 |
Ambient Water Quality Criteria for Colour
1. Definitions, Concepts and Analytical Measurements
1.1 Key Concepts
The observed colour of water is the result of light back scattered upward from the water after it has passed through to various depths and undergone selective absorption. Colour and turbidity determine the depth to which light penetrates in water systems. In water, the light intensity or irradiance at a particular depth (Iz) is a function of the intensity at the surface (Io) to the exponent of the negative extinction coefficient at the depth distance z which is called the Beer-Lambert's Law (Freifelder 1985).
The extinction coefficient is a constant that predicts the attenuation or dissipation of light at a certain wavelength. In pure water, light is highly absorbed in the infrared region of the light spectrum and poorly absorbed in the blue region. Extinction coefficients are influenced by water absorption, suspended organic and inorganic particles, and dissolved compounds (Jerome et al. 1994a; Jerome et al. 1994b). Thus, the visible colour in a water sample is the light that is refracted, reflected or re-emitted by substances in water because it has not been absorbed to produce heat or chemical reactions.
The colour of water and other materials has three main attributes: hue, brightness and saturation (Davies-Colley et al. 1988). Hue refers to whether the water colour is described as blue, green or yellow, for example, and is determined by the dominant wavelength in the visible spectrum. Brightness depends on the amount of energy detected by the human eye, which is most sensitive to green light of wavelength 555 nm. The saturation depends on the spread of energy around the dominant wavelength. Saturation is also referred to as spectral purity (Jerome et al. 1994b).
True colour is due to natural minerals such as ferric hydroxide and dissolved organic substances such as humic or fulvic acids (Hongve and Akesson1996). A great variety of dissolved organic substances originating from anthropogenic sources such as dyes can also contribute to water colouration (McCrum 1984; Brown 1987; Borgerding and Hites 1994). True colour can only be measured once a sample has been centrifuged or filtered (APHA 1992; Environment Canada 1989; Bennett and Drikas 1993).
Colour measured in water containing suspended matter is defined as apparent colour (APHA 1992; Bennett and Drikas 1993). Suspended matter can be in the form of large organic particulates such as plant debris, phyto- and zooplankton (Effler and Auer 1987). For example, a blue-green colour can be due to blue-green algal blooms, a yellow-brown colour to diatoms or dinoflagellates, and reds and purples to Daphnia or copepods (Chapman 1992). Because some of these organisms thrive on anthropogenic releases or disturbances (e.g., fertilizers and forest activities), polluted waters may have a strong apparent colour.
1.2 True Colour
The amount of dissolved organic carbon in streams and lakes is typically about ten times the amount of particulate organic carbon. Dissolved organic compounds or "dissolved colour" greatly affects the absorption of light compared to suspended particles. Organic compounds such as humic acids absorb light and reduce its transmission relative to distilled water (James and Birge 1938). As well, the adsorption will be shifted selectively (Figure 1). Organic compounds contributing to the "dissolved colour" are highly selective and marked by UV, blue and green wavelengths and less so in the red and infrared regions of the light spectrum (see curve C, Figure 1). Light absorption by humic acids has been used extensively in the determination of their concentration in water systems (Reckhow and Singer 1990; Mierle and Ingram 1991).
Figure 1. Percentile absorption of light vs. wavelength. T is the apparent colour, C is the true colour, P is that due to suspended particulates and W is pure water. Ultraviolet and infrared wavelengths are <400 nm and >800 nm, respectively.
1.3 Apparent Colour
The extinction coefficient for particulate matter functions independently from that of dissolved organics. Particles in natural waters generally cause a relatively constant attenuation of light through the visible part of the spectrum in spectrophotometric measurements (see curve P, Figure 1), (James and Birge 1938). Jerome et al. (1994b) refer to suspended particulates as "white scattering centers" because natural waters with high concentrations of particulates have low spectral purity.
Suspended particulates are a natural component of water systems and can be composed of organic detrital carbon and inorganic suspended sediments, both of which contribute to the absorbance readings. The characteristics and rates of sediment and detrital transport depend on the composition of soils and parent materials, local climatic conditions, topography, hydrology, and vegetation of the watershed (Mitchell and McDonald 1995; Häkanson 1993; Heikkinen 1994; Midgley and Schafer 1992). In addition, land use activities tend to accelerate sediment production, thereby increasing the potential for problematic levels of sediment in the water column (Shields and Sanders 1986; Forsberg 1992). For example, logging can increase the concentration of suspended particulates, especially fine debris coming from terrestrial origin (Bilby and Bisson 1992). Included in this debris is a greater input of carbohydrates and proteins that biodegrade quickly with a consequent increase in primary productivity. Additional production of seston can have dramatic impacts on fish populations.
The type and concentration of suspended matter controls the turbidity and transparency of the water and ultimately the depth of the euphotic zone (i.e., the zone with sufficient light to support photosynthesis, generally one percent of ambient light). Measures of apparent colour by, for example, a measure of transmittance can be used to warn against high/low or increasing/decreasing levels of particulate matter.
1.4 Analytical Methods
True Colour
True colour can be measured by comparator and colourimetric methods. Comparator methods rely on visual comparison of a water sample with a standard colour solution or a set of coloured filter disks. The most common comparator method involves matching a water sample with one of a series of dilutions of a standard colour solution of platinum and cobalt chloride salts of molar ratio 2:1 where the platinum concentration in mg/L is equivalent to the colour value in Hazen units (Bennett and Drikas 1993). The Fore-Ule colour scale involves comparisons to alkaline solutions of cupric sulfate, potassium chromate and cobaltous sulfate. The Hazen scale of true colour measurement, however, has been adopted as the reference method by organizations that set standards for water quality analysis, and by many governments in deriving their drinking water quality guidelines (APHA 1992; NH & MRC and AWRC 1987; WHO 1983; EN-ISO 1994).
Colourimetric methods are based on the calibration of absorbance of the water sample at a variety of single wavelengths, usually against the Pt-Co standard (Bennett and Drikas 1993; Hongve and Akesson 1996). Standard measurement comparisons can be made with sealed containers (e.g., the Hellige Aqua Tester). Natural waters range from <5 in very clear waters to 1200 mg/L Pt in dark peaty waters (Kullberg 1992). As some of the compounds determining the colour of water are not very stable, measurements should be made within two hours of collection (Environment Canada 1989).
Comparator or visual assessment methods are not very precise. Most operators find it difficult to distinguish between colours that differ by <5 mg/L Pt (Hongve and Akesson 1996). Inter-laboratory comparisons of colour generally produce a standard deviation of approximately 5 to 10 mg/L Pt (Bennett and Drikas 1993; Hongve and Akesson 1996). Given the drinking water standard in Canada of 15 mg/L Pt (Health Canada 1996), the coefficient of variation for maximum permitted colour is between 33 and 67%. Such uncertainty has necessitated the development of a non-visual, more precise means of quantifying true colour as calibrated against the Hazen scale. A major obstacle, however, has been to match apparatus readings with visual judgments because photometers are confined to measures at defined spectral lines or bands (Hongve and Akesson 1996). The turbidity of natural waters also interferes with the measurement of true absorbance because spectrophotometers are not designed to measure the scattering of light (Bennett and Drikas 1993). Therefore, several steps are required before there can be good agreement between comparator and colourimetric methods. First, turbidity must be removed by either filtration or centrifugation (APHA 1992), or its contribution quantified (Bennett and Drikas 1993; Hongve and Akesson 1996). The common practice of filtration through 0.45 micron filters should be satisfactory for most waters, although repeated filtration may be required for very turbid waters. Bennett and Drikas (1993) found that turbidity is a linear function of absorbance at selected wavelengths between 350 and 700 nm (also see Figure 1, line P). Using this information, they were able to derive the following equation that corrects for turbidity in photometric determinations of true colour (C) in mg/L Pt:
where c is the colour absorptivity (mg/L Pt/cm, E is the coefficient of scattering (NTU/cm, t is turbidity (NTU), A (in proportion) is total absorbance due to dissolved coloured species, and l corresponds to cell path length (cm). Note, however, that the correction for turbidity assumes that particle size distribution in water samples, an inherently variable characteristic in natural waters, is similar to that of the kaolin suspension used in the calibration experiment. Although correction for turbidity would be useful in approximating true colour, it cannot be reliably applied to all waters.
Once turbidity has been removed or quantified, the next step is to select the appropriate wavelength for measuring true colour. In order to produce general agreement with comparator methods, the appropriate wavelength should exhibit equal absorbance when comparing natural coloured waters and the Pt-Co reference solution. For natural waters with high concentrations of humic and fulvic acids, this occurs around 410 nm, and 445-470 nm. The shorter wavelength has better sensitivity and one inter-laboratory comparison between Nordic countries and two inter-laboratory comparisons in Norway have shown that photometer readings at 410 nm and comparator readings gave identical results, with the former method having much better precision (Hongve and Akesson 1996). Bennett and Drikas (1993); however, recommend single wavelength analysis at 456 nm because the influence of turbidity (after filtration) is negligible at this wavelength (see Figure 1). Note that both comparator and colourimetric methods based on the Hazen scale are not appropriate for industrial or other wastewaters that diverge in colour from the Pt-Co standard. Coloured compounds in wastewaters and humic and fulvic acids have different absorption spectra and thus will not exhibit equal absorbance at the same wavelengths as occurs with humic and fulvic acids and Pt-Co reference solutions at 410 nm and 445-470 nm. Several of the methods discussed below are more appropriate for wastewaters.
Crowther and Evans (1981) suggested that analysis across a broad analytical wavelength range (405 to 460 nm) produces colour readings in good agreement with comparator methods. Similar good agreement has been achieved in the analytical wavelength range 445 to 470 nm by Bennett and Drikas (1993). The latter authors argue; however, that the improvement of broad wavelength analysis over single wavelength analysis is marginal with the former method requiring much additional effort.
Colour is dependent on factors that affect the solubility and stability of the dissolved and particulate fractions of the sample such as pH, temperature, exposure to light, and storage time. Although most methods recommend measurement and recording of pH, pH standardization is not desirable because the resultant colour will differ from the colour of water in situ (Bennett and Drikas 1993). Also, to ensure that sample and in situ water colour are the same, most methods recommend that colour samples be analyzed within two hours (Environment Canada 1989).
A specialized method, derivative spectroscopy, can be used to help identify sources of pollution in water samples contaminated by coloured substances (McCrum 1984). Derivative spectroscopy in the UV-visible region of the spectrum records variations in the rate of change of absorbance (A) with wavelength. Thus, in the first order derivative spectrum dA/dwavelength is plotted against wavelength, in the second order derivative spectrum d2A/dwavelength2 is plotted against wavelength and so on for higher order derivative spectra. The most useful characteristic of derivative spectroscopy is the increase in resolution and detail produced. In water analysis, derivative spectroscopy has been applied to the quantitative determination of phenol in wastewater, uric acid in municipal wastewater, nitrate and nitrite in water, and to investigate suspected sources of colour in natural waters (McCrum 1984).
Apparent Colour
Apparent colour is due to dissolved organic matter and suspended particulates in the water such as plant debris, phyto- and zooplankton (Effler and Auer 1987). Despite suspended particulates having relatively non-selective scattering properties, high concentrations of particulate matter of inorganic clays or volcanic ash can produce a yellow to red colour, while high concentrations of blue-green algae and diatoms produce blue-green and yellowish-brown colours, respectively (Wetzel 1975).
Since the absorption spectrum of particulate matter is relatively non-specific, some measures of apparent colour simply estimate the reduction in light transmission or visual clarity resulting from scattering of light by particulates. The most common method for estimating visual clarity is the Secchi disk method. The Secchi disk transparency is the mean depth of the point where a weighted white disk, 20 cm in diameter, disappears when viewed from the shaded side of a vessel during mid-day, and the point where it reappears upon raising it after being lowered beyond visibility (Wetzel 1975). The Secchi disk transparency is a function of the reflection of light from its surface and is therefore influenced by both the absorption characteristics of the water and the presence of dissolved and particulate matter. Most studies indicate, however, that particulate matter influences Secchi disk transparency to a greater extent than does true colour (Wetzel 1975). For example, multiple regression of data from 55 Florida lakes yielded the following close-fitting equation (r2 = 0.89), (Brezonik 1978):
where SD is Secchi disk transparency in meters, T is turbidity in nephelometric turbidity units (NTU) and C is true colour in mg/L Pt. When only one independent variable was included in the regression equations, the r2 value for turbidity was 0.71 and for colour it was 0.10, indicating that both variables contribute to Secchi disk transparency, although turbidity is considerably more important. In these lakes, turbidity was primarily autochthonous and thus closely related to number of algal particles.
An alternative method for measuring visual clarity relies on a black disc viewed horizontally in the water using a right-angle viewer (Davies-Colley and Smith 1992). The black disc method can be used in shallow waters where the Secchi disk cannot be used. The black disc sighting range is inversely proportional to the beam attenuation coefficient, and can be used to estimate this optical property of water (Davies-Colley and Smith 1992). The method is routinely used in water resources investigations and management programs in New Zealand, and may eventually replace the Secchi disk method (Davies-Colley and Smith 1992; Smith et al. 1991; Smith et al. 1995). Davies-Colley and Smith (1992) describe the black disc method, including several adaptations that may be used when in situ measurements are difficult, or for very turbid waters.
A robust means of determining in situ apparent colour is the Munsell system of colour (Davies-Colley et al. 1988). The Pt-Co scale refers only to the yellowness of filtered water samples and does not cope well with blue or green-hued waters. The Munsell system does not have this shortcoming. The system has three coordinates: Munsell hue (H), value (V), and chroma (C). Munsell hues are divided into 100 hue units and are designated as, for example, 10 BG indicating that the sample corresponds to the 10th unit of the blue-green hue range. The Munsell value is a measure of apparent brightness and ranges from 0 (black) to 10 (white). The Munsell chroma is related to colour saturation (or spectral purity) and extends from 0 for neutral grays to values of 20 or more for the most saturated colours. Munsell scores are designated by a notation indicating hue, value and chroma in H V/C format. For example, a bright, low saturation green might be 8.5 G 7/4. Munsell standards are available in a convenient book form that is readily transportable. Although the Munsell system of colour is a comparator method and suffers from some of its shortcomings, the technique shows good correspondence to the results of chromaticity analyses applied to spectro-radiometric scans of near-surface upwelling light in a diverse range of New Zealand lakes (Davies-Colley et al. 1988). Munsell hue is routinely measured in water quality surveys in New Zealand (Davies-Colley and Smith1992; Smith and Davies-Colley 1992; Smith et al. 1991) and elsewhere (Eloranta 1978).
The tristimulus colour system is based on a system of parameters referred to as tristimulus values that can be calculated from up- and down-welling spectral irradiance data (Thomson and Jerome 1975). These parameters plotted on a CIE chromaticity chart numerically define the colour of a particular irradiance measurement in terms of a dominant wavelength and its purity or monochromaticity (Thomson and Jerome 1975; Davies-Colley et al. 1988; Jerome et al. 1994a,b). The tristimulus colour system implies that the dominant wavelength is composed of a combination of three colours: red, blue and green. The tristimulus values of an upwelling irradiance spectrum E(wavelength) are given by:
where x(wavelength), y(wavelength) and z(wavelength) are the CIE colour mixtures for red, green and blue, respectively for equal energy spectra. These may be obtained from CIE tables (Jerlov 1976). The chromaticity coordinates X, Y and Z for red, green and blue, respectively are then obtained from the equations:
Since X+Y+Z = 1, two chromaticity coordinates adequately represent a chromaticity diagram. The loci of all possible (Z,Y) pairs define an envelope which encompasses all possible chromaticity values. For a white spectrum, X = Y = Z = 0.333. This defines the achromatic point S. The dominant wavelength of a particular measured spectrum C is determined by the intersection of the line S-C with the chromaticity envelope indicated by point L. The dominant wavelength is the colourimetric definition of colour in the water body. The spectral purity of the dominant wavelength is defined as the ratio of the line C-S to the line L-S. Thus, spectral purity is a measure of the contribution of the dominant wavelength to the observed optical spectrum (Jerome et al. 1994a,b). A value of 1.0 indicates a monochromatic spectrum at the dominant wavelength, while a value of 0 indicates a white spectrum. The tristimulus colour system is a standard method (APHA 1992) and, although more complicated than the Hazen system, can be applied to a broader range of coloured waters such as industrial wastewaters containing high concentrations of dyes. The tristimulus system can also distinguish between waters with varying amounts of particulate matter (Thomson and Jerome 1975; Jerome et al. 1994a,b). For example, in studies of near-surface upwelling irradiance in Lakes Ontario and Superior, clear waters had a dominant wavelength of 490-530 nm, biologically productive waters had a dominant wavelength of 550-560 nm, and waters with heavy sediment loadings had a dominant wavelength of >565 nm (Thomson and Jerome 1975). This relationship has been noted in other studies (Davies-Colley et al. 1988; Bukata et al. 1983; Jerome et al. 1994a,b; McPherson and Miller 1987).
