoordinatingOffice, Wildlife and Landscape Science Directorate, Science andTechnology Branch, Environment Canada, Canada Centre forInland Waters, 867 Lakeshore Rd, Burlington, ON L7R 4A6,Canadae-mail: [email protected]

T.C. Sharma51B-5305, Glen Erin Drive, Mississauga, ON L5M 5N7, Canadae-mail: [email protected]

J.-F. BibeaultIndicator Development and Integration, Water Quality Monitoringand Surveillance, Water Science & Technology, EnvironmentCanada, 105 rue McGill, 7e étage, Montréal, QC H2Y 2E7,Canadae-mail: [email protected]

centrations of the water quality parameters and then aggre-gating them through an additive model with suitable weightsattached to the parameters involved. The major differencesin various WQIs are based on the mannerism of statisti-cal integration and interpretation of parameter values. A to-tally different approach was adopted in the Canadian WaterQuality Index also known as Canadian Council of Minis-ters of the Environment Water Quality Index (CCME WQI).CCME WQI and is also being used by many countries allover the world and has also been endorsed by United Na-tions Environmental Program (UNEP) in 2007 as a modelfor Global Drinking Water Quality Index (GDWQI). Themost commonly used parameters are dissolved oxygen, pH,turbidity, total dissolved solids, nitrates, phosphates, metalsamong others. All indices have one or other limitation andthe search for a perfect one is still a challenge.

Keywords CCME WQI · Delphi method · NSF WQI ·Water Quality Index

Introduction

The water quality index (WQI) is a single number that ex-presses water quality by aggregating the measurements ofwater quality parameters (such as dissolved oxygen, pH, ni-trate, phosphate, ammonia, chloride, hardness, metals etc.).Usually the higher score alludes to better water quality (ex-cellent, good) and lower score to degraded quality (bad,poor). The index provides a simple and concise method forexpressing the quality of water bodies for varied uses suchas recreation, swimming, drinking, irrigation, or fish spawn-ing, etc. The significance of the WQI can be easily appre-ciated as the water resources play a crucial role in the over-all environment and this index has also been recognized as

12 A. Lumb et al.

one of the 25 environmental performance indicators of theholistic Environmental Performance Index (EPI). The EPIis based on well-established policy categories covering bothenvironmental public health and ecosystem vitality whichfocus on climate change, water quality and quantity, air pol-lution, biodiversity, land-use changes, deforestation and sus-tainability of agriculture and fisheries (EPI 2010). Govern-ments are increasingly being asked to explain their perfor-mance on a range of pollution control and natural resourcemanagement challenges with reference to aforementionedquantitative metrics. The EPI can also provide a benchmarkfor evaluating the success of management strategies, assess-ing relative risks (poor to excellent) for sustaining a specificuse (e.g. aquatic life, irrigation, industrial or source of drink-ing water etc.), and allocating resources for abatement of keystressors on water bodies. The latter is particularly impor-tant, as it allows citizens to be informed and be able to partic-ipate in public policy debates regarding its management andbecome more likely to support policy or other actions to im-prove water quality. The 2010 EPI ranks 163 countries on 25performance indicators tracked across ten well-establishedpolicy categories covering both environmental public healthand ecosystem vitality. These indicators provide a gauge at anational government scale of how close countries are to theestablished environmental policy goals.

The importance of WQI to express the quality componentof water resources has long been recognized and various for-mulations and models have been suggested. This paper re-views the design and development of WQI based on physi-cal, chemical and biological measurements of water qualityand highlights the pros and cons of different formulationsand proposes some avenues for the future development ofwater quality indices.

Chronological History of Evolution of WQI

The concept of water quality to categorize water accordingto its degree of purity or pollution dates back to 1848 inGermany (cited by Sladecek 1973; Ott 1978; Steinhart etal. 1981; Dojlido and Best 1993). Around the same time,the importance of water quality to public health was rec-ognized in United Kingdom (Snow 1854). One of the ear-liest indices of water quality was the saprobic index (SI)which is defined as a degree to specify the loading of easilydegradable organic matter in flowing waters. Different or-ganisms have different saprobic rates and this principle wasthe basis for the determination of SI (Sladecek 1973). TheSI as a measure of the level of organic pollution was thusused in classifying water quality by various European coun-tries such as Germany, Denmark, Czechoslovakia, Hun-gary and United Kingdom (Sladecek 1973; Liebman 1969;Price 1974). The saprobic rates were determined involving

empirical studies. It was noted that all of the above schemes,rather than assigning a numerical value to represent the qual-ity of water, categorize water bodies into several pollutionclasses or levels. The system of grading the water qualityusing SI was found deficient and at times impractical and asearch continued in the succeeding decades for a better sys-tem through a numbering scheme.

Since the birth of the concept of water quality in the formof SI, it took more than a century to develop numerical in-dices to assess the quality of water. In 1965, Horton, of theOhio River Valley Water Sanitation Commission, presenteda new method in the form of an index number system for rat-ing water quality and defined a mathematical form of WQIby selecting, rating and integrating the significant physical,chemical and biological parameters of water in a simple, yetscientifically defensible manner (Horton 1965).

Horton Model of Water Quality Index in 1960s

Horton derived a WQI based on eight characteristics or pa-rameters (sewage treatment, dissolved oxygen (DO), pH,Coliform density, specific conductance, carbon chloroformextract, alkalinity and chlorides). Rating scales (0 to 100for each parameter) were assigned and each parameter wasthen weighed (weighting factor 1 to 4) according to itsrelative impact. The more significant parameter was giventhe weight of 4 such as sewage treatment, DO or pH. Pa-rameters like chloride and alkalinity were given the weightof 1. Two other parameters, namely temperature and obvi-ous pollution, appeared in the form of multiplicative factors(m1 and m2) as shown in (1). Obvious pollution here refersto tangible pollution which includes formation of sludge, de-posits, presence of oil, debris, foam, etc. that creates coloror odor nuisance. The resultant water quality index also hadvalues in the range from 0 to 100 with higher values signi-fying a better quality and vice versa.

In mathematical terms WQI was expressed as

WQI =[w1S1 + w2S2 + w3S3 + · · · + wnSn

w1 + w2 + w3 + · · · + wn

]m1m2 (1)

where S represent the rating number (also referred to assubindex) assigned to a characteristic or parameter rangingfrom 0–100, w’s are weighting factors from 1 to 4, n standsfor number of parameters used for evaluating the WQI. InHorton’s case n was equal to 8, m1 is a correction factor fortemperature (0.5 when temperature is less than 34°C, other-wise 1), m2 is the correction factor for pollution (0.5 or 1).

Water Quality Index Models from North America in 1970s

One of the challenges in Horton’s concept was selection ofthe right choice of parameters to be included in the WQI. Animproved version of the index was proposed by Brown et al.

A Review of Genesis and Evolution of Water Quality Index (WQI) and Some Future Directions 13

(1970) and Deininger and Maciunas (1971) with the supportfrom the National Sanitation Foundation (NSF) of USA.This new index is known as National Sanitation FoundationWater Quality Index (NSFWQI) and is expressed mathemat-ically as

WQI =n∑

i=1

wiSi (2)

where all notations are as defined in (1) with the weightingfactors wi ranging between 0 and 1 such that w1 + w2 +w3 + · · · + wn = 1. The index works well if all of the in-dividual parameters are independent of each other. Specialprocedures were proposed for “pesticides” and “toxic com-pounds”. If any pesticide or toxic compound exceeds its as-signed upper limit (e.g. 0.1 mg/L for pesticides), the WQI isautomatically registered as 0. The threshold levels for toxiccompounds were taken as those prescribed in the US Pub-lic Health Service Drinking Water Standards existing at thattime. Brown et al. (1971) envisioned the feasibility of usinga color spectrum scale to illustrate water quality throughoutthe entire region or state, with dark red signaling poorestquality (WQI = 0–10), a narrow band of yellow color de-picting medium quality (WQI about 50) and the dark blue(WQI = 90–100) representing the best.

The problem of selection of parameters to be included inthe formula of WQI was solved by floating questionnaires(Brown et al. 1971; Deininger and Maciunas 1971) basedon the Rand Corporation’s Delphi technique (Linstone andMurray 1975). The questionnaire was designed to combinethe opinions of a large panel of water experts. Based on therigorous survey, they came up with 11 most significant pa-rameters: DO, BOD5 (biological oxygen demand; 5-days),turbidity, total solids, nitrates, phosphates, pH, temperature,fecal Coliforms, pesticides, and toxic compounds.

It was found that arithmetic or additive formulation, al-though easy to understand and calculate, lacked sensitivityin terms of the effect a single bad parameter value wouldhave on the WQI. This led Brown et al. (1973) to propose avariation of NSFWQI in the following multiplicative form:

WQI =n∏

i=1

Swii (3)

The subsequent investigations tended to show that multi-plicative formulation agreed better with expert opinion thandid the additive one. However, both of them continued tobe in use. Briefly, in the process of selecting the parame-ters for inclusion in evaluating the WQI, marks were to beawarded to a particular parameter in terms of its importance.The most important parameter was given the highest markspurportedly of 100 or close to it and other parameters wererated relative to this parameter. The rating or the marks for a

parameter were awarded by the respondents in the question-naire and hence, involved an element of arbitration.

The other noted contribution in the development of wa-ter quality indices in US is that of McDuffie and Haney(1973) who presented a relatively simple water quality in-dex which they called the River Pollution Index (RPI). Al-though they included eight pollutant variables, either feweror more than eight variables can be included in the index de-pending upon the availability of data. The variables chosenwere: percent oxygen deficit, biodegradable organic mat-ter, refractory organic matter, Coliform count, nonvolatilesuspended solids, average nutrient excess, dissolved salts,and temperature. The RPI was computed as the sum of n

subindices, Ii, times a scaling factor = [10/(n + 1)]. Math-ematically, RPI = [10/(n + 1)](∑ I 2

i ) and the index rangesfrom 100 (natural unpolluted level) to approximately 1000(highly polluted levels). Theoretically it can go as low as 0.

Recognizing the lack of a financial accounting systemrelated with water pollution in the previous WQIs, Dinius(1972) proposed a water quality index that would quantifythe costs and impact of pollution control efforts. The con-ceptual framework followed a similar pattern as the balancesheets used by accountants to describe the assets and lia-bilities of a firm. The index used 11 water quality parame-ters viz., DO, BOD5, total Coliforms, fecal Coliforms, spe-cific conductance, chlorides, hardness, alkalinity, pH, tem-perature, and color. The index value was computed as theweighted sum of the indices, like Horton’s index and theadditive version of NSFWQI and its value ranged from 0(poorest) to 100% (perfect water quality).

About the same time, Walski and Parker (1974) pre-sented a water quality index in which quality was consid-ered specifically with respect to recreational uses of water.They chose the geometric mean (a variant of multiplicativemodel (3)) as the form of index. The parameters chosen forindex calculation were suspended solids, turbidity, nutrients,grease, color, threshold odor, pH, temperature, toxicity, andColiform count. The WQI ranged from 0 (very bad quality)to 1 (very good quality).

The index developed by Brown et al. (1970, 1973) is notreally objective because a panel of experts rates the waterquality parameters to be used (dubbed as Delphi method).There is always a chance that different panels will give dif-ferent ratings, thus lessening comparability and objectivity.The above shortcoming was overcome by an index presentedby Harkins (1974) following the methodology on nonpara-metric multivariate ranking. The major shortcoming of theHarkins’s index is that it has to be recalculated every timenew data become available because comparisons betweendata sets are not possible unless the index values are recal-culated for a merged data set of all values of interest. Thismajor drawback makes this index an untenable choice forregular use on a regional or national level (Landwehr and

14 A. Lumb et al.

Deininger 1974, 1976). The mathematical equation for theindex also tends to be unwieldy that may require effort to beintelligible at first glance.

In Canada, Inhaber (1974) worked on a concept of WQIas an offshoot of the Environmental Quality Index (EQI) onsimilar lines as that of Prati et al. (1971) from Europe. In-haber’s methodology essentially involved the normalizationof concentrations of parameters for means and averagingthem through a root-mean squaring procedure. The resultantindex ranged from 0 “perfect quality” to 1 “objective is be-ing met” to greater than 1 “quality is worse than objective”.Inhaber’s method remained obscure in Canadian usage fornot being more powerful, simpler or more intelligible thanother indices.

The major statistical look to WQI indices was givenby Landwehr (1974), whose doctoral work culminated inthe classification of the water based on the numerical val-ues of indices as follows: very bad: 0–20; bad: 21–45;medium: 46–75; good: 76–90; very good/excellent: 91–100.He also concluded that a multiplicative water quality indexwas a more viable and unbiased estimator of water qualitythat best reflected the consensus of the experts (Landwehrand Deininger 1976; Landwehr 1979). Further Landwehr(1979) concluded that the probability density functions ofthe water quality constituents as well as the structure of rat-ing curves are major determinants of choice of an appropri-ate WQI.

Some States proposed their own formulation. As a keyexample, the Oregon Department of Environment has de-veloped the original Oregon Water Quality Index (OWQI)(Dunnette 1979), which integrates the measurements ofeight water quality variables (temperature, DO, BOD, pH,ammonia + nitrate nitrogen, total phosphates, total solids,and fecal Coliforms). In the United States, the most widelyused indices for rivers are NSF based and are well docu-mented in Brown and McClelland (1974); and McClellandet al. (1976). In summary, between 1965 and early 1980s,more than 20 physico-chemical water quality indices havebeen published (Steinhart et al. 1982), mostly for rivers andstreams, besides numerous biological and trophic state in-dices in North America.

Water Quality Index Models from Europe during the 1970s

Horton’s work spurred interest among scientists worldwideto formulate suitable indices to characterize water quality.In Europe, Liebman (1969) proposed an index (known as theMunich method of water quality evaluation) based on chem-ical and biological parameters and prepared color coded wa-ter quality index maps of the state of Bavaria (western Ger-many). His method relied on the concept similar to that ofHorton but the ratings and weights were given based on indi-vidual opinion. He introduced a system of four water quality

levels which depended not only on the biological classifica-tion of living communities, but also on the overall state ofthe water for aquatic life due to oxygen content.

Since the 1970s there has been gradual growth in the de-velopment work of WQIs by using different concepts andstatistical methodologies. Contemporary to works of Brownet al. (1970) in US, efforts to develop a suitable index of wa-ter pollution were initiated in Europe (Prati et al. 1971). Theindex was primarily designed to express the degree of pollu-tion in surface waters. In developing this index, the authorsfirst reviewed the water quality classification systems thathave been adopted in England, Germany, the Soviet Union,Czechoslovakia, New Zealand, Poland and some states inthe US. The parameters chosen for quality classificationwere: pH, DO, BOD, COD (chemical oxygen demand basedon permanganate or Kubel test), suspended solids, ammo-nia, nitrates, chlorine, iron, manganese, ABS (alkyl benzenesulfonates) and CCE (carbon chloroform extract). For eachparameter, subindices were first computed by transforma-tion of concentrations into new units through mathematicalfunctions such as logarithms to the base 2, squaring, andsquare-rooting. The index was computed as the arithmeticmean of 13 subindices obtained above and it ranged (dimen-sionless number) from 0 to 14 (and sometimes above) withhigher values indicating higher level of pollution and viceversa. In other words lower numbers are good and highernumbers bad. The index was applied to data on surface wa-ters in Ferrara province, Italy (Ott 1978).

Inspired by the work in United States, particularly thatof Brown et al. (1970), the Engineering Division of theScottish Research Development Department initiated the re-search work for developing the Scottish WQI in 1973. Us-ing the questionnaire approach (essentially Delphi method),the department came up with ten parameters for generat-ing their WQI, along with weights (in parentheses), whichwere: DO (0.18), BOD (0.15), free and saline ammo-nia (0.12), pH (0.09), total oxidized nitrogen (0.08), phos-phate (0.08), suspended solids (0.07), temperature (0.05),conductivity (0.06) and E. coli (0.12). Note that sum ofthe weights is 1. Two forms of WQI formulation were rig-orously tested viz., arithmetic (WQI = ∑

wiSi , (2)) andgeometric (WQI = ∏

Swii , (3)). The additive formulation

tended to be less representative at the lower end of the qual-ity scale, whereas the geometric one was found to capturethe scenario better (note none of the parameters should havethe rating value of zero). However, it was suggested that theriver should be classified by simply quoting the WQI to thenearest whole number on the scale 0 to 100 and that the par-ticular WQI formulation (arithmetic or geometric form) bestated (Scottish Research Development Department 1976).Ross (1977) examined data for rivers in the Clyde catch-ment in Great Britain and found that the most significantdeterminants for describing variations in water quality were

A Review of Genesis and Evolution of Water Quality Index (WQI) and Some Future Directions 15

BOD, ammonia, suspended solids, and DO. The Ross indexbased on the above four parameters was intended as a broad-scale indicator of pollution, not a sensitive measure of waterquality.

The index was further improved in the form of a weightedadditive model which was advocated for use in England(Tyson and House 1989). The weighted additive model takesthe form:

WQI = 1

100

n∑i=1

(wiSi)2 (4)

At times an unweighted version has also been used for ag-gregating subindices and has been named the Aquatic Tox-icity Index (Wepener et al. 1992, 1999). In the unweightedversion of the additive model, all subindices are given equalweight such that wi = 1/n; rendering the model (4) to:WQI = [∑(Si/n)2]/100.

The other important European contribution for calculat-ing a WQI came from Spain (Bascaron 1979) using the fol-lowing equation:

WQI =(

n∑i=1

Ci ∗ Pi

)/ n∑i=1

Pi (5)

where n represents the total number of parameters, Ci is thesubindex value assigned to parameters after normalization,and Pi is the weight assigned to parameters (an indicator ofits relative importance for aquatic life/human water use). Itcan be noted that the major ingredient of the above equa-tion is the normalization which has not been used in NSFbased formulations. In several studies emanating from Eu-rope, Bascaron WQI has been widely used in tandem withthe Scottish WQI.

Water Quality Index Models During 1980s

Recognizing that the trophic state of lakes were also an im-portant aspect of water quality but remarkably different fromthe rivers, Steinhart et al. (1982), developed the Environmen-tal Quality Index (EQI) for the Great Lakes of North Amer-ica. Nine variables representing physical (P ), chemical (C),biological (B) and toxic (T ) features were selected for theindex. The variables were: specific conductance, chloride,total phosphorus, fecal Coliforms, chlorophyll a, suspendedsolids, obvious pollution (aesthetic state), toxic inorganiccontaminants, and toxic organic contaminants. Raw datawere converted to subindex values by mathematically de-fined functions based on national or international yardsticksor thresholds. Sub-index values were multiplied by weight-ing factors (a value of 0.1 for chemical, physical and bio-logical factors but 0.15 for toxic substances) and added toyield a final score ranging from 0 (poorest quality) to 100(best quality) with the breakdown as poor <55, fair: 55–69,

good: 70–79, very good: 80–89, excellent: 90–100. Each in-dex number was followed by a letter C, P , B or T indicat-ing which subindex values were most problematic, i.e. lessthan or equal to 50 (threshold). A numerical subscript bythe letter indicates how many in that class are affected. Forexample 70C1P1 indicates that one chemical and one physi-cal variable did not meet the objective criterion. Steinhart etal. (1982) found that for 18 nearshore locations in the GreatLakes, index scores ranged from 98 at two locations in LakeSuperior to 30 (30C2P1B2T3) off Point Mouillee in LakeErie. They projected the utility of the index in evaluating theeffectiveness of the multibillion dollar Great Lakes cleanupefforts conducted during the 1970s.

Most of the work cited above tended to have greater rele-vance to the uses of water for aquatic life or recreational useswith subtle reference to drinking water applications. Bhar-gava (1985) derived the ideas from the concept of WQI eval-uation advanced by Brown et al. (1970) for classifying thewater quality exclusively for drinking purposes. He, how-ever, used the following form of a multiplicative model:

WQI =[

n∏i=1

f i

]1/n

(6)

in which, f i = the sensitivity function value of the ith vari-able (parameter) which included the effect of the concentra-tion and weight of the ith variable in use and varied from 0–1and n is the number of variables considered. Curves basedon requirements of the WQI and involving the weighting ef-fect of each variable on the various uses of water were plot-ted and WQI computed, thus were used for the classifica-tion of river waters for different beneficial uses. The effectson the WQI, due to changes in the concentration of a singlevariable, were depicted through curves to illustrate the effectof different weighs of a variable for different uses.

The variables were divided into groups. The first groupincluded the concentration of Coliform organisms to repre-sent the bacterial quality of drinking water. This variable hasa direct implication on the health of the consumer, and can-not be allowed in excess of the standards set by the variousauthorities. The sensitivity function for this should, there-fore, fall rapidly to a level such that the WQI is signifi-cantly lowered to acceptable levels, i.e., when the concentra-tion of Coliforms exceed permissible level and become dan-gerous. The second group of variables included toxicants,heavy metals, etc. and their permissible concentrations arebased on the physiological effects associated with symp-toms related to concentration levels of these variables. Forsuch variables, a slight deviation from the permissible lev-els may be allowed, to the extent of about twice the allow-able level. The third group of variables includes the materi-als that cause physical effects such as odor, color, turbidity,and the other aesthetic qualities which are important factors

16 A. Lumb et al.

in the public’s acceptance and confidence in a public wa-ter supply system. Their concentrations relate to palatabilityof the water and an excess of these variables would be dis-liked but would not be dangerous for health. The sensitivityfunction for these variables should gradually fall off whentheir concentration exceeds the permissible levels and a de-viation up to three times the permissible limit is allowed forsuch variable. The fourth group of variables includes the or-ganic or inorganic non-toxic substances such as chlorides,sulfates, foaming agents, iron, manganese, zinc, copper, andtotal dissolved solids (TDS), which in some cases addresssimilar concerns as for the third group. Their values exceed-ing permissible limits are not at all dangerous. There are infact health-based drinking water standards for some of thesevariables, including iron, manganese, zinc and copper. How-ever, an upper limit can be established based on local con-ditions. The water quality standards for United States Envi-ronmental Protection Agency existing in 1982 were used inthe analysis. This model was also tested for the drinking wa-ter supply for the city of Delhi, India and worked well. Fordrinking water, a value of WQI of 90 or above based on theabove model was acceptable.

One of the challenges that spurred attention of the waterquality investigators was the selection of the significant pol-lutants and their level of concentration, which was endemicduring the 1970s and 1980s in the fresh waters. Dinius(1987) addressed this problem by evolving a four—roundDelphi process. The inclusion of representative pollutants inthe index, and the relationship between the quantity of thesepollutants in the water and the resulting quality of the waterinvolved consultations among members of a panel of waterquality experts. The panel used a general rating scale as aframe of reference to establish the rate of change in the nu-merical index as the quality of water changed. Evaluationlevels were gathered for six separate water uses (public wa-ter supply, recreation, fish, shellfish, agriculture, and indus-try), but parsimony, pragmatism, and utility suggested theindices be aggregated into one generalized index. A mul-tiplicative index of the form of (3) was used to bring thepollutants together in one system.

A significant advance was made by Smith (1990) in NewZealand, who developed a WQI for four water uses: bathing,water supply, fish spawning and general uses. The salientfeature of Smith’s work has been to make a better use of thewater quality parameter giving the lowest score (or lowestsubindex value) in order to arrive at the final score. The in-dexing system at that point of time tended to integrate expertopinion and water quality standards. The parameters usedfor the WQI evaluation were: DO, suspended solids, turbid-ity, temperature, BOD5 (unfiltered), ammonia, and fecal Co-liforms with the exception that ammonia was only includedin water supply and fecal Coliforms were not included in fishspawning. The index value ranged from 0 to 100 with 80 and

above eminently suitable for all uses and less than 20 totallyunfit for nearly all uses. The system was simple to use andaddition (or subtraction) of determinants (water quality pa-rameters) was a much simpler task. The index was used forwater quality legislation and dissemination of water qual-ity information in New Zealand. Several water authorities inNew Zealand are presently using this as a planning tool andas a simple means of disseminating water quality informa-tion.

In the United Kingdom, House (1989, 1990) suggestedanother water quality index which conceptually was simi-lar to NSFWQI. The following parameters and their rela-tive weightings were used in the calculations: DO (0.20),BOD (0.18), ammonia nitrogen (0.16), total Coliforms(0.11), suspended solids (0.11), pH (0.09), nitrate (0.09),chloride (0.04), and temperature (0.09). These weights wereestablished by using a questionnaire (Delphi procedure)which was sent out to operational management personnelin the pollution prevention organization of the UK. House’sgraphs were established from various quality standards suchas EC (European commission) directives or maximum de-sirable concentrations. The WQI values result in the rangefrom 10 to 100 with higher values pointing to a better qual-ity.

Water Quality Index Models during the 1990s and 2000s

The Florida Stream Water Quality Index (FWQI) was de-veloped in 1995 under the Strategic Assessment of Florida’sEnvironment Indicators Project (SAFE 1995). It is an arith-metic average of water clarity, turbidity, total suspendedsolids, dissolved oxygen, BOD, COD, total organic carbon,nutrients (phosphorus and nitrogen), bacteria (total and fe-cal Coliforms), and biological diversity. Values of this indexclassified the water as follows: good: 0 to <45, fair: 45 to 60,poor: 60 to 90.

At the same time, attempts to refine the mathematicalstructure of the water quality index proposed by Brown etal. during the 1970s were still continuing with the additionalinput from Dojlido et al. (1994) in the form of a harmonicmean square root formula, or simply known as the harmonicmodel formula, expressed as:

WQI =[

1

n

n∑i=1

S−2i

]−0.5

(7)

Equation (7) has shown promise for the Vistula River inPoland, and water quality classes have been defined as: veryclean: 75–100; clean: 50–75; polluted: 25–50; and very pol-luted: 0–25.

The Oregon Water Quality Index (OWQI) also takes theform as represented by (7). The details of the index includ-ing the rating functions for various parameters are well doc-umented in Cude (2001, 2005, 2008). The water quality

A Review of Genesis and Evolution of Water Quality Index (WQI) and Some Future Directions 17

classes in terms of OWQI scores are different from thosebased on NSFWQI shown by (2) and (3). The water qual-ity classes based on OWQI scores are excellent: 91–100;good: 85–90; fair: 80–84; poor: 60–70; and very poor: 0–59.The harmonic model shown in (7) looks promising, althoughthe band of OWQI scores is too narrow in good and fair cat-egories.

One major gap encountered in many WQIs until the be-ginning of 1990s was the problem where all the subindicesare considered acceptable yet the overall index is not. In thesame vein an eclipsing problem existed where the overall in-dex is insensitive to a single crucial variable (or parameter).This problem was addressed by Swamee and Tyagi (2000)among others and they proposed full range of equations forwater quality subindices for a number of water quality pa-rameters in order to facilitate the computer adaptation of theaggregation process.

In Canada, a new water quality index was introduced inthe mid 1990s by the province of British Columbia (Roc-chini and Swain 1995; Zandbergen and Hall 1998) and usedit as the basis for reporting to the public and identifying wa-tersheds for priority action, such as by Manitoba Ministry ofEnvironment (SOE Report 1997). Recognizing the need toassess the suitability of water for diverse uses in tandem withair quality, the Water Quality Guidelines Task Group of theCanadian Council of Ministers of the Environment (CCME)modified the original British Columbia Water Quality In-dex and endorsed it as the CCME WQI in 2001. This in-dex was originally called the Canadian Water Quality Index(CWQI) as referred in CCME (2001). Since its inception andendorsement for use in Canadian jurisdictions, the CCMEWQI has been implemented in British Columbia, Alberta,Saskatchewan, Manitoba, Northern territories, the Atlanticprovinces and Newfoundland & Labrador (Cash et al. 2001;CCME 2001; Husain 2001; Sharma 2002; Khan et al. 2003;Mercier and Leger 2003) and nationally (CESI 2008). Since2005 there has been a spurt of studies published on the ap-plication of CCME WQI supporting its widespread use andversatility (Khan et al. 2005; Environment Canada 2005;Mercier and Leger 2006; Dube et al. 2006; Lumb et al. 2006;Tobin et al. 2007; Statistics Canada 2008; Rickwood andCarr 2009; de Rosemond et al. 2009). From the experiencewith the CCME WQI implementation, both strengths andchallenges of the index have come to light, including is-sues related to monitoring, communication and public ex-pectation, inputs, interpretation, and the long-term feasibil-ity studies of the quality aspects of water resources and fi-nancial commitment by governments.

Conceptually CCME WQI comprises three factors andis well documented (CCME 2001). Factor 1 (F1) deals withscope that assesses the extent of water quality guideline non-compliance over the time period of interest. Factor 2 (F2)

deals with frequency i.e. how many occasions the tested or

observed value was off the acceptable limits or the yard-sticks. Factor 3 (F3) deals with the amplitude of deviationor the amount by which the objectives are not met. The in-dex value is computed using the following formulation:

CCMEWQI = 100 −(√

F 21 + F 2

2 + F 23

1.732

)(8)

The factor of 1.732 has been introduced to scale the indexfrom 0 to 100. The above formulation produces a value ofCCME WQI between 0 and 100 and gives a numerical valueto the state of water quality. Note a zero (0) value signifiesvery poor water quality, whereas a value close to 100 signi-fies excellent water quality. The assignment of CCME WQIvalues to different categories is somewhat subjective processand also demands expert judgment and public’s expectationsof water quality. The water quality is ranked in the followingfive categories: excellent: 95–100; good: 80–94; fair: 65-79;marginal: 45–64; poor: 0–44.

An equally noteworthy contribution in the form of thewater quality index dubbed as Indice de qualite bacteri-ologique et physicochimique (IQBP) has been tested rig-orously in the province of Quebec, Canada (Hèbert 1996,2005) with the following formulation:

IQBP = min(IF1, IF2, IF3, . . . , IF7) (9)

in which IF1 = subindex for fecal Coliform; IF2 = subindexfor total phosphorus; IF3 = subindex for nitrite-nitrate,IF4 = subindex for ammonical nitrogen; IF5 = subindexfor chlorophyll; IF6 = subindex for turbidity; and IF7 =subindex for suspended solids. A value of subindex rangesfrom 0 (very poor) to 100 (good). IQBP varies between0–100 and five water quality classes are defined: good(80–100; generally good for all uses including bathing);satisfactory (60–79; generally satisfactory for most uses);questionable (40–59; questionable quality with some usesmay compromise); bad (20–39; poor water quality, mostuses may compromise); very poor (0–19; very poor waterquality, all uses may compromise). The IQBP was primarilymeant to assess the water quality for swimming and recre-ational activities (E. coli as key parameter), and protectionof aquatic life including protection against eutrophication.IQBP was used in the 2000s for assessing all major rivers inQuebec, Canada.

One noteworthy development in the recent years has beenthe introduction of statistical multivariate analysis for dis-cerning the significant parameters for evaluation of the wa-ter quality index (Liou et al. 2004; Debels et al. 2005;Qian et al. 2007; Sedeno-Diaz and Lopez-Lopez 2007;Saeedi et al. 2010) vis-à-vis the traditional Delphi method.The first decade of 21st century has also witnessed theemergence of software for computing water quality indices(Sarkar and Abbasi 2006). Also some objective methods

18 A. Lumb et al.

based on mathematical logic have been suggested to re-place the Delphi method of selection of parameters and theirweights (Parparov and Hambright 2007; Kumar and Dua2009).

In the USA, the NSFWQI continues to enjoy the status ofa popular indicator of water quality, which has been furtherassessed and altered to address the specific needs of individ-ual states. The notable variants of NSFWQI such as OregonWater Quality Index (OWQI) is used to aid in the assessmentof water quality for fishing and swimming and in turn towater quality management in major streams (Mrazik 2007;Cude 2008). One important feature of the OWQI is that itincludes E. coli as the bacterial parameter for assessment ofthe index. On lines similar to Oregon, states of Washing-ton (Hallock 2002), California (Thomson et al. 2007), andIowa (Iowa DNR Report 2006) etc. have adopted modifiedversions of NSFWQI to suit their needs. Another index dif-ferent from NSFWQI was introduced by Said et al. (2004)from the University of South Florida and its mathematicalform is expressed as:

WQI = log

((DO)1.5

(3.8)TP(Turb)0.15(15)Fcol./10000 + 0.14(SC)0.5

)

(10)

where DO is the dissolved oxygen (% saturation), Turb is theTurbidity (Nephelometric turbidity units, NTU); TP is thetotal phosphates (mg/L); Fcoli is the fecal Coliform bacteria(counts/100 ml); and SC is the specific conductivity (MS/cmat 25°C).

One recent model of water quality index is put forthby the Malaysian Department of Environment (Shuhaimi-Othman et al. 2007). The mathematical structure of themodel is as follows:

WQI = 0.22(DOsi) + 0.19(BODsi) + 0.16(CODsi)

+ 0.15(ANsi) + 0.16(SSsi) + 0.12(pHsi) (11)

where subscript si stands for the subindex function associ-ated with each of these parameters (DO, BOD, COD, am-monical nitrogen, suspended solids and pH).

A few other indexing schemes have also been devel-oped deriving the ideas from the pioneering NSF based con-cepts and those coming from Europe. One such index isthat of Fulazzaky (2009). The WQI scores based on thismethod classifies the water in five classes viz. excellent: 80–100; good: 60–80; moderate: 40–60; bad: 20–40; and verybad: 0–20. The method has been successfully used to clas-sify the water quality of Selangor River in Malaysia (Fu-lazzaky et al. 2010). The concepts based on grey relationalmethod (Ip et al. 2009) and fuzzy logic analysis (Kung et al.1992; Lermontov et al. 2009; Jinturkar et al. 2010) as usedin hydrology have also been introduced in developing the

WQI. The other novel attempt in the recent past has been theintroduction of the probabilistic approach in development ofwater quality index for river quality assessment (Nikoo et al.2010), which entailed the concepts of fuzzy inference sys-tems, Bayesian networks and probabilistic neural networks.These concepts have been used in modeling the hydrolog-ical processes and water resources systems. The approachworked well for water quality index evaluation of the Ja-jrood River, Iran (Nikoo et al. 2010).

The emerging technology of remote sensing has alsobeen introduced in the realm of water quality indices. Onesuch attempt is that of Vignolo et al. (2006), who have suc-cessfully used this technique to assess the water quality ofthe Medrano Creek, Argentina. In particular, the techniqueseemed to have strong potential for tracing the organic con-tamination associated with dyes in fresh water systems.

There are other water quality indices which in principleare derived from the NSF based concept. One such WQI(the score range 0 to 10) has been used to evaluate the qual-ity of the Mekong River for aquatic life, human impact andagricultural uses (MRC 2008). For instance for aquatic usesthe water quality classes are: high quality (WQI: 10–9);good quality (WQI: 9–9.5); moderate quality (WQI: 7–9);and poor quality (WQI: <7). Different score ranges havebeen applied to categorize the Mekong River water for agri-cultural uses and human impact evaluations (MRC 2008).Likewise Tipping et al. (2002) have proposed a concept ofsustainable development indicators for managing the waterquality in the Mekong River. Another recent developmenthas been the emergence of some macro-invertebrate basedbiotic indices (Bhatt and Pandit 2010). Succinctly duringthe first decade of 2000, competing models do exist as con-sequences of different legal and political context on imple-mentation, environmental concerns, data availability, multi-ple ways of formulating mathematical relationships betweenvariables, and mannerism on inculcating the role of expert inthe models.

A Note on Indexing the Ground Water Quality During the1990s and 2000s

Most of the work on the WQI has been devoted to surfacewater such as streams and lakes with the intent of classify-ing the water for aquatic and recreational uses. There havebeen some attempts to categorize the water for drinking pur-poses, particularly the groundwater which comprises a ma-jor source for drinking applications. To determine the suit-ability of the groundwater for drinking purposes, WQI iscomputed by adopting the method which is formulated as(Tiwari and Mishra 1985):

WQI = anti logn∑

i=1

wi logqi (12)

A Review of Genesis and Evolution of Water Quality Index (WQI) and Some Future Directions 19

where wi is weight or the weighting factor = K/Oi and K isa constant = 1/(

∑1/Oi); and Oi correspond to the WHO

(World Health Organization) or Indian Council of MedicalResearch (ICMR) standard value of the parameter, which istantamount to the objective function or the threshold value.

The quality rating (qi) is calculated as

qi = [(Vactual − Videal)/(Vstandard − Vactual)

] × 100 (13)

where qi = quality rating of ith parameter for a total ofn water samples; Vactual = value of the water quality pa-rameter obtained from the laboratory analysis; Vstandard =value of the water quality parameter obtained from the stan-dard tables; Videal for pH = 7 and for the other parametersit is equivalent to zero. The index was specifically devel-oped for ranking ground water for drinking purposes andhas been used in subsequent years implying its acceptance(Ramachandramoorthy et al. 2010) on the Indian subconti-nent.

Backman et al. (1998) presented an index for evalua-tion and mapping the degree of groundwater contamina-tion and its applicability in southwestern Finland and cen-tral Slovakia. A simple WQI involving nine parameters iscreated by Soltan (1999) to indicate the quality of ground-water from 10 artesian wells located near Dakhla Oasis inthe Egypt. Likewise Stambuk-Giljanovic (1999) has indexedgroundwater in conjunction with surface water in Dalma-tia, Crotia. The robust indexing approach for ground wa-ter has been developed by Stingter et al. (2006). The mostrecent contribution in developing the WQI for ground wa-ter is from Iran (Saeedi et al. 2010) in which a similar ap-proach as that of NSF-WQI is adopted. In stead of using thesubindex values they have normalized the concentrations ofeach parameter and an additive aggregation model is used toevolve the index. The weighting factors for each parameterare chosen based on their perceived importance. The index(GWQI, ground water quality index) values range from 0to 1 with water quality classes as high (GWQI > 0.15); suit-able (GWQI between 0.04 and 0.15) and low (GWQI <

0.04).

Reviewing the Current Use of Various WQI

There is no standardized use or unique convergence towarda particular model for indexing the quality of water for aparticular use. The use of NSFWQI or its variants such asBhargava’s (1985) WQI model for drinking water are popu-lar not only in USA but in many other countries. The studiesfrom Croatia suggest the superiority of the geometric formof NSFWQI (Stambuk-Giljanovic 1999, 2003). Other stud-ies have been reported in the literature demonstrating the useof NSF based indices across countries worldwide (Abrahao

et al. 2007—Brazil; Sedeno-Diaz and Lopez-Lopez 2007—Mexico; Bordalo and Savva-Bordalo 2007—Guinea-Bissau;Stojda and Dojlido 1983—Poland; Soltan 1999—Egypt;Bordalo et al. 2006—Portugal; Giuseppe and Guidice 2010–Italy; Avvannavar and Shrihari 2008; and Chaturvedi andBhasin 2010—India).

The indexing approach from Smith (1990) and Nagleset al. (2001) has been implemented to quantify the qualityof waters of Lake Kinneret (Israel) and the Naroch lakesof Belarus. Some work has also been initiated to index thewaters for recreational uses in New Zealand (Nagles et al.2001). The Bascaron model of WQI has found acceptancein Latin American countries (Pesce and Wunderlin 2000 inArgentina; Debels et al. 2005 in Chile). The other impor-tant feature inherent in these applications is to weigh theparameters for their relative importance and include themin calculation for the composite index (Pesce and Wunder-lin 2000; Stambuk-Giljanovic 2003; Sargaonkar and Desh-pande 2003; Tsegaye et al. 2006).

The Water Framework Directive (WFD) was adopted in2000 and is viewed as a substantial piece of the EuropeanCommunity (EC) legislation that establishes an integratedapproach to the protection, improvement, and sustainableuse of Europe’s water (surface and ground). Of special men-tion is the formation of the Technical Advisory Group in2001 (UKTAG 2007) to assist in implementing the agendaof WFD within United Kingdom. Yet, no clear and uniqueindex formulation has emerged, but the river classificationsystem is one of the key aspects related to EU Framework.

Through the Canadian Environmental Sustainability In-dicators (CESI) initiative, the Canadian government reportson sustainable environmental indicators that track three is-sues of key concern to Canadians which are Air Quality,Water Quality and Greenhouse Gas Emissions (CESI 2008).The CCME WQI continues to be used in Canada on a na-tional basis and by provincial and territorial environmentdepartments and watershed organizations. It has also sup-ported development of subindices for nutrients, pesticides,bacteria and metals in Alberta for reporting in 2005–2006and 2006–2007 (AB 2007). The existing Canadian index isnot perfect and there is always room for improvement basedon the historical development of WQIs since the 1960s. Themajor strength of the CCME WQI is that it is able to takethe flexible number of parameters sampled at irregular timeintervals, and choose relevant parameters at each site withsite specific guidelines. Technically, it can also allow forinclusion of many variables based on existing environmen-tal quality guidelines or management objectives (e.g. wa-ter column, but also sediment and fish community as wasdone in the past in British Columbia). However, the indexrequires at least four parameters sampled four times duringthe desired time span in context of a 3 years roll-up av-erage (more frequent sampling if WQI would become an-nual). In terms of main weaknesses, the Canadian WQI,

20 A. Lumb et al.

as the vast majority of other indices, suffers from the in-sensitivity toward a particular parameter in the process ofaggregation, which mutes its role. That means that a lowscore for one variable that is severely limiting water use maybe masked when aggregated with relatively high scores forother variables. For example, a high level of fecal contam-ination should make water unsuitable for bathing, but thecorresponding low index score for fecal indicator bacteriamay be “swamped” in the overall index by a high score forthe other variables. The above noted deficiency was circum-vented in the late 1990s and early part of the first decade of21st century by using a subindex approach (Bhargava 1985;Nagles et al. 2001; Hèbert 2005). The WQI model can workin complementary mode with subindex as part of a dash-board approach to water quality, and also support key pa-rameters for trends analysis. It would be feasible by estab-lishing connection between the overall index and measure-ments of water quality ingredients. The water quality as-sessed using the CCME WQI that confirmed its suitabilityfor drinking, aquatic life, irrigation and other purposes hasbeen applied not only in Canada but also in many othercountries all over the world (Zoabi and Gueddari 2008;Aloui and Gueddari 2009 in Tunisia; Avvannavar and Shri-hari 2008; Joseph and Parameswaran 2005 in India; Boya-cioglu 2007, 2010 in Turkey).

Globally, much of the work being conducted on wa-ter quality is linked to one of the UN Millennium Devel-opment Goals (UNEP GEMS 2007): sustainable access tosafe drinking water by world population by 2015. Anotherinitiative from Mekong River Commission (MRC 2008),which has established the drinking water categories basedon their own WQIs to assess the water quality of LowerMekong Basin for human use. GEMS/Water Program sup-ported the development of a global drinking water qualityindex (GDWQI) developed by Rickwood and Carr (2009),based on CCME WQI model. Notwithstanding developmentof other indices, it seems that the CCME WQI still holdssome promise in terms of improvements for the future andcan support additional works.

Future Research Directions

More of a constraint than an intrinsic weakness, the ef-ficiency and the accuracy of all indices bank on existingmonitoring network, prevalent methods of physico-chemicalanalysis and guidelines. Notably, it is prudent that a long-term core network for indexing status and trends (e.g. defin-ing selection criteria and implementing them) be estab-lished. Also, there exists imminent need to develop morecost-effective ways of monitoring (e.g. automated sampling,rotational base sampling) of water quality parameters. Au-tomated samplers, which typically measure a more limited

suite of parameters (e.g. DO, pH, turbidity, temperature andconductivity) continue to expand in popularity. The suite ofmeasured parameters also continues to expand with tech-nological advances, as do the ability to predict the valuesof non-measured parameters through correlation modeling.The potential thus exists to translate these data into informa-tion for the public on a real-time basis through indices as isbeing done for air quality in many cities around the world.

Depending upon the guidelines, conflicting interpretationand messages can emerge on the quality characteristics ofthe water. To reduce such a problem, efforts should be di-rected to develop site-specific guidelines adapted to the pre-vailing ecological conditions (e.g. taking into considerationhigh background concentration of phosphorus in the Prairiesof Canada). In some cases, it is necessary to not just con-sider toxic or acute effects of contaminants, but to look forchronic effects such as eutrophication delineated through amore ecological approach applied to background level of nu-trients in the natural waters. This calls for extended partner-ship and cooperation between various institutions dealingwith water for implementation and development of the in-dexing methodologies and approaches uniformly.

As the suitability of water can vary greatly depending onits intended use, there is a need to provide use-specific as-sessments (e.g. for aquatic and recreation uses versus sourcewater use for human consumption). For being more trans-parent with the level of protection related to water qualityand making connection with policy, one would need to useat least one index by use (cannot use one index that “fits all”water uses). It would thus be important for the indices tobe specifically adapted to provide the required informationon each use, and for long- term monitoring networks for in-dexing status and trends to consider these aspects in theirdesign.

Evolution of these indices can also take the form ofsubindices based on parameters by types of effects (e.g. eu-trophication or bio-accumulative property) or sources (e.g.agricultural or industrial) to provide information on morespecific water quality issues (e.g. lakes or rivers) and pre-sented as a type of dashboard to the public and for priori-tizing actions. Although the development of such an indexposes a challenge, it is feasible by establishing connectionbetween the overall index and measurements of water qual-ity parameters. With the continued growth in water qualitydata, the selection of the water quality parameters for indexcalculation should be based on more robust techniques suchas the statistical multivariate analysis rather than the sub-jective Delphi method (as used in many states). This wouldimprove robustness of WQIs (and support better knowledgeof intervals of confidence) and also credibility of status andtrends regarding water quality objectives and guidelines.

For aquatic life, the scope of WQI could also be enhancedby incorporating the biological indicators that must form the

A Review of Genesis and Evolution of Water Quality Index (WQI) and Some Future Directions 21

part of the routine monitoring network. Index of Biotic In-tegrity (IBI) was first developed by Karr (1981) to help re-source managers to assess the biological health (conditionof living organisms, benthic, native fish or algae etc.) of thewatersheds, rivers, lakes and streams. To characterize ben-thic conditions, macro-invertebrate index of biotic integrity(M-IBI) and multimetric index and more recently the Refer-ence Condition Approach (RCA) has been widely used bybenthic ecologists (Reynoldson et al. 1997, 2003). This typeof index could either be used as such and as complement toWQI, or be aggregated to form a more comprehensive index.

In a comparative study of 36 WQIs and WPIs (Water Pol-lution Indices) it was noticed that appreciable differencesexist between classifications given by different indices onthe same water sample (Fernandez et al. 2004). These dif-ferences arise primarily because of differing parameter num-bers, calculations and aggregation formulas. Overall, apply-ing WQI to water quality data must be done with due regardto how the index is formulated and used for the purposes itwas designed for. There is no simple procedure to comparethe performance of indices, but one would look at comple-mentarities of the information, credibility of measurements,transparency of indices formulation, relevancy of key pa-rameters selected and comparability of results, looking forstatus and trends.

There is need to launch a study to compare the perfor-mance of various water WQI models in terms of efficiency,adequacy, parsimony and flexibility with available data. Sev-eral countries, including Canada, possess data on water qual-ity parameters that are lying in archives and such data shouldbe used in developing a universal WQI. In an equal vein, thescore system used in various WQI models must be linkedappropriately to reflect on the right class of water quality.

Concluding Remarks

The concept of water quality index is a relatively new, andHorton (1965) is to be credited to give a first formal defi-nition to it. Since then numerous variations of water qual-ity indices have been addressed in literature over the pastfive decades. The NSF based additive and multiplicativeWQI formulations (Brown et al. 1970, 1973) seem to rulethe roost with modifications from various countries. A dis-tinctly diverse concept has been introduced by Canada in de-veloping WQI (CCME 2001), which is widely used acrossprovinces in Canada, and is gaining popularity globally.Most indices rely on normalizing/standardizing data param-eter by parameter according to expected concentrations andsome interpretation of “good” versus “bad” concentrations.Parameters are often then weighted according to their per-ceived importance to overall water quality and the indexis calculated as the weighted average of all observations

of interest, and therefore dependant on existing monitor-ing data. The ranking of water quality classes based on theWQI scores from various WQI models is also not uniform.The commonly used NSF based and CCME based scoresrange from 0 to 100 alluding higher scores to a better qual-ity. There is no unique model of water quality index that hasbeen universally applicable.

In many cases of index evaluation the parameter selec-tion has been made using Delphi method. It is only in thecurrent decade that statistical tool of multivariate analysishas been introduced to perform this task. The majority ofindices involve nine or more parameters with commonal-ity of DO, BOD, pH, total dissolved solids, and fecal Co-liforms or E. coli. The indices are largely developed for sur-face waters (rivers and lakes) with a greater bias to aquaticuses (Simões et al. 2008) and less intent to drinking, recre-ational and irrigational uses. The indexing for agriculturalapplications is at an infant stage (Cooper et al. 1998; Wrightet al. 1999; Almeida et al. 2008; NL 2008). Some attemptsexist on indexing ground water quality for drinking appli-cations. There is no globally acceptable composite index ofwater quality, though some countries and regions have usedaggregated water quality data in the development of WQI.

In the future, it may be expected that additional effortfor integration will be fostered in the context of WQIs, in asomewhat similar fashion as with economic indicators sincethe 1930s—from a global index to more specific and ex-tended subindices and dashboard indicators for sustainablesocieties. There is immediate need to compare the effec-tiveness, ease, adequacy etc. of the existing WQI models toevolve a universally applicable model flexible enough to cutacross the available data for assessing the water quality forvaried uses.

Acknowledgements The authors wish to thank Vincent Mercier, En-vironmental Indicator and Reporting Specialist, Environment Canada,Environmental Science Centre, University of Moncton, Moncton, NB,Canada, E1A 3E9, for very constructive suggestions and a critical re-view of the manuscript.

References

AB (2007) Water for life: healthy aquatic ecosystem informationsynthesis and initial assessment of the status and health ofaquatic ecosystems in Alberta: surface water quality, sedimentquality and non-fish biota. Technical Report # 278/279-01, Al-berta Environment, Edmonton, AB, Canada. Can be viewed athttp://environment.gov.ab.ca/info/library/7861.pdf

Abrahao R, Carvalho M, da Silva Junior WR, Machado TTV, GadelhaCLM, Hernadez MIM (2007) Use of index analysis to evaluatethe water quality of a stream receiving industrial effluents. WaterSA 33:459–466

Almeida C, Quintar S, Gonzalez P, Mallea M (2008) Assessment ofirrigation water quality: a proposal of a quality profile. EnvironMonit Assess 142:149–152

22 A. Lumb et al.

Aloui BZ, Gueddari M (2009) Long-term water quality monitoring ofthe Sejnane reservoir in northeast Tunisia. Bull Eng Geol Environ68:307–316

Avvannavar SM, Shrihari S (2008) Evaluation of water quality indexfor drinking water purposes for river Natravati, Manglore, SouthIndia. Environ Monit Assess 143:279–290

Backman B, Bodis D, Lahermo P, Rampant S, Tarvainen T (1998) Ap-plication of a groundwater contamination index in Finland andSlovakia. Environ Geol 36(1–2):55–64

Bascaron M (1979) Establishment of a methodology for the determi-nation of water quality. Bol Inf Medio Ambient 9:30–51. CIMA,MOPU, Madrid

Bhargava DS (1985) Expression for drinking water supply standards.J Environ Eng. Div 111(3):304–316

Bhatt JP, Pandit MK (2010) A macro-invertebrate based new bioticindex to monitor river quality. Curr Sci 99(2):196–203

Bordalo AA, Savva-Bordalo J (2007) The quest for safe drinking wa-ter: an example from Guinea-Bissau (West Africa). Water Res41:2978–2986

Bordalo AA, Teixeira R, Wiebe W (2006) A water quality index ap-plied to an international shared river basin: the case of the DouroRiver. Environ Manag 38:910–920

Boyacioglu H (2007) Development of a water quality index based on aEuropean classification scheme. Water SA 33:101–106

Boyacioglu H (2010) Utilization of the water quality index methodas a classification tool. Environ Monit Assess 167:115–124.doi:10.1007/s10661-009-1035-1

Brown RM, McClelland NI (1974) Up from chaos: the water qualityindex as an effective instrument in water quality management. Na-tional Sanitation Foundation, Ann Arbor, p 27

Brown RM, McClelland NI, Deininger RA, Tozer RG (1970) A waterquality index—Do we dare? Water Sew Works 117(10):339–343

Brown RM, McClelland NI, Deininger RA, O’Connor M (1971) A wa-ter quality index—crashing the psychological barrier. Presented at138th meeting of the American association for the advancementof science, Philadelphia, Pennsylvania

Brown RM, McClelland NI, Deininger RA, Landwehr JM (1973). Vali-dating the WQI. The paper presented at national meeting of Amer-ican society of civil engineers on water resources engineering,Washington, DC

Cash KJ, Saffran KA, Wright CR (2001) Application of Canadian wa-ter quality index (CWQI) to PPWB monitoring program. Techni-cal Report, March 2001

CCME (2001) Canadian water quality guidelines for the protection ofaquatic life. CCME water quality index 1.0, User’s Manual, Win-nipeg, Manitoba, Canada. Can be viewed at http://www.ccme.ca/assets/pdf/wqi_usermanualfctsht_e.pdf

CESI (2008) Water quality: national perspective. EnvironmentCanada. It can be viewed at http://ec.gc.ca/indicateurs-indicators/default.asp?lang=En&n=A073189E-1

Chaturvedi MK, Bhasin JK (2010) Assessing the water qual-ity index of water treatment plant and bore holes inDelhi, India. Environ Monit Assess 163:449–453. doi:10.1007/s10661-009-0848-2

Cooper J, Rediske R, Northup M, Thogerson M, van Denend A (1998)Agricultural water quality index. Robert B Annis Water ResourcesInstitute, Grand Valley State University, Allendale, p 75. WRIPublication # MR-98-1

Cude CG (2001) Oregon water quality index: a tool for evaluating wa-ter quality management effectiveness. J Am Water Resour Assoc37(1):125–137

Cude CG (2005) Accommodating change of bacterial indicators inlong term water quality data sets. J Am Water Resour Assoc41(1):47–54

Cude CG (2008) Interpretation and communication of water qualitydata using the Oregon water quality index (OWQI). State of Ore-

gon Department of Environmental Quality, Laboratory and En-vironmental Assessment Division, Portland. Can be viewed atwww.deq.state.or.us/lab/wqm/wqindex.htm

de Rosemond S, Duro DC, Dube M (2009) Comparative analysis of re-gional water quality in Canada using water quality index. EnvironMonit Assess 56:23–240. doi:10.1007/s10661-008-0480-6

Debels P, Figueroa R, Urrutia R, Barra R, Niell X (2005) Evaluationof water quality in the Chillan river (Central Chile), using physi-cal and chemical parameters and a modified water quality index.Environ Monit Assess 110:301–322

Deininger RA, Maciunas JM (1971) Water quality index for public wa-ter supplies. Department of Environment and Industrial Health,University of Michigan, Ann Arbor

Dinius SH (1972) A social accounting system for evaluating water.Water Resour Res 8(5):1159–1177

Dinius SH (1987) Design of water quality index. Water Resour Bull,Am Water Resour Assoc 23(5):823–843

Dojlido JR, Best GA (1993) Chemistry of water and water pollution.Ellis Horwood, London

Dojlido J, Raniszewski J, Woyciechowska J (1994) Water quality indexapplication for river in Vistula River basin in Poland. Water SciTechnol 30(10):57–64

Dube M, Johnson B, Dunn G, Culp J, Cash K, Munkittrick K, WongI, Hedley K, Booty W, Lam D, Resler O, Storey A (2006) De-velopment of a new approach to cumulative effects assessment: anorthern river ecosystem example. Environ Monit Assess 113:87–115

Dunnette DA (1979) A geographically variable water quality indexused in Oregon. J Water Pollut Control Fed 51(1):53–61

Environment Canada (2005). Application and testing of the CCME wa-ter quality index for selected water bodies in Atlantic Canada. Canbe viewed at www.ec.gc.ca/soer-ee/English/resource_network/status_report_e.cfm

EPI (2010) Yale Center for Environmental Law & Policy, YaleUniversity, Center for International Earth Science Informa-tion Network (CIESIN) Columbia University, in collaborationwith World Economic Forum, Geneva, Switzerland and JointResearch Centre of the European Commission, Ispra, Italy.Can be viewed at http://sedac.ciesin.columbia.edu/es/epi/papers/2008EPI_mainreport_july10.pdf

Fernandez N, Ramirez A, Sonalo F (2004) Physico-chemical waterquality indices—a comparative review. Bistua, Rev Fac Cienc Ba-sic 2(1):19–30. ISSN 0120-4211

Fulazzaky MA (2009) Water quality evaluation system to assessthe Brantas River water. Water Resour Manag 23:3019–3033.doi:10.1007/s11269-009-9421-6

Fulazzaky MA, Seong TW, Masirin MIM (2010) Assessment of wa-ter quality status for the Selangor River in Malaysia. Water AirPollution 205:63–77. doi:10.1007/s11270-009-0056-2

Giuseppe B, Guidice RL (2010) Application of two water quality in-dices as monitoring and management tools of rivers. Case study:the Imera Meridiopnale River Italy. Environ Manag 45:856–867.doi:10.1007/sD00267-010-9450-1

Hallock D (2002) A water quality index for ecology’s stream monitor-ing program. Washington State Department of Ecology, Olympia.Publication No. 002-03-052

Harkins RD (1974) An objective water quality index. J Water PollutControl Fed 46(3):588–591

Hèbert S (1996). Dévelopment d’un indice de la qualité bactéri-ologique et physico-chimique de l’eau pour des rivières duQuébec. Report of the ministière de l’environnement et de lafaune, Québec, QC, Canada

Hèbert S (2005) Comparison between the index of overall quality ofwater in Quebec (IQBP) the water quality index CCME (WQI)for the protection of aquatic life, Quebec. Ministry of SustainableDevelopment, Environment and Parks Department in Monitoringthe State of Environment, Québec. ISBN:2-550-45900-8

A Review of Genesis and Evolution of Water Quality Index (WQI) and Some Future Directions 23

Horton RK (1965) An index number system for rating water quality. JWater Pollut Control Fed 37(3):300–306

House MA (1989) A water quality index for river management. J WaterEnviron Manag 3:336–344

House MA (1990) Water quality indices as indicators of ecosystemchange. Environ Monit Assess 5:255–263

Husain T (2001) Canadian water quality index determination for threeEMAN sites. In: Ecological monitoring and assessment network.Environment Canada, Burlington

Inhaber H (1974) An approach to a water quality index for Canada.Water Res 9:821–833

Iowa DNR Report (2006) Iowa’s ambient water monitoring program,water fact sheet 2006-8. Iowa Department of Natural Resources,Geological Survey, Iowa City

Ip WC, Hu WQ, Wong W, Xia J (2009) Application of grey relationalmethod to river environment quality evaluation in China. J Hydrol(Amst) 379:284–290

Jinturkar AM, Deshmukh SS, Agarkar SV, Chavan GR (2010) Deter-mination of water quality index by fuzzy logic approach: a case ofground water in an Indian town. Water Sci Technol 61(8):1987–1994

Joseph J, Parameswaran G (2005) Water quality index of different areasin Calicut City. Asian J Chem 18:740–742

Karr JR (1981) Assessment of biotic integrity using fish communities.Fisheries 6(6):21–27

Khan F, Husain T, Lumb A (2003) Water quality evaluation and trendanalysis at selected watersheds of the Atlantic Region of Canada.Environ Monit Assess 88:221–242

Khan A, Tobin A, Paterson R, Khan H, Warren R (2005) Application ofCCME procedures for deriving site specific water quality guide-lines for the CCME water quality index. Water Qual Res J Can40:448–456

Kumar A, Dua A (2009) Water quality index for assessment of wa-ter quality of river Ravi at Madhopur (India). Glob J Environ Sci8(1):49–57

Kung H, Ying L, Liu Y (1992) A complementary tool to water qualityindices: fuzzy clustering analysis. Water Resour Bull 28(3):525–533

Landwehr JM (1974) Water quality indices: construction and analysis.PhD thesis, University of Michigan, Ann Arbor, Michigan, USA

Landwehr JM, Deininger RA (1974) An objective of water qual-ity index. Environ Monit Assess, J Water Pollut Control Fed46(7):1804–1807

Landwehr JM, Deininger RA (1976) A comparison of several waterquality indices. J Water Pollut Control Fed 48(5):954–958

Landwehr JM (1979) A statistical view of class of water quality in-dices. Water Resour Res 15(2):460–468

Lermontov A, Yokyama L, Lermontov M, Machado RDI (2009) Riverquality analysis using fuzzy water quality index: Ribeira doIguape river watershed, Brazil. Ecol Indic 9:1188–1197

Liebman H (1969) Atlas of water quality, methods and practical con-ditions. Oldenbourg, Munich

Linstone HA, Murray T (1975) The Delphi method: techniques andapplications. Addison Wesley, Reading

Liou SM, Lo SL, Wang SH (2004) A generalized water quality indexfor Taiwan. Environ Monit Assess 96:35–52

Lumb A, Halliwell D, Sharma T (2006) Application of the CCME wa-ter quality index to monitor. Water quality: a case study of theMackenzie river basin, Canada. Environ Monit Assess 113:411–429

McClelland NI, Brown RM, Deininger RA (1976) WQI—enhancingappreciation of quality improvement. National Sanitation Foun-dation, Ann Arbor

McDuffie B, Haney JT (1973) A proposed river pollution index. Amer-ican Chemical Society, Division of Water, Air and Waste Chem-istry, New York

Mercier V, Leger D (2003) Water quality survey. NB Department ofEnvironment. Can be viewed at http://www.gnb.ca/0009/0371/0013/index-e.asp

Mercier V, Leger D (2006) An assessment of the application of test-ing of the WQI of the CCME in the Atlantic Provinces. A dis-cussion paper, Knowledge Integration Strategies Division, Envi-ronment Canada, Gatineau, Quebec K1A 0H3. Can be viewed atwww.ec.gc.ca/soer-rec/English/recourse_network/CCME_e.cfm

Mrazik S (2007) Oregon water quality index summary report, wateryears 1997–2006. State of Oregon Department of EnvironmentalQuality, Laboratory Division, Portland, DEQ07-Lab-0047-TR

MRC (2008) An assessment of water quality in the lower Mekongbasin. MRC Technical Paper No. 19, Mekong River Commission,Vientiane, 70 p

Nagles JW, Davies-Colley RJ, Smith DG (2001) A water quality in-dex for contact recreation in New Zealand. Water Sci Technol43(5):285–292

Nikoo MR, Kerachian R, Malakpor-Estalaki S, Bassi-Azghadi SN,Azimi-Ghadikolaee MM (2010) A probabilistic water quality in-dex for river quality assessment: a case study. Environ Monit As-sess. doi:10.1007/s10661-010-1842-4

NL (2008) Agrifood water quality index (AFWQI). Water ResourcesManagement Division, Department of Environment and Conser-vation, Newfoundland and Labrador, St. John’s. Can be viewed athttp://www.env.gov.nl.ca/Env/env/waterres/Groundwater/Real_Time/afwqi.pdf

Ott WR (1978) Environmental quality indices: theory and practice.Ann Arbor Science, Ann Arbor

Parparov A, Hambright KD (2007) Composite water quality: evalua-tion and management feedbacks. Water Qual Res J Can 42(1):20–25

Pesce SF, Wunderlin DA (2000) Use of water quality indices to verifythe impact of Cordoba City (Argentina) on Suquia River. WaterRes 34:2915–2926

Prati L, Pavanell R, Pesarin F (1971) Assessment of surface water qual-ity by a single index of pollution. Water Res 5:741–751

Price DHA (1974) Water quality monitoring systems: the practice andexperience in the United Kingdom. In: Deininger RA (ed) Designof environmental information systems. An Arbor Science, AnnArbor

Qian Y, Migliaccio W, Wan Y, Li Y (2007) Surface water quality evalu-ation using multivariate methods and a new water quality index inthe Indian River Lagoon, Florida. Water Resour Res 43:W08405

Ramachandramoorthy T, Sivasankar V, Subramaniam V (2010) Theseasonal status of chemical parameters in shallow coastal aquifersof Rameswaran Island, India. Environ Monit Assess 160:127–139

Reynoldson TB, Rosenberg DR, Day KE, Norris RH, Resh VH (1997)The reference condition: comparison of multimetric & multivari-ate approaches to assess water-quality impairment using benthicmacroinvertebrates. J N Am Benthic Soc 16:833–852

Reynoldson TB, Logan C, Pascoe T, Thompson SP (2003) Cana-dian aquatic biomonitoring network (CABIN). Invertebrate Bio-monitoring, Field & Laboratory Manual National Water ResearchInstitute, Environment Canada, Burlington. Can be viewed athttp://cabin.cciw.ca/Application/Downloads/cabin_protocol.doc

Rickwood CJ, Carr JM (2009) Development and sensitivity analysisof a global drinking water quality index. Environ Monit Assess156:73–90

Rocchini R, Swain LG (1995) The British Columbia water qualityindex. Water Quality Branch, Environmental Protection Depart-ment, British Columbia Ministry of Environment Land and Parks,Victoria, p 13

Ross SL (1977) An index system for classifying river quality. WaterPollut Control 76(1):113–122

24 A. Lumb et al.

Saeedi M, Abessi O, Sharifi F, Meraji H (2010) Development ofground water quality index. Environ Monit Assess 163:327–335.doi:10.1007/s10661-009-0837-5

SAFE (1995) Strategic assessment of Florida’s environment indica-tors. Florida stream water quality index (WQI). Can be viewedat http://www.pepps.fsu.edu/safe/pdf/swq3.pdf

Said A, Stevens DK, Sehlke G (2004) An innovative index for evaluat-ing water quality in streams. Environ Manag 34(3):406–414

Sargaonkar A, Deshpande V (2003) Development of an overall indexof pollution for surface water based on a general classificationscheme in Indian context. Environ Monit Assess 89:43–67

Sarkar C, Abbasi SA (2006) Qualidex—a new software for generatingwater quality index. Environ Monit Assess 119:201–231

Scottish Research Development Department (1976) Development ofa water quality index. Applied Research & Development ReportNumber ARD3, Engineering Division, Edinburg, UK, 61 p

Sedeno-Diaz JE, Lopez-Lopez E (2007) Water quality in the RioLerma. Mexico: an overview of the last quarter of the twentiethcentury. Water Resour Manag 21:1797–1812

Sharma TC (2002) Canadian water quality index determination for 4sites in the Mackenzie River Basin. Contract KW405-02-0308,Ecological Monitoring and Assessment Network, EnvironmentCanada, Burlington, ON, Canada, p 58, September

Shuhaimi-Othman M, Lim EC, Mushrifah I (2007) Water qualitychanges in Chini Lake, Pahang West Malaysia. Environ MonitAssess 131:279–292

Simões FS, Moreira AB, Bisinoti MC, Nobre Gimenez SM, Yabe MJS(2008) Water quality index as a simple indicator of aquacultureeffects on aquatic bodies. Ecol Indic 8(5):476–484

Sladecek V (1973) System of water quality from biological point ofview. In: Mai V (ed) Archiv fur Hydrobiologie. Advances inLimnology, vol 7. E. Schweizerbart‘sch Verlagsbuch-handling,Stuttgart, p 218. ISBN:978-3-510-47005-1 paperback

Smith DG (1990) A better water quality indexing system for rivers andstreams. Water Res 24(10):1237–1244

Snow J (1854) On the mode of communication of cholera. Churchill,London. Can be viewed at http://www.ph.ucla.edu/epi/snow.html

SOE Report (1997) The state of the environment report. Pro-cess and criteria for indicator selection. Can be viewed athttp://www.manitoba.ca/conservation/annual-reports/soe-reports/soe97/append.html

Soltan ME (1999) Evaluation of groundwater quality in Dakhla Oasis(Egyptian Western Desert). Environ Monit Assess 57:157–168

Stambuk-Giljanovic N (1999) Water quality evaluation by index inDalmatia. Water Res 33(16):3423–3440

Stambuk-Giljanovic N (2003) Comparison of Dalmatian water valuesindices. Water Environ Res 75(5):338–405

Statistics Canada (2008) Behaviour study on the water quality index ofthe Canadian Council of Ministers of the Environment (CCME).It can be viewed at www.statcan.ca/english/research/16-001-MIE2007003.htm

Steinhart CE, Schierow LJ, Chesters G (1981) A review of waterquality and related Indices. Great Lakes Environmental PlanningStudy contribution No 38, Water Resources Centre, University ofWisconsin, Madison, Wisconsin, 53706, USA

Steinhart CE, Schierow LJ, Sonzogni WC (1982) An environmentalquality index for the Great Lakes. Water Resour Bull 18(6):1025–1031

Stingter TY, Ribeiro L, Carvalho Dill AMM (2006) Application ofa groundwater quality index as an assessment and communica-tion tool in agro-environmental policies—two Portuguese case

studies. J Hydrol (Amst), 327:578–591. doi:10.1016/j.hydrology.2005.12.01

Stojda A, Dojlido J (1983) A study of water quality index in Poland.WHO. Water Anal Bull 1:30–32

Swamee PK, Tyagi A (2000) Describing water quality with aggregateindex. J Environ Eng Div ASCE 126(5):450–455

Thomson B, Melwani A, Lowe S, Greenfield B (2007) Developmentof indicators of anthropogenic contamination in the San FransiscoEstuary. A report to the San Fransisco Estuary Project, San Fran-sisco Estuary Institute, Oakland, CA, December 17

Tipping WD, Cude C, Long SC, Boyce JK (2002) Sustainable develop-ment indicators for transboundary river basin systems. In: Waterintelligence on line, January 2002. Papers from the IWA WorldWater Congress, Berlin, 2001

Tiwari TN, Mishra M (1985) A preliminary assessment of water qual-ity index of major Indian rivers. Indian J Environ Prot 5(4):276–279

Tobin A, Khan A, Khan H, Moores L, Taylor J (2007) Forestry waterquality index—a planning tool for the assessment and commu-nication of the impact of forestry activities on water quality. ForChron 83:207–214

Tsegaye T, Sheppard D, Islam KR, Johnson A, Tadess W, Atalay A,Marzen L (2006) Development of chemical index as a measureof in stream water quality in response to land-use and land-coverchanges. Water Air Soil Pollut 174(1):61–79

Tyson JM, House MA (1989) The application of a water quality indexto river management. Water Sci Technol 21:1149–1159

UKTAG (2007) Recommendations on surface water classificationschemes for the purposes of the water framework directive. UKTechnical Advisory Group on the Water Framework Directive,December

UNEP GEMS (2007) Global drinking water quality index. De-velopment and sensitivity analysis report. Can be viewed athttp://www.gemswater.org/publications/pdfs/gwqi.pdf

Vignolo A, Pochettino A, Cicerone D (2006) Water quality assessmentusing remote sensing techniques: Medrano Creek, Argentina.J Environ Manag 81(4):429–433

Walski TM, Parker FL (1974) Consumer’s water quality index. J Envi-ron Eng Div ASCE 100(3):593–611

Wepener V, Euler N, van Vuren JHJ, du Preez HH, Kohler A (1992)The development of an aquatic toxicity index as a tool in the oper-ational management of water quality in the Olifants River (KrugerNational Park). Koedoe 35(2):1–9

Wepener V, van Vuren JHJ, du Preez HH (1999) The implementationof an aquatic toxicity index as a water quality monitoring tool inthe Olifants River (Kruger National Park). Koedoe 42(1):85–96

Wright CR, Saffran KA, Anderson AM, Neilson RD, MacAlpineND, Cooke SE (1999) A water quality index for agriculturalstreams in Alberta. In: The Alberta agricultural water quality in-dex (AAWQI). Alberta Environmentally Sustainable AgricultureProgram Alberta Agriculture, Food and Rural Development, Ed-monton, p 35

Zandbergen PA, Hall KJ (1998) Analysis of the British Columbia waterquality index for watershed managers: a case study of two smallwatersheds. Water Qual Res J Can 33(4):519–549

Zoabi B, Gueddari M (2008) Long term water quality monitoring ofthe Sejane reservoir in northeast Tunisia. Bull Eng Geol Environ.doi:10.1007/s10064-009-0186-1