You are on page 1of 32

The Science of the Total Environment 314 316 (2003) 303334

A review of dissolved oxygen modelling techniques for lowland rivers


B.A. Cox*
Department of Geography, The University of Reading, Whiteknights, Reading RG6 6AB, UK Accepted 1 January 2003

Abstract This review introduces the methods used to simulate the processes affecting dissolved oxygen (DO) in lowland rivers. The important processes are described and this provides a modelling framework to describe those processes in the context of a mass-balance model. The process equations that are introduced all require (reaction) rate parameters and a variety of common procedures for identifying those parameters are reviewed. This is important because there is a wide range of estimation techniques for many of the parameters. These different techniques elicit different estimates of the parameter value and so there is the potential for a significant uncertainty in the models inputs and therefore in the output too. Finally, the data requirements for modelling DO in lowland rivers are summarised on the basis of modelling the processes described in this review using a mass-balance model. This is reviewed with regard to what data are available and from where they might be obtained. 2003 Elsevier Science B.V. All rights reserved.
Keywords: Dissolved oxygen; Biochemical oxygen demand; Photosynthesis; Respiration; Mass-balance model; Rate parameter

1. Introduction A sufficient supply of dissolved oxygen (DO) is vital for all higher aquatic life. The problems associated with low concentrations of DO in rivers have been recognised for over a century. The impacts of low DO concentrations or, at the extreme, anaerobic conditions in a normally welloxygenated river system, are an unbalanced ecosystem with fish mortality, odours and other aesthetic nuisances. When DO concentrations are reduced, aquatic animals are forced to alter their breathing patterns or lower their level of activity.
*Tel.: q44-118-9316553; fax: q44-118-9755865. E-mail address: b.a.cox@reading.ac.uk (B.A. Cox).

Both of these actions will retard their development, and can cause reproductive problems (such as increased egg mortality and defects) andyor deformities. The variability in DO in rivers is caused by the influences of many factors, and the major influences can be categorised as being either sources or sinks of DO in rivers. The major sources of DO include: Reaeration from the atmosphere; Enhanced aeration at weirs and other structures; Photosynthetic oxygen production; The introduction of DO from other sources such as tributaries.

0048-9697/03/$ - see front matter 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0048-9697(03)00062-7

304

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

The main causes of oxygen depletion, or sinks, are: The oxidation of organic material and other reduced matter in the water column; Degassing of oxygen in supersaturated water; Respiration by aquatic plants; The oxygen demand exerted by river bed sediments. If the volume of water in a water body is V and the concentration of DO is C, then the variation of DO in that water body may be described as: Accumulation s Mass in y Mass out q sources y sinks, or in simple mathematical terms as: dM sMiyMoq(PyR)qCRyBODySOD dt yCD"DS

profiles along natural streams. BOD is a mixture of many different elements and so if one uses BOD the model strictly cannot close the massbalance. However, this deficiency is generally assumed acceptable given the alternative of an over-parameterised, overly complex model that may be impossible to implement. 2. Influences on DO concentrations in lowland rivers In the UK, stream DO studies have been undertaken since the late nineteenth century Theriault (1927). By the turn of the century, the science of DO measurement and interpretation had progressed rapidly following research in the UK by the Royal Commission on Sewage Disposal while in the United States, studies on the Ohio River were conducted between 1914 and 1916. These investigations provided the basis of the classic mathematical modelling of DO by Streeter and Phelps (1925). This model incorporates the two primary mechanisms governing the fate of DO in rivers receiving sewage, the decomposition of organic matter and atmospheric aerationyreaeration. According to their theory, biochemical oxidation is the only sink and atmospheric reaeration is the only source of oxygen (i.e. photosynthesis and respiration are ignored). The simplest manifestation of the StreeterPhelps model is for a reach in steady-state (i.e. time invariant) characterised by plug flow with constant hydrology and geometry. Mass-balances may be written as: U dL dD syKrLU sKdLyKaD dx dx (2)

(1)

where t is the time, Mi is the mass flux of DO entering the water body, Mo is the mass flux leaving, CR represents the aeration and reaeration processes, BOD is the biochemical oxygen demand representing the oxidation of organic material, SOD is the sediment oxygen demand, CD represents degassing of oxygen and DS represents changes in the water body due to transport from external sources. In this review, the major processes are examined in more detail and are expressed in mathematical terminology in the form of differential equations. The list of processes described above is limited and one could attempt to construct a mathematical model based on descriptions of the movement of each chemical element in the system. Such a model would explicitly simulate all of the chemical reactions and biological processes affecting each elementbe they dissolved, adsorbed on particles, or a part of some plant or animal. However, such a model would be prohibitively complex and few of the required data would be available as they would not be measured, or perhaps not even be measurable. As a result, most mechanistic or processbased models are actually semi-empirical although they do represent individual processes. For example, many models of DO usually employ modified or extended versions of the classical equations of Streeter and Phelps (1925) for the BOD and DO

where U is average stream velocity, L is the amount of oxidisable organic material as oxygen equivalents (i.e. the BOD), x is the distance along the reach moving downstream and D is the DO deficit (the difference between the DO concentration if saturated and the actual concentration). The other terms are the rates of processes affecting the DO: Ka is the surface reaeration rate; Kr is the total removal rate of organic matter; Kd is the decomposition (i.e. oxidation) rate in the stream; and KrsKdqKs, where Ks is the net rate of sedimen-

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

305

Fig. 1. How an open system affected by a BOD load is described by the StreeterPhelps model. DO is the DO concentration, Cs is the saturation concentration of DO and L is the BOD concentration.

tation and re-suspension of BOD. If LsL0 and DsD0 at time zero then these equations can be solved for: LsL0e(yKryU)x DsD0e(yKayU)xq KdL0 (yKryU)x (yKayU)x ye e . (3) KayKr

These two equations constitute the classic StreeterPhelps model of BOD and DO profiles along a river. A simple application of this model describes the classic DO sag curve that occurs below sewage discharges in streams as illustrated in Fig. 1. This application represents a stream that is originally unpolluted and so has DO concentrations near saturation. A large BOD load is then added, for example, from untreated sewage, and this elevates the levels of dissolved and solid organic matter. As oxygen levels drop, atmospheric reaeration takes place to compensate for the oxygen deficit. Initially, the reaeration is dwarfed by the oxidation of the BOD as the organic matter is consumed. However, with time, the amount of organic matter not assimilated decreases and so the rate of oxygen loss decreases. At some point, the oxygen depletion and reaeration will balance and at this point, the lowest or critical level of oxygen is reached. Beyond this point, reaeration dominates and so

oxygen levels begin to rise again towards the saturation concentration. Traditionally wastewater engineers have been content with this description and have concentrated on how best to estimate the rate parameters Ka, Kd and Ks so that the position and extent of the DO minimum for a given polluting load can be predicted. However, while the equations of Streeter and Phelps (1925) have been found to be adequate to give a schematised representation of the variation of DO concentration downstream of a point discharge, subsequent studies have pointed out that other sources and sinks need to be considered (Dobbins, 1964; Camp, 1963; Owens et al., 1964; Edwards and Owens, 1965). Thus, the processes suggested for inclusion in a model of DO are: the removal of BOD by sedimentation or adsorption; the addition of BOD by the re-suspension of bottom sediments or by the diffusion of partially decomposed organic matter from the bed sediments into the water above; the addition of BOD by local runoff; the removal of oxygen from the water by the action of gases in the sediments; the removal of oxygen by the respiration of plankton and fixed plants; the addition of oxygen by the photosynthesis of plankton and fixed plants;

306

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

the addition of oxygen by atmospheric reaeration; the redistribution of BOD and DO by the effect of dispersion, particularly when the polluting load varies suddenly; the removal of oxygen by nitrifying bacteria. Of these processes, the dominant ones which relate to the oxygen balance of a river are (Bennett and Rathburn, 1972) the oxygen demand of the carbonaceous and nitrogenous wastes in the water; the oxygen demand of the bottom deposits; any immediate chemical oxygen demand (COD); the oxygen required for plant respiration; the oxygen produced by plant photosynthesis; the oxygen gained from atmospheric reaeration. 3. Process descriptions The six dominant processes given by Bennett and Rathburn (1972), as listed above, must be formulated as mathematical equations if they are to be used to simulate DO in a river using a mathematical model. This procedure is familiar to all water quality models and many models share similar mathematical descriptions of the physical, chemical and biological processes in a river. 3.1. BOD oxidation, sedimentation and resuspension The BOD is the amount of oxygen required by micro-organisms (to respire) as they consume organic matter. The amount of organic waste in a water body is measured by its demand on the waters oxygen resources. In rivers that pass through populated areas, the majority of BOD in rivers will generally come from effluent discharges of organic matter (for example, from sewage treatment works (STW)), but it can also be sourced from dead algae and other organisms. Although generally small, the input of BOD to a river reach from dead organisms can be very significant after algal blooms or large fish kills causing further problems downstream of the event itself. BOD concentrations vary with time and location along

a river because as oxygen is used in oxidising the waste, the amount of waste material decreases and so the BOD drops. BOD can also be removed from the water by sedimentation or it can be added to the water by scouring and re-suspension of bottom deposits. All of these factors must be considered in order to model the BOD distribution along a river. The rate at which biochemical oxidation takes place is usually assumed a first-order process, i.e. the rate of oxidation is proportional to the amount of organic matter remaining in the water. The net rate of sedimentation and scour is also generally assumed proportional to the amount of BOD present. Therefore, an equation for the distribution of BOD can be written as (Thomann and Mueller, 1987): L L qux sKadwAlgxyKdLyKsL t x (4)

where L is the mean BOD concentration in the water column; x is the distance downstream, ux is wAlgx is the algal the mean longitudinal velocity; biomass contribution; Kad is the rate constant for the BOD addition due to algal death; Kd is the rate constant for the biochemical oxidation; and Ks is the rate constant for the net rate of settling and re-suspension of BOD. In terms of DO concentrations then, the rate of change of DO due to BOD oxidation alone could be expressed as a firstorder process in a well-mixed system as: dC syKdL dt (5)

where C is the DO concentration, and which is obtained by the solution of Eq. (5) above once the rate Kd has been estimated. Because the oxidation is carried out by organisms, the rate will also depend on nutrient availability, but this is rarely considered in water quality models. The BOD is usually measured using a 5-day test (BOD5), but this may not account for the whole of the oxygen demand that the organic matter in a sample can exert given sufficient time (i.e. the ultimate BOD or BODu). The BOD data

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334 Table 1 Typical values of BOD decomposition rates for various levels of treatment (Chapra, 1997) Treatment Untreated Primary Activated sludge Kd (dayy1) at 20 8C 0.35 (0.200.50) 0.20 (0.100.30) 0.075 (0.050.10) BOD5 yBODu 0.83 0.63 0.31

307

used in the DO models will invariably be the BOD5 (because of the availability of these data). However, it might be of merit to estimate BODu and use this when modelling DO if the time-scales being simulated are sufficiently long. If the firstorder decomposition model holds, an extrapolation can be made to estimate BODu by using the firstorder decay (or oxidation) rate Kd (Thomann and Mueller, 1987): BODus BOD5 1yey5Kd (6)

Using the assumption that the BOD5 and BODu can be conceptualised as fast and slow components of BOD, Chapra (1997) provides estimates of the relative proportions of fast and slow BOD for STW effluent (Table 1). This estimate can be further refined by using the concept of a fast and slow BOD such that each component has its own reaction rate, such that the ultimate BOD would be expressed by: BODus BOD5 1yw(1ya)ey5Kfbodqaey5Ksbodx (7)

represent the demand by resident invertebrates for respiration. There are two aspects to this sediment demand: oxygen diffusing into the sediments then being consumed, and reduced organic matter entering the water column where it is oxidised. The main influencing factors on the effect of the SOD are: the concentration of DO in the overlying water; temperature; the characteristics of the bed (physical, chemical and biological); the velocity of the water in the stream and the characteristics of the interstitial (or pore) water, nut nutrient availability may also be a factor. Commonly used methods for modelling SOD use a zero-order or constant source term, but this is not ideal because it treats SOD as a model input rather than as a simulated part of the in-stream processes (Chapra, 1997). Alternative models have been developed that relate the SOD to the sedimented BOD and divide the sediments into anaerobic and aerobic layers (Di Toro et al., 1990) so that the oxidation of methane and ammonium can be simulated. However, these methods are complex and can involve many parameters that may not be measured and so must be evaluated by calibrating the model to fit observed data. A simpler approach is to assume that the SOD uptake is related to the transfer of oxygen between the overlying water and the sediments, i.e.: V dC dC K K syKAsC sy B E Cfy C dt dt V H C F D As G (8)

where, a is the proportion of slow BOD, Kfbod is the reaction rate constant for the oxidation of fast BOD and Ksbod is the same reaction rate, but for slow BOD. 3.2. Sediment oxygen demand Sediment (or benthic) oxygen demands (SOD) result from organic matter being deposited and incorporated in the channel bed. The sources of SOD may be allochthonous (external) such as leaf litter and humic substances, or autochthonous (internal) such as the settling of BOD. It may also

where K is the transfer coefficient (m dayy1) and C is the DO concentration in the water (mg O2 ly1), V is the water volume and As is the bed area in contact. VyAs can be approximated by the river depth, but it actually represents the availability of oxygen to the bed. Variations of the reaction rate (K) with temperature (T) are commonly predicted using the following Eq. (9): KsKrefu(TyTref) (9)

where the most frequently used Tref is 20 8C. This form is based on the Arrhenius equation, which

308

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

provides the temperature dependence of a reaction based on its activation energy. Assuming that water temperatures will only vary over a limited range (say 040 8C), u is defined as a constant: lnusy E RTrefT (10)

3.3. Nitrification and the nitrogenous biochemical oxygen demand Nitrogenous matter in waste generally consists of organic compounds such as proteins, urea, ammonia and nitrate together with intermediate decomposition products of the proteins such as amino acids, amides and amines in varying amounts. With time, the proteins are broken down by hydrolysis in a series of steps into amino acids in a process called deamination. In this process, ammonia (NH3) is released and this can then combine with hydrogen ions in the water to form the ammonium ion (NHq). Thus, ammonia in 4 natural waters is invariably the result of either the direct discharge of the material in wastewaters or the decomposition of organic matter (Thomann and Mueller, 1987). The ammonium in water can be oxidised under aerobic conditions to nitrite (NOy) by bacteria of 2 the genus Nitrosomonas as described by the following equation (Gaudy and Gaudy, 1980):
Nitrosomonas q 2NH4 q3O2

where T is the temperature in 8C, R is the universal gas constant (8.314 J Ky1 moly1) and E is the activation energy of the reaction. The temperature can also be specified in degrees Kelvin if consistent use is made (it is the difference from the reference temperature that is important). Here, degrees Celsius are used. Zison et al. (1978) has reported a range of 1.041.13 for u and a value of 1.065 is commonly employed. However, below 10 8C, the rate will decline faster than indicated by this equation and from 0 to 5 8C it will approach zero. The influence of the oxygen in the overlying sediments is clear. If the DO concentration in the water goes to zero, the SOD will cease. Conversely, above a certain concentration it is usually assumed that the SOD is independent of the oxygen concentration in the overlying water. Baity (1938) found this to be the case when DO concentrations were above just 2 mg O2 ly1. This can be represented by a saturation relationship, KSOD(C)s C KSOD KhsqC (11)

4Hqq2H2Oq2NOy 2

(13)

From the stoichiometry of this reaction, 3.43 g of oxygen will be utilised in the oxidation of each gram of nitrogen to nitrite and the nitrite formed in this reaction can then be oxidised further to nitrate (NO2y) by bacteria of the genus Nitrobac3 ter (Gaudy and Gaudy, 1980):
Nitrobacter

2NOyqO2 2 where KSOD is the new oxygen-dependent SOD rate, C is the concentration of DO in the overlying water and Khs is the half-saturation value for this dependence. Thomann and Mueller (1987) used the data of Fillos and Molof (1972) to estimate Khs and obtained a value of 0.7 mg O2 ly1, which corresponds to independence at oxygen concentrations greater than approximately 3 mg O2 ly1. A simple SOD model would therefore be as follows: dC KSOD1.065(Ty20) s dt H (12)

2NO2y 3

(14)

This reaction consumes an extra 1.14 g of oxygen per gram of nitrite-nitrogen in its oxidation to nitrate. Thus, the total oxygen requirement for the process of nitrifying ammonium to nitrate is 4.57 g for each gram of ammonium-nitrogen. However, some of this ammonium may be used directly in cell production, and so the actual oxygen utilisation may be closer to 4.2 g oxygen per gram of ammonium oxidised (Gaudy and Gaudy, 1980) provided: there are adequate numbers of the nitrifying bacteria available;

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

309

there is an alkaline environment to neutralise the resulting acids; there is sufficient phosphate; there is sufficient oxygen, i.e. at least 12 mg O2 ly1 (Chapra, 1997). The source of nitrifying bacteria is usually the surrounding soils, where one would expect to find them in large numbers. In streams and rivers, the presence of bacteria will generally depend on the available substrate and the nature and treatment of the wastewaters. In order to model the uptake of oxygen, there are two main approaches available. One approach uses a nitrogenous BOD (NBOD) to characterise the process, while the other approach mechanistically models the organic nitrogen, ammonia and nitrate explicitly. The NBOD method uses an overall oxidation rate of the organic and ammoniacal nitrogen together (called the total Kjeldahl nitrogen or TKN). Since no attempt is made to simulate the actual reactions using the NBOD method, this is a semi-empirical approach, but by using an oxygen utilisation of 4.57 g O g Ny1, the NBOD (LN) can be estimated by: LNs4.57TKN (15) and therefore: dC syKnLN (16) dt where Kn is the rate of NBOD oxidation. Clearly, such an approach is rather simplistic and it will not be able to accurately simulate the timings of the real reactions when it uses one process to describe a set of separate reactions. The mechanistic approach which models organic nitrogen (No), ammonium (NH4), nitrite (NO2) and nitrate (NO3) involves the following equations (Chapra, 1997): dNo syKoa dt dwNH4x sKoaNoyKaiwNH4x dt dwNO2x sKaiwNH4xyKinwNO2x dt dwNO3x sKinwNO2x (17) dt

where Koa is the rate of the reaction converting organic nitrogen to ammonium-nitrogen, Kai is the reaction rate of the first stage of nitrification (that oxidises ammonium to nitrite) and Kin is the reaction rate of the second stage of nitrification where nitrite is converted to nitrate. Under low DO conditions (and in the anaerobic conditions of muddy sediments), a process of denitrification may also occur. This denitrification process involves a bacterial reduction of nitrate to nitrogen gas, but the oxygen produced is used by bacteria in respiration and so this will not act as an oxygen source. An approach such as that described by Eq. (17) for nitrogen has been shown to be reasonable (Di Toro, 1976) and, using this method, the rate of consumption of DO due to the nitrogenous oxygen demand can be expressed as: dC s3.43KaiwNH4xq1.14KinwNO2x dt (18)

where C is the concentration of DO. Whilst this method is more elegant than the NBOD method, three parameters rather than one are required. The method also requires greater amounts of information on input concentrations of all four nitrogen species. This means that the semi-empirical method is more suitable in situations where there are insufficient data available to use the mechanistic model. Curtis et al. (1974) studied nitrification in the rivers of the Trent Basin in the UK and found that the growth rates of both Nitrosomas and Nitrobacter bacteria were essentially identical. However, laboratory experiments carried out by Alexander (1965) showed that the Nitrobacter were five times as efficient as the Nitrosomonas in transforming nitrite and ammonium, respectively. This would suggest that the ammonium concentration is the controlling factor for the rate and that the process might be modelled based only on the temperature, and the ammonium and Nitrosomonas concentrations (Knowles and Wakeford, 1978). Denitrification is a biological process, which takes place under anaerobic conditions in the river, for example, in mud and bacterial films on the surfaces of stones gravel and leaves. During this process, nitrate is transformed into nitrogen gas by

310

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

(predominantly Pseudomonas) bacteria, which can escape into the atmosphere due to its low solubility in water. The oxygen produced in this reduction is used by the denitrifying bacteria because of the anaerobic conditions and so does not have an influence on the DO concentrations, but denitrification can be an important process for nitrogen removal in rivers. Thus, in order to have a proper nitrogen balance, this process must be included wherever it is known to occur. 3.4. Chemical oxygen demand Many types of industrial effluents contain substances such as iron sulphite and aldehyde that are oxidised relatively rapidly by DO in a water body. These substances all exert an oxygen demand on the waters DO content and this demand is equivalent to a fast COD. In some rivers, this demand can be very important, but in others where industrial effluent discharges are relatively minor, this is not the case. It should also be noted that the technique for measuring the COD uses high temperatures and strong oxidising agents and so does not reflect natural conditions. Thus, even where data are available, they may not actually reflect a significant sink of oxygen in the natural environment. 3.5. Atmospheric aerationyreaeration Reaeration is the process of absorption of atmospheric oxygen into the water and is regarded as one of the most important factors controlling the waste assimilation capacity of a river because photosynthesis is the only other source of oxygen replenishment and this is limited to daylight hours only. In order to understand the basic mechanism of the transfer of oxygen to any water body from the atmosphere, one can consider a conceptual two-phase system such as a beaker of water. The container of water is open to the atmosphere and can therefore interact with it. If the water is allowed to come to equilibrium with the atmosphere above it, the concentration of DO that will be reached will be fixed for a given temperature and pressure (for that body of water). This is known as the oxygen saturation concentration and

is described by Henrys law, which states that, the mass of any gas that will dissolve into a given volume of liquid, at a constant temperature, is directly proportional to the pressure that the gas exerts above that liquid. Thus, psHeCs (19)

where p is the partial pressure of O2 (mmHg), Cs is the saturation concentration of DO in the liquid (mg O2 ly1) and He is Henrys constant (mmHg/ (mg O2 ly1)). The saturation concentration of DO will be dependent on the air pressure and the temperature and salinity of the water. In lowland freshwater rivers, the most important effect on oxygen saturation is the temperature and various methods are available for calculating its value. The most frequently used equation in water quality modelling is that developed by Elmore and Hayes (1960) for distilled water: Css14.652y(0.41022T) q0.007991T2.y7.7774=10y5T3. (20)

where T is the temperature in degrees Celsius. If the liquid in the conceptual system has its DO concentration decreased very rapidly, say by an input of a reduced pollutant or BOD, the driving force for the restoration of the equilibrium DO condition is the difference between the actual and saturated concentrations. The rate of change of DO with respect to time can be approximated as being proportional to the oxygen deficit, with the exchange taking place across the interfacial area of the waters surface. The proportionality constant is called the reaeration constant Ka and the derivation of this exchange is based on a two-film theory where it is assumed that a gaseous film is at the atmosphere side of the airwater interface and a liquid film present on the side of the water. In order to transfer oxygen from the atmosphere to the water, it is necessary for a parcel of water to travel from the bulk liquid to the interface. Oxygen is then able to diffuse to the water through the gaseous and liquid films. In each film, resistances are encountered and the relatively high Henrys constant for oxygen reflects a high partitioning of oxygen into the gaseous phase relative

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

311

to the liquid phase. Therefore, the gas film has relatively low resistance compared to the liquid film and the liquid film is said to be the limiting control for the exchange. Assuming complete mixing, the flux of oxygen through the controlling liquid film equals the time rate of change of DO in the vessel: Ns i.e. V dC sKLACsyC. dt (22) 1 dm V dC s sKLCsyC. A dt A dt (21)

and the reaeration rate coefficient (Ka) will depend on: internal mixing and turbulence due to velocity gradients and fluctuation; temperature; wind mixing; waterfalls, dams and rapids; surface films because each of these processes will affect the turbulence in the surface layer of the water and thus control the rate of oxygen absorption. However, it is more common for the effects of waterfalls and other flow structures to be dealt with separately such that they act as a direct source of DO into the system. 3.6. Plant respiration and photosynthesis The presence of aquatic plant in a water body will have a marked effect on the DO resources and the variability of DO throughout a day or from day to day (Thomann and Mueller, 1987). Phytoplankton, macrophytes and periphyton all contribute to this effect and are important because of their ability to photosynthesise. During photosynthesis, the chlorophyll-containing plants use solar energy to convert carbon dioxide into carbohydrates and release oxygen in the process. This subjects the water to pure oxygen unlike atmospheric reaeration at the surface that occurs with an atmosphere of only approximately 21% oxygen. Since saturation values refer to the standard atmosphere, photosynthesis can result in supersaturated values and DO concentrations of 150200% of the air saturation are common in highly productive streams (Thomann and Mueller, 1987). Depending on the depth of the stream, two types of plant will tend to grow. In deeper rivers, plants within the water column (phytoplankton) dominate and their distribution declines with depth in relation to decreasing light penetration. In the shallower streams where light can reach the bottom, bottom plants tend to dominate. These plants can include rooted and attached macrophytes and microfloral growths called periphyton. Unfortunately, frequent measuring of the plant activity directly in a stream is not a trivial task, particularly

where, N represents the rate of mass transfer per unit time per unit area, dmydt is the time rate of the mass transfer, A is the area through which diffusion occurs, V is the volume of the liquid and KL is the DO interfacial transfer coefficient (m dayy1) and C is the actual DO concentration in the water (mg O2 ly1). This can also be written as dC sKaCsyC. dt (23)

where Ka is the volumetric reaeration coefficient (dayy1), which reflects the proportionality already discussed, and is given by Kas or Kas KL H (25) KLA V (24)

where H is the ratio of the surface area to volume (m) usually approximated as the depth. The source term for reaeration in a DO massbalance equation such as Eq. (2) may be represented as: KaCsyC. (26)

312

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

Fig. 2. A typical plot of DO concentrations measured by an automatic water quality monitor at Mildenhall on the River Kennet (Williams et al., 2000; Neal et al., 2001).

for shallow streams dominated by periphyton and macrophytes. Because this process requires solar energy, it will only occur during daylight hours. However, the plants and algae will consume oxygen as they respire and this will occur continuously. In a massbalance DO model, therefore, the oxygen terms for representing photosynthesis (P) and respiration (R) will be PyR, and for these two processes only: dC sPyR dt (27)

The photosynthetic production of oxygen depends on several factors including the light intensity, the optical density of the water, water temperature, carbon dioxide and nutrient availability, and the quantity (and type) of plants (Edwards and Owens, 1965). The rate of respiration is often assumed constant, but clearly will also be dependent on oxygen and nutrient availability, and the quantity (and type) of plants. The combination of these two processes can produce strong diurnal (as shown in Fig. 2) and seasonal effects on the DO concentrations. Photosynthesis therefore begins at dawn and ends at dusk and the total photosynthesis of the water column roughly follows the approximately sinusoidal variation of the solar irradiance during the day (Kirk, 1994). However, there can be photo-

inhibition effects close to the surface that will reduce the photosynthetic rate per unit volume. If dinoflagellates are the dominant organisms, there might also be a reduction in the photosynthetic rate during the middle part of the day due to their downward migration to areas of lower light intensity (Kirk, 1994). In temperate and ArcticyAntarctic latitudes, there is a marked seasonal variation in aquatic photosynthesis. This is caused by and is the cause of the seasonal variation in plant biomass. Biomass and photosynthesis rates are both low in the winter due to the decrease in irradiance and temperature. In deeper waters, flow circulation may also take phytoplankton to depths where the light intensity is insufficient for photosynthesis to take place. In spring, irradiance and temperature increase and this, combined with the availability of nutrients, leads to a bloom in phytoplankton and increased photosynthesis. The behaviour during the rest of the seasons tends to be complex and highly variable (Kirk, 1994), affected by zooplankton grazing, fluctuations in nutrient availability, or major changes in species composition due to changes in the water quality. For example, higher pH values in late summer tend to favour bluegreen algae (or cyanobacteria) rather than diatoms. Thus, overall, photosynthesis will tend to dominate during the growing season whilst decomposition and respiration may dominate outside of this growing season. During the growing season, the

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

313

swings in stream oxygen concentration due to diurnal light variations can be enough to leave the water supersaturated in the afternoon and yet severely depleted just before dawn (Chapra, 1997). It is therefore crucial that, particularly in small streams, the effects of plants are not ignored in any river oxygen modelling that is carried out, although very shallow streams might show restricted effects due to interactions with the atmosphere (Kirk, 1994). This applies especially with regard to predicting the effects of waste input to the river system, for example, in simulating the impact of a new STW. In general, values of the average daily areal production rate range from approximately 0.6 to 6 g (O2) my2 dy1 for moderately productive streams. Highly productive streams can range from 6 to 40 g (O2) my2 dy1 (Chapra, 1997). 4. Parameter estimation for modelling DO in lowland rivers The processes affecting DO in rivers that have been described in Section 3 provide a modelling framework to simulate those processes in a massbalance model. The process equations described there all require (reaction) rate parameters and so common procedures for identifying those parameters are reviewed here. Any of the parameters can be evaluated by a process of calibration where the parameter values are adjusted until the simulated data fit observed data. However, in multi-parameter systems it may not be true that the parameters are independent, and so it will not be possible to ensure that a unique set of parameter values is obtained. A variety of statistical methods for parameter estimation also exist, including the extended Kalman filter (Jazwinski, 1970; Whitehead et al., 1981), instrumental variable estimation (Young and Whitehead, 1977), and general sensitivity analysis (Spear and Hornberger, 1980). However, in this section, alternative methods for obtaining estimates of the rate parameters are described that are based on field measurements and theoretical considerations, or empirical relationships.

4.1. BOD oxidation, sedimentation and re-suspension rates The usual method for estimating the BOD rate is by laboratory experiment and involves a similar technique to that used to measure BOD. In the assay, a sewage sample is added to series of closed-batch reactors or bottles containing water with a known DO content. Small amounts of bacteria are then added to each bottle and they are sealed and placed in an incubator. Values of Kr can vary from 0.01 to 0.5 dayy1 depending on the type of waste and the degree of stabilisation although this could rise up to 2.0 dayy1 immediately downstream of effluent discharges (Williams, 1993). Another method of determining the rate constant is to use measurements of BOD at successive stations on a river. This process can be described by solving the BOD Eq. (28) here limited to a constant flow within a channel with a constantgeometry: L L syU yKrL t x (28)

where U is the mean reach velocity and Kr is the total removal rate (dayy1) of BOD, which is composed of oxidation and settling, i.e. KrsKdqKs (29)

Given the constant-flow assumption, at steadystate, this becomes U L syKrL x (30)

and, if complete mixing is assumed at the top station, an initial concentration can be calculated as the flow-weighted average of the upstream concentration in the river (subscript r) and any other input (subscript i): QiLiqQrLr QiqQr

L 0s

(31)

314

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

dation of readily decomposable organics and the settling of sewage particulates. The coefficient Ks for the rate of settling and re-suspension of BOD needs to be determined from BOD profile measurements even though settling and re-suspension are physical processes, governed by the flow conditions and particle size. One might assume that the removal by settling can be estimated given a depth (H) and settling velocity (vs), i.e. Kss vs H (34)

Fig. 3. A plot of BOD downstream from a point source of organic pollution (after Chapra, 1997).

Using this initial condition one can solve Eq. (28) for: LsL0e(yKryU)x (32)

However, one can also use this solution to estimate the removal rate, Kr by taking logarithms, i.e.: x lnLslnL0y Kr U (33)

using typical settling velocities such as 0.10.5 m sy1. However, particulate matter can flocculate into different sizes depending on the particle population in the receiving water (Kranck, 1974) and this will have a marked effect. Under steady-state conditions one might assume that the amount of settling and scour should be equal, but even in this scenario the BOD concentrations in the material settling and that being scoured might not be equal. Because of these factors, the coefficient Ks is usually just an average for the river reach being studied and Dobbins (1964) provides a method for its calculation provided that Kd has already been determined, for example, from laboratory measurements. This technique is similar to the graphical method shown in Fig. 3, but it also accounts for addition of BOD along the reach. 4.2. SOD oxidation rate SOD in rivers is usually measured either by modelling the observed oxygen levels or by direct measurement. The modelling approach is extremely common and requires a DO model to be developed for the water body where all of the rates except for the SOD have been determined. The SOD rate can then be estimated by adjusting its value until the simulated DO values match those observed. Although widespread, this calibration method is flawed, in principle, because it assumes that all of the other model parameters are known with some certainty, when this is rarely the case. The direct measurement of SOD is made by enclosing the sediments and some overlying water

and this has the form of a straight line. Thus, if BOD measurements at successive stations are plotted on a log-scale against xyU (which is also known as the time constant, t), the gradient will be equal to the combined rate Kr if the BOD diminishes as a fixed first-order decay. Fig. 3 shows a typical pattern that might be observed in a stream receiving sewage. In this case, a single straight line is not obtained. Instead, the rate is higher immediately below the discharge. These higher rates are usually caused by the fast degra-

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334 Table 2 SOD values (from Thomann, 1972; Rast and Lee, 1978) Bottom type and location Average value Sphaerotilus (10 g-dry wt. m
y2 20 KSODyg O2 my2 dayy1

315

Range 210 12 12 0.21 0.051 0.062

7 4 1.5 1.5 0.5 0.07

Municipal sewage sludge: Outfall vicinity Further downstream (or aged) Estuarine mud Sandy bottom Mineral soils Areal hypolimnetic oxygen demand (AHOD) in lakes

in a chamber and measuring the concentration of DO in the water over a period of time. This can be carried out in the field or laboratory but has the same flaws as the similar BOD and lightdark bottle tests also described in this review. Studies by Rolley and Owens (1967) showed that this parameter varied considerably from river to river; and that the rates found ranged from 0.144 to 9.84 g O2 my2 dayy1 where there was an overlying DO concentration of 7 mg O2 ly1 at 15 8C. However, they do not explain this variation. Typical values are listed in Table 2 and, in general, values from approximately 1 to 10 g O2 my2 dy1 are probably indicative of enriched sediments (Chapra, 1997). The organism Sphaerotilus sp. (sometimes called sewage fungus despite it being an attached filamentous higher bacterium) is included to indicate that the processing of organic matter in sediments is not the sole domain of simple bacteria. In fact, freshwater (or Zebra) mussels (Dreissena polymorpha) can also have a significant effect (Effler and Siegfried, 1994). 4.3. Nitrification rates If the oxidation of nitrogen compounds is well predicted by an NBOD model, the NBOD rate can be determined in the same manner as described in Section 4.1 for the carbonaceous BOD rate. If this is done, the values found for the rate parameter Kn are approximately the same as for the CBOD (Thomann and Mueller, 1987). Thus, values of between 0.1 and 0.5 dayy1 would be reasonable in deeper rivers, with smaller streams perhaps

displaying rates as high as 1 dayy1 at 20 8C (Thomann and Mueller, 1987). The effect of temperature on the rate coefficient is given by: Kn(T)sKn(20)1.08(Ty20) (35)

for 10(T(30, where T is the temperature in 8C, but values for the temperature coefficient (1.08) have been reported to be in the range 1.0548 1.0997 (Zison et al., 1978). The temperature limitations are given, because the bacteria are not actively multiplying below 10 8C and above 30 8C their activity is severely limited. Therefore, outside of this range, Kn may be assumed zero. If, instead, one assumes that the nitrification rate is limited by the oxidation of ammonium by Nitrosomonas bacteria (Alexander, 1965; Knowles and Wakeford, 1978), then, provided there are sufficient bacteria, the rate can be described by a first-order decay rate such as: dwNH4x dt

sKanwNH4x1.08(Ty20)

(36)

where wNH4x is the concentration of ammonium and Kan is the ammonium nitrification rate at 20 8C found by Knowles and Wakeford (1978) to be between 0.01 and 0.5 dayy1. The process of denitrification may also be assumed to be at a first-order decay rate proportional to the nitrate concentration in the water. Other effects will be due to both the temperature and the bed area since the denitrifying bacteria

316

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

will be predominantly on the river bed. If the effects of the bed area and substrate type are combined into the rate parameter and the temperature dependency is taken to be that of Toms et al. (1975) then: dwNO3x dt and KdensKden,20(1.06981)=100.0293T (38) s(yKden yH)wNO3x (37)

reach. An inert radioactive gas such as Krypton is used as a tracer for oxygen and correlates the rate of desorption of the tracer gas with the rate of adsorption of oxygen. The latter methods are used most commonly in the literature. However, the DO balance method can be very useful since these methods do not require specific experiments to be carried out. All that is required is the information provided by automatic DO monitorsprovided that the data are of a sufficient frequency and quality. Since the latter methods are relatively self-explanatory; only the DO balance methods for measuring the rate parameter are described here. 4.4.1. DO balance methods Techniques have been developed that use the diurnal variation in DO concentrations to estimate reaeration, photosynthesis and respiration rates. These techniques are based on one of two familiar equations, depending on whether it is for day-time or night-time measurements:night-time: dC sKaCsyC.yR dt day-time: dC sKaCsyC.qPyR dt (40) (39)

where wNO3x is the nitrate concentration, Kden is the denitrification rate coefficient, Kden,20 is the rate coefficient at 20 8C and T is the water temperature in degrees Celsius. 4.4. Atmospheric reaeration rates The transfer of oxygen between the water and the atmosphere, as a function of internal mixing and turbulence, has been the subject of much study and investigation. This is because, in most models, the addition of oxygen to the water from the atmosphere is the most significant source of DO. Hornberger and Kelly (1975) expressed that estimates of the (rate of) exchange of oxygen between the atmosphere and water are critical in several areas of water quality control and monitoring, and that these rates are needed for both predicting the effects of waste disposal on a water body and for measuring the rates of community photosynthesis and respiration. Three general techniques are used for measuring the reaeration coefficient Ka of a stream or river. These are (a) the DO balance, (b) the disturbed-equilibrium technique and (c) the tracer technique. The DO balance method consists of measuring the various sources and sinks of DO and determining the amount of reaeration needed to balance the equation. The disturbed-equilibrium method relies on producing an artificial DO deficit (by adding a reduced chemical such as sodium sulphite to the stream) and then measuring the subsequent recovery rate by measuring the DO concentration upstream and downstream of the

where R is the respiratory oxygen uptake rate and P is the photosynthetic oxygen production rate. Solutions to these equations based on continuous measurements of DO have been provided by Odum (1956), Hornberger and Kelly (1975) and Chapra and Di Toro (1991). The oldest, simplest and most widely used method is that of Odum (1956). It estimates Ka from the rates of DO change and saturation deficits just before dawn and just after sunset:

B DC E

C
Kas w
x

Dt

FSSyC
G D

B DC E

Dt

GSR
|

CsyC.SSyCsyC.SRz ~

(41)

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

317

Here the subscripts SS and SR denote the times just after sunset and just before sunrise, respectively; C is the concentration of DO and Cs is the DO saturation concentration. No specific limitations are given on this method, but it will not perform well if the oxygen deficit (CsyC) does not change significantly over the dark period. The method of Odum (1956) only considers the conditions at the beginning and end of the dark period, but more complex methods can be devised that make use of the data from the whole of the dark period or the whole of the light period. Hornberger and Kelly (1975) created two methods of estimating the reaeration rate. The first and simpler method, like that of Odum (1956) uses only that part of the DO curve where there is no influence by sunlight and, therefore, no influence by photosynthesis. This method derives leastsquares estimates of R and Ka by using repeated attempts to predict the post-sunset DO concentrations at the end of sequential time intervals from the concentrations at the beginning of each interval. As with Odum (1956), the starting equations are Eqs. (39) and (40) given at the start of this section and, assuming that the respiration rate has a constant value R0, integrating from time tj to tjq1 yields: R0 Cjq1yCjsCs,jyCj.y 1yeyKd. Ka where dstjq1ytj (43) (42)

where the subscript j refers to sequential time intervals during the night, it can be seen that the value of K that gives the best fit will be obtained by minimising a function, J for the dark hours where:
o o Jj1,j2.s8dj yaj j1qj1j2.2 j

(45)

and the superscript o refers to observed data. A direct minimisation of this function yields:
B
C

j 8a28djy8aj8ajdjE D G
F

j2sy

(46)

N8ajdjy8aj8dj
j j j

and

8dj
j1s
j j

(47)

8ajyNj2
Therefore: Kasy and R0sKaj2 (49) 1 log1yj1. d (48)

and Cs,j is the mean saturation concentration of DO over that interval, calculated from the water temperature. The values of R0 and K may then be estimated by choosing values so that the estimate of Cjq1 best fits (by least-squares methods) the measured data. Simplifying the notation by letting: djsCjq1yCj ajsCsjyCj j1s1yeyKad R0 j2s Ka (44)

This method is very similar to that of Odum (1956), but it uses all of the dark-hours data, and, as with the Odum method, this method will not work well if (CsyC) does not vary much during the night. The second method by Hornberger and Kelly (1975) includes the light hours (or daytime) measurements and so involves solar radiation to estimate the photosynthetic rate, and therefore to estimate the reaeration rate. As before, this involves fitting parameters to an exact solution of the linear DO mass-balance equation Eq. (40) and then using these parameters to predict the DO at the end of sequential time intervals from the

318

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

concentrations at the beginning of each interval. To do this, (PyR) is assumed to be linearly related to the light intensity and that: PsaI and RsR0

(50)

8IjyNm 8dj 8ajyNb bs 8dj


ms where

(56)

where I is the incident radiation and a is a constant of proportionality. Integrating Eq. (40) gives: a R0 Cjq1yCjsCCs,jyCjq Ijy F D K KG 1yeyKd.
B E

8dj8Ij2 8Ijy 8djIj ms


A

(51) bs and

8ajy

To ease notation, if the same values of j1 and j2 as before are used along with a j3s K (52)

8dj8ajIj 8djIj
A

(57)

Ny8dj8Ij As

the best-fit will be obtained by minimising the new J function:


o o o Jj1,j2,j3.s8dj yaj j1yIj j1j3qj1j2. j 2

8djIj

(58)

(53)

So once j1, j2 and j3 have been calculated from Eq. (54), Ka and R0 may be computed from: j1s1yeyKad R0 j2s Ka a j3s Ka

A direct minimisation yields: g b j2smj3qb 1 j1s mj3qb. j3sy where: asym8djIjqmm8djq8I2qNm2y2m8Ij bsymb ajdj djyb djIjqmb dj

(59)

(54)

q28ajIjy2m8ajq2Nmby2b8Ij 2 gsyb ajdjq aj qbb djy2b ajqNb2

(55) and:

The equations are rather long, but do not require solving using iterative calculations like some other day-time solutions. It should be noted that typographical errors were found in three of the intermediate terms as presented on page 734 of Hornberger and Kelly (1975), but that the correct derivations are given here (G.M. Hornberger, personal communication, 2001). A more recent technique has been developed by Chapra and Di Toro (1991) and is mainly used for estimating the rates of photosynthesis and respiration in a water body based on the diurnal DO curves. However, it does produce an estimate of the reaeration rate in the process of doing so.

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

319

The method is described in more detail in Section 4.5. However, it is worth noting here that this method for computing Ka is particularly sensitive when the difference between the time of the minimum DO deficit and the time of the solar noon (i.e. the time lag) is low. Because of this, some authors (e.g. Williams et al., 2000) have attempted to use another estimate of Ka in order to calculate the photosynthesis and respiration rates. A review of these productivity methods by Kosinski (1984) found that all of the techniques were highly accurate, and that this was particularly true for the day-time methods. This review also considered the impacts of noise, data resolution, extreme low or high productivities, temperature variations, and fluctuating light conditions. Noisy data and long observation intervals were not a serious problem for most of the methods and could be improved with the use of a moving DO average instead of single DO data points (Kelly et al., 1974), but low productivity or high reaeration rates were found to cause serious impairments in all of the methods. Therefore, it is possible to obtain a value of the reaeration rate parameter by experimentation or data analysis, but this is not acceptable from a modelling perspective, because it is known that the rate parameter will vary under different stream conditions. 4.4.2. Semi-empirical methods based on mean hydraulic parameters Because the reaeration rate parameter is so important to DO modelling, many attempts have been made to estimate this from other parameters that can be simulated andyor from those that may be assumed not to vary greatly under the expected conditions. Such studies have generally used one of the experimental measurement techniques to estimate the reaeration rate in real streams or flumes in which many hydraulic parameters are also measured. These data are then used either to form an empirical relationship or to test theoreticalyconceptual models of atmospheric aeration and reaeration. Unfortunately, none of the presently available models of the oxygen absorption process in open-channel flows is sufficiently developed to accurately predict the reaeration coefficient from

mean hydraulic parameters alone. As a result, water quality models, instead, generally rely on semi-empirical and empirical regression equations to estimate this coefficient. Numerous studies have been made and many prediction equations have been presented for the reaeration coefficient, Ka. The studies have ranged from theoretical investigations (OConnor and Dobbins, 1956) to empirical field studies (Churchill et al., 1962; Owens et al., 1964; Edwards and Owens, 1965). For the field studies, the general procedure was to measure the reaeration coefficient indirectly by measuring changes in DO under closely controlled conditions (i.e. disturbed-equilibrium techniques). The various studies have covered a wide range of river situations from shallow short-run streams in England, to deep, wide and slow moving rivers in the US. All of the empirical, semi-empirical and conceptual studies are based on a general differential equation of the oxygen balance in a river: dD sKdLyKaD dt and, ignoring BOD oxidation, dD syKaD dt (61) (60)

where D is the DO deficit, L is the BOD concentration, Kd is the BOD oxidation rate and Ka is the reaeration rate. When integrated this becomes: DtsD0eyKat (62)

which is usually presented in the literature as: DtsD010ykat (63)

where D0 is the deficit at time zero and ka is a different reaeration coefficient, related to Ka by Kas2.303ka. This crucial factor is sometimes overlooked in the literature. It needs to be remembered that the rate required for a model is Ka (Eqs. (2) and (22)), which is equivalent to K2 in the original StreeterPhelps equations, and not ka. In this

320

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

section, all parameter estimations quoted will be for the reaeration coefficient Ka in units of dayy1 based on the original work. In addition, most empirical methods are based on the Ohio River findings of Streeter and Phelps (1925), where the reaeration coefficient ka is influenced by the hydraulic and physical characteristics of the river channel. The relationship was formulated empirically as: kas EUn1 H2 (64)

where U is the flow velocity (m sy1), H is the depth (m) and E and n1 represent the (undetermined) empirical constants dependent on the physical and hydraulic conditions of the channel. In the original description, n1 was related to the function mean velocity increase per 5-foot increase in river stage, and the constant E, was related to the slope of the channel and to an irregularity factor, which is a measure of the relative roughness of the channel bed. Unfortunately, with each experiment and theorisation that followed the work by Streeter and Phelps (1925), came a different method to estimate the rate parameter. This is highlighted in the work by Parkhurst and Pomeroy (1972) who say that the literature shows a multiplicity of reaeration formulas with little consensus, and that these formulas cannot all be right. Many of the empirical prediction equations appear to only fit the few sets of data from which they were derived, while the conceptual models suffer because they all contain coefficients that depend on a number of different flow variables. Nonsensically, the combination of coefficients and variables chosen for a given prediction equivalent may not even maintain dimensional consistency. Lau (1972) found from his laboratory experiments that the dimensionless variable Kahyu is a function of both the friction factor f and the Reynolds number uhyy where y is the kinematic velocity of the watershowing that the simple empirical prediction equations that relate Ka to u and h cannot be satisfactory. However, in a number of the semi-empirical cases the coefficient may incorporate one or more physical parameters.

Wilson and Macleod (1974) tested 16 prediction equations against published data covering a wide range of hydraulic variables and found that even the ones with the best correlations gave very unreliable prediction rates. They suggested that the poor performance of these equations might be attributable to one or more missing variables (based on dimensional analysis). Of the equations tested by Wilson and MacLeod, those of Dobbins (1964) and of Parkhurst and Pomeroy (1972) gave the best-fit overall to all of the data investigated. However, Lau (1972) warns that the variation of ka (and therefore Ka) cannot be covered by a single equation. Therefore, the selection of an equation for Ka is usually based on the users judgement by considering the desired complexity, hydraulic situation and available data. However, as a part of this work it has been found that many reviews on this topic have reproduced one or more equations containing errors in that they are not found to be consistent with the original published equations. Some of the confusion arises because, as already noted, theoretical considerations require that the parameter is actually logarithmic and the base of the logarithm is not always reported. In addition, the units of time can be confused between days, hours and seconds, for example. Because of the confusion that exists in the literature, this review has re-examined the original texts for each of the major studies in this field. This was done to ensure that the equations presented here are correct and that consistent use of units is made. Table 4 summarises the results of this extensive review of work covering a period of more than 40 years and the terms used in the equations are described in Table 3. 4.4.3. Summary Within the large range of equations presented here, some patterns have been found. For example, those equations incorporating only simple hydraulic parameters of stream velocity and depth exhibit a tendency to predict higher values of Ka than those observed, except of course for the data set used to formulate that equation (Wilson and Macleod, 1974). It has also been found that equations based on flume data tend to predict higher rates than those based on river data for similar condi-

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334 Table 3 Description of the terms used in Table 4 B C4 F H Q T u* 0.976q0.0137(30-T)1.5 0.9qF Froude numbersUy(6gH) Mean depth Mean flow rate Temperature Shear velocitys6(HSg) CA E g Ad S U W

321

1.0qF 2 Energy dissipation per unit mass of flowsSUg Acceleration due to gravity 9.68q0.054 (T-20) Slope Mean velocity Width

tions (Bennett and Rathburn, 1972). Of the simple empirical equations mentioned in this paper, the best results probably come from the modified equation by Churchill et al. (1962) provided by Isaacs and Gaudy (1968). However, none of the equations incorporating only the simple hydraulic parameters of stream velocity and depth allow accurate predictions of Ka for all flow situations (Wilson and Macleod, 1974). Given that the variation in the predicted coefficients by empirical means appears to be larger than could be attributed to experimental error, it is suggested that variables may have been omitted from the equations (Rathburn, 1977). The equations involving energy terms perform somewhat better than those that do not, especially with regard to the differences between river and laboratory situations (Wilson and Macleod, 1974). Of these methods, those of Dobbins (1964, 1965) and Parkhurst and Pomeroy (1972) give the most generally reliable predictions over the whole range of data, with the Parkhurst and Pomeroy equation producing the lowest standard error of prediction (Wilson and Macleod, 1974). The reviews by Rathburn (1977) and Bennett and Rathburn (1972) also rate the equation by Dobbins to be the most reliable. However, even these best predictions are found to be often lacking when compared with observations and in many cases the observed rates could be several times larger or smaller than those predicted. Wilson and Macleod (1974) suggest that this might be due to uncertainties in the oxygen balance determination. However, it seems likely that the discrepancies arise because of errors in the theory. For example, the equations might lack one or more of the physical variables affecting the reaeration rate.

Using dimensional analysis techniques, the estimates by both Dobbins (1964, 1965) and Parkhurst and Pomeroy (1972) contain incomplete sets of variables. The missing variables could for example, relate to surface-active contaminants. These compounds can lead to a reduction in the observed reaeration rate and yet not one of these estimation methods could predict this effect. Examining partial correlation coefficients, it has been found that of the variables usually incorporated in the rate estimates, the most important was depth, then average velocity and then slope and that the width was negligible (Bennett and Rathburn, 1972). If only the flume data were examined, the order of importance was slightly different with slope, width, depth and average velocity in descending order of importance. There is therefore, no consensus by which reaeration rates can be estimated with certainty on a river without measuring directly (say by a tracer experiment). The productivity methods would appear to be more accurate than those using a relationship based on hydraulic parameters (Kosinski, 1984), but they are far more data intensive and require data of a high quality. Furthermore, productivity methods are not suitable for use in predictive models since they require the data that such a model would be designed to simulate! However, these methods could be useful as a means of improving the simulation of primary production in rivers, where one has access to (even brief) records of high-intensity measurements at important times of the year such as spring and summer (Williams et al., 2000). If temporal variability is very low, then it would also be possible to use the estimates from the more complex productivity methods in a model based on a single

322

Table 4 A summary of the available reaeration coefficient prediction equations from the literature Type of system Equation Range of hydraulic variables used to generate or verify the equation Discharge Mean velocity (m3 sy1) (m sy1) S0.25 1.016(Ty20) H1.25 U0.5 Kas3.952 1.5 1.016Ty20 H Kas11.057 U0.969 Kas5.0140 1.673 1.0241(Ty20) H E0.408 Kas68.4 0.66 1.016(20yT) H Kas B 2.7510BE0.125 E CAAdE0.375 F 1.7535 cothC 1.5 D G C4 H C0.5 4 U0.73 Kas6.9152 1.75 1.0241(Ty20) H U Kas5.1340 1.33 1.0182(Ty20) H B U E Kas4.7531C 1.5 F1.0241(Ty20) DH G U H1.5 uU Kas154 H Kas4.05 Kas3.60 U H1.5
F

Mean depth (m)

Energy slope (my1000 m)

Ka(20) (dayy1)

Number of data points

Reference

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

Conceptual model

0.27 0.061.28 11.28

0.0273.6 0.0411.1

43

OConnor & Dobbins (1956)

Large rivers Recirculating flume

27489

0.126 0.561.52 0.653.48 2.35 0.070.65 0.020.06 0.7541

0.5212.8 24265

30 means of 509 Churchill et al. measurements (1962) Krenkel and 58 Orlob (1962) Dobbins (1964, 1965) Owens et al. (1964) Langbein and Durum (1967) Isaacs and Gaudy (1968) Eloubaldy (1969)*

Conceptual and flume Small streams Large rivers and flume Recirculating cylindrical flume Recirculating 0.61 m flume

0.156 0.081.03 0.040.56 0.120.74 10.6 71.2 15705 0.551.62 1.1615.5 0.090.51 0.300.45 0.170.50 0.150.46 0.170.34 0.050.15 0.431.0

Approximately 0.1100 0.71113 0.172.46 3.0829.8 3174 32 )14 52 12

Recirculating cylindrical flume Recirculating 0.20 m flume Flumes and large rivers

1.311.7

48

Kas10.9C

B U E0.85 DHG B uU E DHG


F1.016(Ty20)

0.200.58 0.050.15 0.01 0.061.52 11.28 0.03 20.38

2043 15.32168

18 131

Isaacs et al. (1969)* Negulescu and Rojanski (1969)* Thackston and Krenkel (1969)

Kas24.86081qF0.5.C

Table 4 (Continued) Type of system Equation Range of hydraulic variables used to generate or verify the equation Discharge Mean velocity (m3 sy1) (m sy1) Mean depth (m) Energy slope (my1000 m) Ka(20) (dayy1) Bennett and Rathburn (1972) Holtje (1972)* Number of data points Reference

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

Small and large rivers Small, mountain streams

U0.607 Kas5.5773 1.689 H Kas2.30(1955.2Ey1657Sq20.87) Kas4440E1.483 Kas2506.7 Kas4.52C Kas


B

0.08489 0.003 0.007

0.126 0.041.52 0.123.48 10.16 0.100 0.030.56 0.020.27 370

0.52113 4.48160

62 97

Flumes and large rivers Recirculating 0.76 m flume

U B uU E3 C F HD U G

0.001 0.003 1.0212


(Ty20)

0.071.52 0.013.48 0.6541

0.52265

140

U E0.703 F D H1.5 G

0.020.14 0.030.19

1.723

239

Lau (1972) Padden and Gloyna (1972)* Parkhurst and Pomeroy (1972) Bansal (1973) Alonso et al. (1975)* Foree (1976)* Grant (1976)* Tsiovoglou and Neal (1976)* Tsiovoglou and Neal (1976)* Thyssen and Jeppesen (1980)* Thyssen et al. (1987)

Sewers Medium to large rivers Recirculating 0.359 m flume Small to medium sized streams Small streams Non-tidal streams including pools and waterfalls Small streams

23.0400

1q0.17F2.(SU)0.375
H
0.6

0.01 0.249 27489

U Kas4.1528 1.4 1.016(Ty20) H uU Kas123 H Kas0.8880.30q5570.8S0.5. Kas22700SU Kas31200SU Kas3170S U0.76(1qF)2.66S1.13 H0.60 8784U0.734S0.93 kas H0.42 Kas23000

0.453.60 0.050.48 0.7168.4 2.21155 0.27 0.061.52 11.28 0.0273.6 0.0412.8 24113 0.3420.7 1.849

74 73 19 20 59

0.290.88 0.030.11 0.013 11.6 07.90 0.011.05 0.080.38 0.213.3 0.095 56.8

0.00830 0.008 0.23

-0.67

0.08305

605 28

Small, lowland streams Shallow macrophyte rich streams

0.390 0.020.35 0.060.32 0.130.50 2.03 0.022.9 0.060.52 0.121.37 0.37.4

1.015.0 0.297.7

29 144

323

324

Table 4 (Continued) Type of system Equation Range of hydraulic variables used to generate or verify the equation Discharge Mean velocity (m3 sy1) (m sy1) (US)0.524 UQ0.242 0.556, pool and riffle (US)0.528 Kas596 0.136 UQ 0.556, pool and riffle (US)0.313 Kas88 UH0.353 0.556, pool and riffle (US)0.333 Kas142 0.66 0.243 UH W 0.556, channel control Kas517 Large streams Mean depth (m) Energy slope (my1000 m) Ka(20) (dayy1) Number of data points Reference

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

0.0028 210

0.003 1.83

0.0457 3.05

0.0160

99

Melching and Flores (1999)

References with asterisks have information taken from the review by Thyssen et al. (1987), not from the original source. All terms are as used in Table 3.

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334 Table 5 Flow structure parameters after Department of the Environment (1973) Dam type Flat broad-crested regular step Flat broad-crested irregular step Flat broad-crested vertical face Flat broad-crested straight slope face Flat broad-crested curved face Round broad-crested curved face Sharp-crested straight slope face Sharp-crested vertical face Sluice gates with submerged discharge b 0.70 0.80 0.80 0.90 0.75 0.60 1.05 0.80 0.05

325

4.6. Plant respiration and photosynthesis rates Various methods have been described for determining the gross primary production, total community respiration rates in flowing water systems. These include light and dark bottle techniques, 14 C-bicarbonate uptake techniques (e.g. Kevern and Ball, 1965) and the methods based on DO evolution in the river that were described in the previous section (e.g. Odum, 1956; Owens et al., 1969; Kelly et al., 1974; Hornberger and Kelly, 1975; Chapra and Di Toro, 1991). 4.6.1. Light and dark bottle method The light and dark bottle technique is similar in principle to the measurement of BOD, in that a bottle environment is used to simulate the chemical environment of the water body. The technique involves placing a sample of the water into two bottles. One bottle is clear or light allowing both photosynthesis and respiration to occur. The other bottle does not allow light to penetrate and is thus dark. This second bottle should exhibit respiration only. The DO concentration in each bottle is determined and then they are sealed and placed in the river. Some time later the bottles are opened and the new DO concentration measured. The difference in the concentrations can then be used to estimate the photosynthesis and respiration rates in the following way. If the experiment was conducted over a period of time t, and the initial and final concentrations in the light bottle are denoted Cli and Clf, then if one assumes a zero-order (i.e. constant rate) reaction, the net photosynthetic rate can be determined by: Pnets ClfyCli t (66)

set of continuous data. If this is not the case, then it is likely that one will have to rely on the hydraulic methods where suitable data are more likely to be available. 4.5. Enhanced aeration at flow structures Structures present in the river such as weirs and dams can influence aeration and increase DO concentrations by 13 mg O2 ly1 over very short distances (Bowie et al., 1985). Numerous predictive equations have been developed to simulate the effects of these structures and these are reviewed by Butts and Evans (1983). A report by the Department of the Environment in the UK (1978) identified nine classes of structures, and quantified an aeration coefficient (b) for use in the following formula: CsyCu s1q0.38abh(1y0.11h) CsyCd =(1q0.046T)

rs

(65)

where Cs is the saturation concentration of DO, Cu is the concentration above the structure and Cd is the concentration below the structure. The parameter a is a water quality factor and ranges from 0.65 for grossly polluted streams to 1.8 for clean streams, h is the head loss over the structure in meters and T is the water temperature in degrees Celsius. Values for the parameters b are based on the type of structure as described in Table 5.

where Pnet is defined as: PnetsPbyRc (67)

and the subscript b is used to distinguish that this is a bottle rate. Rc is the community respiration rate, i.e. it reflects plant respiration, but also any other oxygen consuming reactions such as bacterial respiration.

326

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

The community respiration in the dark bottle is determined in a similar fashion: Rcs CdiyCdf t (68)

where the initial and final concentrations of DO in the dark bottle are Cdi and Cdf, respectively. Pb can then be calculated using Eqs. (66) and (67). The community respiration can be corrected to plant respiration (or Rb, because this is a bottle rate) by accounting for the BOD, i.e.: RbsRcyKdLfi (69)
Fig. 4. The variation in photosynthesis over the diurnal cycle as represented by a half-sinusoid.

where Kd is the BOD oxidation rate described in Section 4.1 and Lfi is the initial filtered BOD. This technique provides average rates over the period of the study, but whilst the respiration might be assumed constant it is clear that this is unlikely to be the case for photosynthesis. If constancy is the case, then some manipulation of Pb is required in order to obtain maximum and average photosynthesis rate. The main drive for this variation is the variation in light intensity over a diurnal cycle, which can be idealised as a half-sinusoid (Chapra, 1997). In most water-quality models, photosynthesis is assumed directly proportional to the available energy from sunlight (although this is not strictly true (Kirk, 1994), i.e.: P(t)AI(t) (70)

By integrating Eq. (71) between the beginning (t1) and end (t2) of the test and setting the result equal to Pb(t2yt1), the maximum photosynthesis rate can be found. So: p | P sinx fT tyt .|dtsP tyt .
t2 t1 m
y w z ~

(72)

and integrating gives: PmsPbt2yt1. fTp W w T p z p z p cosx |t1ytr.ycosx |t2ytr.X T y fTp ~ y fTp ~ Y
S T U T V w

(73)

where P is the rate of primary oxygen production and I(t) is the available light. If this holds true, the photosynthesis rate can also be represented by a half-sinusoid: P(t)sPmsinwvtytr.ztr-t-ts y ~
x |

The average daily rate can then be found by taking an average over the whole day, i.e.:

| Ps
(71)

Tp 0

P(t)dt 2f sPm Tp p

(74)

P(t)s0

otherwise

where Pm is the maximum rate, v is the angular frequency, tr is the time of sunrise, ts is the time of sunset, f is the photoperiod (i.e. the fraction of the day subject to sunlight) and Tp is the daily period as illustrated in Fig. 4 below.

However, the lightdark bottle method for measuring gross primary production of the community suffers from the same problems as BOD bottle measurements. Thus, the rate found is seldom applicable in flowing waters because much of the community is benthic and heterogeneous rather than planktonic. Furthermore, any measure-

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

327

ment made without the normal turbulent flow may be questioned on the grounds that production is a function of current flow, Odum (1956). 4.6.2. Dissolved oxygen balance An alternative method for estimating rates of photosynthesis and respiration is based on a linear DO mass-balance equation and high-quality continuous DO data as described in the previous section (e.g. Odum, 1956). Such methods are less expensive and therefore ideal for small systems, but they do have limitations and do require highquality inputs in order to get good results. In the years since Odums work others such as OConnor and Di Toro (1970), Schurr and Ruchti (1975), Erdmann (1979a,b) have extended and refined this idea and more recently Chapra and Di Toro (1991) have given a graphical method. The methods of Schurr and Ruchti (1975) and Chapra and Di Toro (1991) are described here, but the methods of Odum (1956) and Hornberger and Kelly (1975) described in the previous section may also be used for the purpose of estimating community photosynthesis and respiration rates. Schurr and Ruchti (1975) employed two different methods in their study. The first, a solution of rate equations is similar to the methods already described and the second is a cross-correlation computational technique on DO records. Both methods are explained in this investigation and applied to several Swiss rivers to obtain estimates of the oxygen exchange constant, and the rates of photosynthesis and respiration. The basic principle for both methods comes from an idea of Einstein (1920): When a chemical reaction in the steady-state is subjected to a cosinusoidal (in time) external perturbation of any of the thermodynamic variables, the periodic response of the reactants to the perturbation will lag behind in phase (or time) by an amount depending upon the finite speed of the chemical process itself, and will exhibit an amplitude of oscillation reflecting the fundamental susceptibility or inducibility of the system to the perturbation. In this situation, the chemical process is reaeration, the steady-state condition is that maintained in the presence of constant respiration, and the

cosinusoidal perturbation is brought about by photosynthesis. The simple model used by Schurr and Ruchti (1975) is: C syKCqKCAqaI(t)yb t (75)

where K is the outgassing constant or exchange constant, K is the invasion constant, CA is the (constant) concentration of oxygen in the atmosphere, b is the constant total respiration rate, I(t) is the instantaneous light intensity at time t on the surface of the water and aI(t) is the instantaneous rate of photosynthesis at time t. It is assumed that the light intensity is given by a constant plus a 24-h cosinusoidal function, i.e.: I(t)s I0 (1qcos Vt) 2 (76)

where Vs2py24, and t is in hours. Separating both C and I into a mean plus a time-dependent part, produces the following pair of equations: C(t)sCqDC(t) I0 I0 I(t)sIqDI(t)s q cos Vt 2 2 (77)

and substituting the initial simple model equation reveals: aI0y2 b aI0y2 KCA b Cs yq sCsyq K K K K K (DC) aI0 qKDCs cos Vt dt 2

(78)

Solving this resulting pair of equations using standard methods such as Fourier transforms gives: DC(t)sA cos V(tyT) (79)

where the time delay, T is the difference between the time of the DO maximum and the time of

328

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

maximum sunlight (or solar noon) given by: Ts


BVE 1 tany1CF DKG V

(80)

and the amplitude of the response, A is one half of the range of the DO variations, given by: aI0y2 As B E 2 2 DV qK G0.5
C F

(81)

Thus, measurement of the three parameters T, A and O in conjunction with the oxygen saturation concentration will provide estimates of the rate parameters: K, K, aI0 y2 and b, i.e. the rates of oxygen exchange, photosynthesis and respiration. The implications of using an ideal sunlight model are recognised by the authors (Schurr and Ruchti, 1975) and so they also provide a method that incorporates a random opacity factor to represent the random fluctuations in cloud cover and atmospheric dust, etc. that one might experience. However, it was found to be unsuccessful in that the random noise can (and did) prevent the identification of secondary maxima, which would negate the use of the simple model. Instead, the authors (Schurr and Ruchti, 1975) suggest that the second method is used that effectively averages out the random effects of the clouds. This crosscorrelation method uses the cross-correlation function between the ideal sunlight and the measured DO concentrations to produce estimates of K and A (actually the amplitude of the cross-correlation function), and then uses the same set of equations as before to obtain the remaining rates. Both techniques were applied to data collected from several rivers in Switzerland, where the data had previously been manipulated so that missing data points would not upset the cross-correlation calculations. It was found that, where the model was successful, the two methods gave very similar estimated of the rates so long as the data only exhibited single maxima each day. However, it was also found that both methods and the first method in particular were very poor at producing reliable estimates when the DO concentrations in

the rivers were affected by other influences such as power stations and dams. The Delta method of Chapra and Di Toro (1991) is similar to that of Schurr and Ruchti (1975) in that it too uses the time-lag and amplitude of the diurnal DO variations to estimate values of the reaeration rate (Ka), average daily photosynthetic rate (Pav) and daily average respiration rate (R). In this method, the rates are calculated from a series of diurnal DO curves using a piecewise solution. This is based on a mass-balance DO model provided by OConnor and Di Toro (1970), but simplified for the case where the DO deficit, D does not vary spatially i.e.: dD qKaDsRyP(t) dt (82)

where D is the oxygen deficit (mg O2 ly1), t is the time in days and P(t) is the primary production rate (mg O2 ly1 dayy1). The components of the diurnal DO cycle are illustrated in Fig. 5 where D is equivalent to twice the amplitude in the Schurr and Ruchti (1975) technique and f is equivalent to the time-delay, T. The rates are then calculated by carrying out three sequential steps. Firstly, the reaeration rate (Ka) is found from the time difference, f between the minimum DO deficit and the solar noon. From the above model (Eq. (82)), there is a unique relationship between f and Ka, such that f is controlled solely by the reaeration rate. An analytical expression of this relationship is difficult to obtain as it involves the solution of a transcendental equation, but a numerical solution is relatively straightforward and is presented by Chapra and Di Toro (1991) as a series of curves relating ka to f for various day lengths (f). It is, however, possible to recalculate these curves as look-up tables using the latitude and longitude for the area of interest. Once the reaeration rate has been specified, the average daily photosynthetic oxygen production rate Pav is calculated using the value of Ka and the deficit range (D). As before, there is a unique relationship between these parameters and again the analytical expression of this relationship is difficult to obtain, but Chapra and Di Toro (1991)

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

329

Fig. 5. A graphical description of the components of the diurnal DO cycle used in the Delta method to estimate the rates of photosynthesis, reaeration and respiration in a stream (after Chapra and Di Toro, 1991).

present a series of approximations for different ranges of Ka, i.e.:


B f E2 D fTyfq0.2C F DTG Pav

on to the estimates of the other rates. It is suggested here that under those conditions an alternative method is chosen for calculating the reaeration rate. 4.6.3. Summary The simple DO model used by the DO balance methods, and their solutions, involve some significant assumptions and limitations. Since photosynthesis is represented as a half-sinusoid, the solution will not account for shading by clouds, banks and valleys or overhanging plants. Furthermore, respiration is assumed constant throughout the day and night (this assumption is also in the other productivity methods). An improvement is suggested here whereby actual solar radiation data could be used to account for overcast skies, but this may not help significantly if the actual radiation input does not follow the assumed half-sinusoid pattern. The rates of photosynthesis and respiration are not easily obtainable from theoretical considerations and that bottle measurements may not be characteristic of the whole stream. It is therefore not surprising that some authors have attempted to derive more simplistic empirical estimates. Investigations by the Water Pollution Research Laboratory (1968) of photosynthesis in freshwater streams found that, in streams not suffering from nutrient limitation, the rate of oxygen production

Ka-1.0 dayy1

1yey0.5Ka.2 D s q0.05110.1(Ka(5.0 dayy1 Pav 0.5Ka1yeyKa. (83) D p y1 s Ka)5.0 day 2 Pav yKa q(2p)2
where T is the period (i.e. 1 day) where, if f (the length of the day) is given in hours, T will be 24 h, but if f is given in days then T will be 1 day. Finally, the respiration rate may be obtained from the calculated estimates of Pav and Ka by: RsPavqKaD (84)

where D is the mean daily oxygen deficit (mg y1 O2 l ). It should be noted that the calculations of the Delta method are highly dependent on an accurate estimate of fespecially when the time-lag is short (i.e. the stream has a high reaeration rate) and the relationship becomes extremely sensitive. In such situations, any errors in Ka will be passed

330

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

was related to the incident solar radiation and the concentration of chlorophyll-a in the water. A least-squares analysis revealed that the following equation fitted the observed data best: Ps(0.95q31.7Chl)I0.28 (85)

106CO2q16NHqqHPO2yq108H2O 4 4 C106H263O110N16Pq107O2q14Hq

(87)

or the following equation for nitrate based systems: 106CO2q16NOyqHPO2yq122H2O 3 4 q18Hq C106H263O110N16Pq138O2

where P is the quantity of oxygen produced per unit area per day (g O2 my2 dayy1), Chl is the concentration of chlorophyll-a (mg ly1 ) and I is the total light energy incident at the water surface (Cal cmy2 dayy1). A study on the Rivers Thames and Kennet near Reading by Kowalczewski and Lack (1971) suggested that the net oxygen production rate is related to: temperature, total solar radiation, chlorophyll concentration and euphotic depth (i.e. the depth to which sufficient light is transmitted). Regressions performed on the data found that the net production in the River Thames could be expressed by: Pnetsy0.05Iq0.02ChlaD.y0.47 (86)

(88)

If the photosynthetic quotient were one, the mass conversion from oxygen to carbon would be 12y32 or 0.375. Adjusting for the primary nitrogen source the mass conversion rates become: Ammonium source, 0.3715 Nitrate source, 0.2880 and the carbon biomass production rate can be calculated from the photosynthetic rate by multiplying by the appropriate factor. Since most measurements of algae in rivers use a measure of the pigment in the algae (chlorophyll-a) rather than the algal biomass itself, the chlorophyll-a (Chl-a) may be estimated using a fixed carbon to Chl-a ratio of 1 g of carbon to 12 mg of Chl-a (Chapra, 1997). 5. Modelling dissolved oxygen in lowland rivers From the preceding discussion, it is clear that any (mechanistic) model that attempts to simulate the concentration of DO in a lowland river will, at the very least, have to simulate the processes described in this review and summarised in Fig. 6 below. These processes will occur all along any river that is being modelled, and most models will use a discretisation whereby the river is split into a series of reaches. Within a river reach conservation of mass must be maintained, but changes in the DO concentration may occur due to physical transport and transformation processes such as those described in this review. The description of those processes in a mathematical model requires a hydraulic model in order to simulate the transport of the solutes along the river as well as including any biological, chemical and physical conversion processes that are to be simulated. The model may

where Pnet is the net production rate (g O2 my2 dayy1), I is the total solar radiation (kCal my2 dayy1), Chla is the areal concentration of chlorophyll (mg Chl-a my2) and D is the euphotic depth (m). If a model requires the simulation of organic carbon as well as oxygen, the average composition of a plant will need to be considered since this will not be CH2O, due to the presence of proteins, lipids and nucleic acids, etc. as well as carbohydrate. Thus, to simulate the growth in say an algal population in-stream, the ratio O2:CO2 (known as the photosynthetic quotient) that should be utilised will not be exactly one. In most models, the nutrient levels are assumed to be sufficient enough to not be a limiting factor and in many lowland rivers with relatively high effluent discharge this is reasonable. In such situations, the photosynthetic quotient depends only whether the primary nitrogen source is ammonium (NHq) or nitrate 4 (NO2y). In either case, the summary reaction for 3 photosynthesis (or indeed respiration) will be given by the following equation for ammonium-based systems:

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

331

Fig. 6. A schematic of the major processes influencing the concentration of DO in rivers.

be described by well-known extended transport equations such as (Rauch et al., 1998):

B E c c yv c yw c q C c F syu x t x y z x D x G B B E E Cy c Fq Cz c FqDc q y D y G z D z G

(89)

where is an multi-dimensional mass concentrac tion vector for each of the determinants; t is the time; x, y and z are spatial coordinates; u, v and w are the corresponding velocity components, x, y and z are turbulent diffusion coefficients for the directions x, y and z, respectively; and Dc is a term representing the rates of change of determinants due to internal transformations in the reach (e.g. nitrification or reaeration). This partial differential equation (or PDE) can be solved numerically, but is usually simplified to some extent before solving. For example, if the concentrations in the reach do not vary greatly over the depth or the cross-sectional area, the number of dimensions may be reduced and this leads to the so-called advectiondispersion approach. Alternatively, the

system might be assumed to consist of a number of interconnected, perfectly mixed tanks or elements, which leads to a set of ordinary differential equations (or ODEs). A model of this type (simulating the processes in Fig. 6) will invariably require a large amount of data, especially if it is to simulate time-varying (i.e. dynamic) changes in the system. The first set of data required describes the physical characteristics of the system such as the layout of the network, the division of the river system into reaches, and the widths, depths and lengths of those reaches. The reach descriptions will also include the type and dimensions of any flow structures such as weirs whose effects are intended to be simulated. At the top of the system, a model needs the upstream boundary conditions of flow, DO, BOD, ammonium, nitrate and chlorophyll-a (or some other indicator of the plant density) and in addition to this, the pH and temperature will be required. The temperature determines the rate at which the various reactions will take place, and, if this is to be a dynamic model, these conditions will need to be time series for each determinant for the period of interest. In more complex hydraulic models, boundary conditions must also be supplied at the bottom of the system. At each

332

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334

reach a similar set of data will be required for any influences such as tributaries or effluent discharges that enter the river at that point, together with the flow rates of any abstractions. 5.1. Data availability Often models are designed with little regard to the data requirements for that model. This is unfortunate, because such models will be impractical if it turns out that certain parameters are not generally measured or worse are impossible to measure. Thus, it is common for field studies to accompany modelling exercises so that the required data might be obtained, but if a model can be designed that is satisfactory using only commonly measured determinants then this will clearly be preferable. Much of the required data may be available having been collected by organisations such as the Environment Agency or research institutes, or as a part of an ongoing study such as LOIS. Within each reach, these determinants may be transformed according to the processes described, and the rate parameter for each reaction in each reach will be required. It has been shown in Section 4.3 that these rates are all obtainable either from theoretical considerations or by experiment, if the necessary data are available. They may also be obtained by calibration of each rate in turn in a specific order, so that the act of variation of one parameter has as small an affect as possible on the determinants that have already shown a good fit to observed data. However, this blind calibration method should ideally require a large number of observed data points in order to ensure a good fit. Furthermore, this method should only be used as a last resort because one cannot be certain that there will be a unique set of parameters that facilitates a good fit of the simulated and observed data. It is therefore suggested that as many parameters as possible should first be estimated using theoretical or experimental techniques before resorting to blind calibration, although it is perhaps reasonable that minor adjustments may be made to estimates of rate coefficients once they have been determined if this improves model performance

6. Conclusions This review has described the processes that influence DO, in lowland river systems and the data requirements for using models that simulate these processes have been outlined. It is has been shown how mathematical models can be used to describe these processes so that one can simulate both the condition and the variability in these systems. Although mechanistic models have been described, it is clear that there is some degree of uncertainty in these equationsrelating to the reaction rate parameters in particular. The most commonly used techniques for estimating the rates of these processes have been described, but it is clear that there will be inherent uncertainty in applying any model based on the process formulations described here. Therefore, the effects of uncertainty must be analysed in any water quality model so that the results obtained may be interpreted correctly before any management decisions are made based on those results. Acknowledgments This review results from a NERCyEnvironment Agency funded Ph.D. (GT2y97y3yEA), and the author is grateful to NERC and the Environment Agency for the financial support granted, also to Paul Whitehead (The University of Reading) and Colin Neal (CEH, Wallingford) for supervising the project. References
Alexander M. Nitrification. Agronomy 1965;10:307 343. Alonso CV, McHenry JR, Hong J-CS. The influence of suspended sediment on the reaeration of uniform streams. Water Res 1975;9:695 700. Baity HG. Some factors affecting the aerobic decomposition of sewage sludge deposits. Sewage Works J 1938;10(2):539 568. Bansal MK. Atmospheric reaeration in natural streams. Water Res 1973;7:769 782. Bennett JP, Rathburn RE. Reaeration in open-channel flow. USGS Professional paper. 1972;737:75. Butts TA, Evans RL. Small stream channel dam aeration characteristics. J Environ Eng 1983;109(3):555 573. Bowie GL, Mills WB, Pocella DP, Campbell CL, Pagenkopf JR, Rupp GL, Johnson KM, Chan PWH, Gherini S, Cham-

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334 berlain CE. Rates, constants, and kinetic formulations in surface water quality modelling. Athens, Georgia, USA: US Environmental Protection Agency, 1985. Camp TR. Water and its impurities. New York: Reinhold Book Corp, 1963. p. 355 Chapra SC, Di Toro DM. Delta method for estimating primary production, respiration, and reaeration in streams. J Environ Eng 1991;117(5):640 655. Chapra SC. Surface water-quality modeling. McGraw-Hill International editions, 1997. Churchill MA, Elmore HL, Buckingham RA. Prediction of stream reaeration rates. Int J Air Water Pollut 1962;6:467 504. Curtis EJC, Durrant K, Harman MMI. Nitrification in rivers in the Trent Basin. Water Resour 1974;9:255 268. Department of the Environment. Notes on water pollution. In: Aeration at Weirs. Department of the Environmental Water Research Laboratory, Elder Way, Stevenage, Herts., England, June 1973;61. Di Toro DM. Combining chemical equilibrium and phytoplankton models. In: Canale RP, editor. Modeling biochemical processes in aquatic ecosystems. Ann Arbor, MI: Ann Arbor Science, 1976. p. 233 255. Di Toro DM, Paquin PR, Subburama K, Gruber DA. Sediment oxygen demand model: methane and ammonium oxidation. J Environ Eng Div ASCE 1990;116(5):945 986. Dobbins WE. BOD and oxygen relationships in streams. J Sanit Eng Div ASCE 1964;90(SA3):53 78. Dobbins WE. Closure to BOD and oxygen relationships in streams. J Sanit Eng Div ASCE 1965;91(SA5):49 558. Edwards RW, Owens M. The oxygen balance of streams. Ecology and the Industrial Society, Fifth Symposium of the British Ecological Society. Oxford: Blackwell; 1965:149 172. Effler SW, Siegfried C. Zebra mussel populations in the Seneca River: impact on oxygen resources. Environ Sci Technol 1994;28(12):2216 2221. Einstein A. Schallausbreitung in teilweise dissoziierten Gasen. Sitzber Preuss Akad Wiss Physik Math Kl 1920;380. Elmore HL, Hayes TW. Solubility of atmospheric oxygen in water. Twenty-ninth progress report of the committee on sanitary engineering research. J Sanit Eng Div ASCE 1960;86(SA4):4153. Eloubaldy AF. Wind waves and the reaeration coefficient in open channel flow. Ph.D. Thesis, Colorado State University. Fort Collins, Colorado; 1969. Erdmann JB. Systematic diurnal curve analysis. L Water Pollut Control Fed 1979a;15(1):7886. Erdmann JB. Simplified diurnal curve analysis. J Environ Eng Div ASCE 1979b;105(6):10631074. Fillos J, Molof AH. Effect of benthal deposits on oxygen and nutrient economy of flowing waters. J Water Pollut Control Fed 1972;44(4):644 662. Foree EG. Reaeration and velocity prediction for small streams. J Environ Eng Div ASCE 1976;102:937 952. Gaudy AF, Gaudy ET. Microbiology for environmental scientists and engineers. New York: McGraw-Hill, 1980. p. 736

333

Grant RS. Reaeration-coefficient measurement of 10 small streams in Wisconsin using radioactive tracers. With a section on the energy dissipation model. Geol Surv Water Resourc Div, Madison, Wisconsin; 1976. Holtje RK. Reaeration in small mountain streams. Ph.D. Thesis, Oregon State University. Oregon; 1972. Hornberger GM, Kelly MG. Atmospheric reaeration in a river using productivity analysis. J Environ Eng Div ASCE 1975;101(EE5):729 739. Isaacs WP, Gaudy AF. Atmospheric oxygenation in a simulated stream. J Sanit Eng Div ASCE 1968;94(SA2):319 344. Isaacs WP, Chulavachana, Bogart R. An experimental study of the effect of channel surface roughness on the reaeration rate coefficient. Proceedings of the 24th Industrial Waste Conference, Purdue University. 1969. p. 14641476. Jazwinski AH. Stochastic processes and filtering theory. New York: Academic Press, 1970. Kelly MG, Hornberger GM, Cosby BJ. Continuous automated measurements of rates of photosynthesis and respiration in an undisturbed river community. Limnol Oceanogr 1974;19:305 312. Kevern NR, Ball RC. Primary productivity and energy relationships in artificial streams. Limnol Oceanogr 1965;10:74 87. Kirk JTO. Light and photosynthesis in aquatic environments, 2nd ed. Cambridge University Press, 1994. p. 509 Knowles G, Wakeford AC. A mathematical deterministic river quality model: Part I: formulation and description. Water Res 1978;12:1149 1153. Kosinski RJ. A comparison of the accuracy and precision of several open-water oxygen productivity techniques. Hydrobiologia 1984;119:139 148. Kowalczewski A, Lack TJ. Primary production and respiration of the phytoplankton of the River Thames and Kennet at Reading. Freshwat Biol 1971;1:197. Kranck K. The role of flocculation in the transport of particulate pollutants in the marine environment. Proceedings of the International Conference on the Transport of Persistent Chemicals in Aquatic Ecosystems. Ottawa, Ontario, Canada; 1974. Krenkel PA, Orlob GT. Turbulent diffusion and the reaeration coefficient. J Sanit Eng Div ASCE 1962;88(SA2):53 84. Langbein WB, Durum WH. The aeration capacity of streams. USGS Circul 1967;542:6. Lau YL. Predicition equation for reaeration in open-channel flow. J Sanit Eng Div ASCE 1972;98(SA6):1063 1068. Melching CS, Flores HE. Reaeration equations derived from US geological survey database. J Environ Eng 1999;125(5):407 414. Neal C, Watts C, Williams RJ, Neal M, Hill L, Wickham H. Diurnal and longer term patterns in carbon dioxide and calcite saturation for the River Kennet, south-eastern England. Sci Total Environ, 2001;282:205231. Negulescu M, Rojanski V. Recent research to determine reaeration coefficient. Water Res 1969;3:189 202.

334

B.A. Cox / The Science of the Total Environment 314 316 (2003) 303334 Thackston EL, Krenkel PA. Reaeration prediction in natural streams. J Sanit Eng Div ASCE 1969;95(SA1):65 94. Theriault EJ. The oxygen demand of polluted rivers. I Bibliographical: A critical view. II Experimental: The rate of deoxygenation, vol. 173. Washington DC: Treasury Dept. USPHS Public Health Bulletin, 1927. p. 185 Thomann RV. Systems analysis and water quality management. New York: McGraw-Hill, 1972. Thomann RV, Mueller JA. Principles of water quality modeling and control. NY: Harper Collins, 1987. Thyssen N, Erlandsen M, Jeppesen E, Ursin C. Reaeration of oxygen in shallow, macrophyte rich streams: IDetermination of the reaeration rate coefficient. Int Revue ges Hydrobiol 1987;72(4):405 429. Thyssen N, Jeppesen E. Genluftning I mindre vandlb. Vatten 1980;36:231 248 (In Danish). Toms IP, Mindenhall MJ, Harman MMI. Factors affecting the removal of nitrate by sediments from rivers, lagoons and lakes. Medmenham, UK: Water Research Centre, 1975. Tsiovoglou EC, Neal LA. Tracer measurement of reaeration: IIIPredicting the reaeration capacity of inland streams. J Water Pollut Control Fed 1976;48:2669 2689. Water Pollution Research Laboratory. Water Pollution Research. HMSO. 1968. p. 3560. Whitehead P, Beck B, OConnell E. A systems model of streamflow and water quality in the Bedford Ouse river systemII. Water quality modelling. Water Res 1981;15:1157 1171. Williams RJ. QUASAR Technical guide V2.0: Notes on calibrating QUASAR. Institute of Hydrology, UK; 1993. Williams RJ, White C, Harrow ML, Neal C. Temporal and small-scale variations of dissolved oxygen in the Rivers Thames, Pang and Kennet, UK. Sci Total Environ 2000;251y 252:497 510. Wilson GT, Macleod N. A critical appraisal of empirical equations and models for the prediction of the coefficient of reaeration of deoxygenated water. Water Res 1974;8:341 366. Young PC, Whitehead PG. A recursive approach to time-series analysis for multivariable systems. Int J Control 1977;25:457 482. Zison SW, Mills WB, Mills, Diemer D, Chen CW. Rates, constants and kinetic formulations in surface water quality modelling. ORD, Athens, GA, ERL: Tetra Tech, Inc. for USEPA. EPA 600-3-88-105. 1978. p. 317.

OConnor DJ, Di Toro DM. Photosynthesis and oxygen balance in streams. J Sanit Eng Div ASCE 1970;96(2):547 571. OConnor DJ, Dobbins WE. Mechanism of reaeration in natural streams. Trans Am Soc Civil Eng 1956;123:641 667. Odum HT. Primary production in flowing waters. Limnol Oceanog 1956;1(2):795 801. Owens M, Edwards R, Gibbs J. Some reaeration studies in streams. Int J Air Water Pollut 1964;8:469 486. Owens M, Knowles G, Clark A. The prediction of the distribution of dissolved oxygen in rivers. Advances in water pollution research. Proceedings of the Fourth International Conference. Pergammon; 1969. p. 125147. Padden TJ, Gloyna EF. Simulation of stream processes in a model river. Technical Report No. 2 (EHE-70-23, CRWR72). Austin, Texas: Texas University, Center for Reasearch in Water Resources; 1972. Parkhurst JD, Pomeroy RD. Oxygen absorption in streams. J Sanit Eng Div ASCE 1972;98(SA1):101 124. Rast W, Lee GF. Summary analysis of the North American (US Portion) OECD Eutrophication Project: Nutrient loading-lake response relationships and trophic state indices. Corvallis, OR: USEPA, Corvallis Environmental Research Laboratory. EPA-600y3-78-008. 1978. p. 454. Rathburn RE. Reaeration coefficients of streamsState of the art. J Hydraulics Div ASCE 1977;103(HY4):409 424. Rauch W, Henze M, Koncsos L, Reichert P, Shanahan P, Somlyody L, Vanrolleghem P. River water quality modelling: I. State of the art. Water Sci Technol 1998;38(11):237 244. Rolley HJL, Owens M. Oxygen consumption rates and some chemical properties of river muds. Water Res 1967;1:759 766. Schurr JM, Ruchti J. Kinetics of oxygen exchange, photosynthesis, and respiration in rivers determined from timedelayed correlations between sunlight and dissolved oxygen. Schweiz Z Hydrol 1975;37(1):144 174. Spear RC, Hornberger GM. Eutrophication in Peel InletII. Identification of critical uncertainties via generalized sensitivity analysis. Water Res 1980;14:43 49. Streeter HW, Phelps EB. A study of the pollution and natural purification of the Ohio River. III Factors concerned in the phenomena of oxidation and reaeration. US Public Health Serv Public Health Bull 1925;146:75.

You might also like