FjordEnv and MOM presentation at ECASA Workshop - WP

FjordEnv and MOM presentation at ECASA Workshop - WP4 - 2006-01-26
Anders Stigebrandt

The holding capacity is determined by environmental effects caused
by the turnover of organic matter in fish farms, in particular effects on the bottom beneath the farm (A scale) in the fish cages (A-scale) on the surrounding inshore environment (B-scale)

Two computer programs have been developed
FjordEnv - computes water exchange of different strata in fjords and other inshore areas and how this is influenced by changes of e.g. the topography of the mouth. It also computes how outlets of particulate organic matter and nutrients, e.g. from fish farming, influence Secchi Depth and deepwater oxygen conditions in the whole system. MOM – computes the holding capacity of sites for fish farming based on acceptable changes of the state of the local environment (the sea bed) and on expected extreme states with respect to oxygen and ammonium concentrations in the fish cages. The holding capacity is given by the lowest of the estimates from the local and the regional aspects, i.e. the lowest of the estimates given by MOM and FjordEnv, respectively.

The MOM Model for estimation of the holding capacity of sites
for fish farming The holding capacity is estimated with respect to three basic environmental requirements: The benthic fauna at a farm site must not be allowed to disappear due to accumulation of organic material; The water quality in the net pens must be kept high; The water quality in the areas surrounding the farm must not deteriorate. All these requirements must be fulfilled, and the holding capacity is determined by the lowest of the three estimates.

The dispersion sub-model
Excess feed and faeces will be dispersed and for a large part settle under or in the neighborhood of the farm. Where and how much will settle depends on the amount and the disintegration of the effluent, the sinking velocity of the particles, the current speed and the water depth. Particle tracking models have been used to estimate the settling from farms. The MOM model considers the settling from single net pens. The spatial distribution of particle sedimentation under a fish farm then becomes a function of the pen size, the separation between pens and their configuration. The dispersion model computes how the sedimentation at the distance r from the cage centre F2(r) (g m-2 day-1) is related to the emission from the net pen F1 (g m-2 day-1) The sedimentation of carbon out of a net pen F1C is given by F1feed and F1faeces can be obtained from the fish model

The dimensionless dispersion function μ(r) attains values between 0 and 1. It is called
the normalized sedimentation or loading function. If the mean current vanishes, maximum sedimentation occurs beneath the centre of a net pen (r=0). We use the variance of the current, σ2, to estimate the dispersion of particles. The dispersion increases with the variability of the current and with the sinking time T=H/w of the particles. Here H is the distance between the net pens and the bottom and w is the sinking speed of the particles. The dispersion capacity of a location is given by the dispersion length σT= σH/w (m).

The normalised loading (sedimentation) function (r) versus distance r from
the centre of the cage for different values of T (m) (see legend). The mean current vanishes and the cage size is15x15 m2. (from Stigebrandt and Aure, 1995).

The following conclusions were drawn by Stigebrandt & Aure (1995);
1) The dispersion at a site may be described by the dispersion length σT. By using feed with lower sinking speed or feed which disintegrates easily into smaller particles, the sinking time T may be increased and thereby the dispersion of excess feed. 2) The sedimentation on the seabed outside the vertical projection of a single net pen increases and the maximum loading μ(0)F1 beneath the net pen decreases with increasing dispersion length and decreasing pen area. 3) The maximum loading under a fish farm decreases if i) the separation between net pens is increased, ii) pen size is decreased, iii) the number of net pen rows is decreased. The pen rows should be oriented perpendicular to the direction of the strongest currents. The holding capacity of a site which is limited by the assimilative capacity of the benthic community may thus be increased in several ways.

To compute the maximum sedimentation rate at a certain site, the dispersion
sub-model uses filed estimates of the (r) function for single net pens, for the values of T and L that are applicable to the farm. Note that because of its much lower sinking speed, T is much larger for faeces than for conventional feed. The sedimentation rate at any point on the seabed is obtained by adding the sedimentation from all net pens in the farm with specified number of rows R and distance S between net pens. The dispersion sub-model computes the maximum values feed and faeces for excess feed and faeces, respectively, under the farm. The maximum carbon flux F2Cmax to the sediment under the farm (gC m-2 day-1) is computed as: where feed (faeces) is the maximum loading with excess feed (and/or faeces) that takes into account contributions from all cages as computed using the dispersion sub-model.

Maximum normalised loading (sedimentation) versus distance S between
cages for a standard farm with 1, 2 or 3 rows (see legend). For these computations T=10 m and L=11 m (from Stigebrandt and Aure, 1995).

If the current speed above the bottom occasionally exceeds a certain threshold
value, accumulated organic material will be resuspended and may be transported away from the site. Cromey et al. (2002) estimated that the threshold current value for resuspension of organic matter from fish farms is about 10 cm s-1. This is in accordance with Panchang and Newell (1997) who estimated that the threshold is in the interval between cm s-1. Current speeds in this range should occur occasionally if the variance of the bottom current is 4-6 (cm s-1)2, and the frequency of such events should increase with the current variance, as shown in Stigebrandt and Aure (1995).

The benthic sub-model. Aerobic decomposition of organic matter in sediments requires that oxygen be supplied to the sediments from the overlying water. Sediments beneath marine fish farms are susceptible to oxygen depletion if the sedimentation rate of excess feed and faeces reaches a critical level. With insufficient oxygen supply to the sediments, anaerobic decomposition will prevail and the sediments may produce high concentrations of hydrogen sulphide, resulting in azooic sediments. The task of the benthic sub-model is to compute the maximum rate of loading of organic matter that does not lead to extinction of the benthic infauna. Stigebrandt and Aure (1995) developed the benthic impact model used in MOM. They assume that aerobic benthic metabolism is limited by the maximum rate of oxygen delivery to the sediments. They argue that the latter is determined by the turbulent diffusion of oxygen across the turbulent bottom boundary layer. Oxygen consumption by infauna cannot be greater than oxygen delivery and it requires oxygen concentrations to be sufficiently high over time. This determines the maximum rate of sedimentation of particulate matter from the farm.

The flux of organic material from the farm that settles at the bottom (F2) does not
necessarily represent the amount of organic material being decomposed in the sediment. Some of the material may be transported away by strong bottom currents and by animals and oxidise outside the farm area. The fraction of the particulate organic matter from the farm that is oxidised within the farm area is called  (0<<1). The vertical oxygen flux necessary to completely decompose the settled material will be  F2, where  is the amount of oxygen necessary to oxidise one gram of organic carbon to carbon dioxide and water. If the specific flux of oxygen to the sediment is FO2 (g O2 m-2 s-1) one expects at steady state that: Thus, if one knows ,  and FO2, F2 can be computed. In the MOM model, we use for  the standard value 2.7 gO2/gC. The following formula for FO2 is from Stigebrandt and Aure (1995) where O2i is the oxygen concentration just above the turbulent benthic boundary layer, O2bent is the oxygen concentration at the sediment surface and Ubent the horizontal current velocity just above the turbulent benthic boundary layer. Ubent may be looked upon as the effective vertical velocity that transfers oxygen to the bottom. Theoretically, this should be equal to the effective vertical velocity CDUbent that transfers horizontal momentum towards the bottom. The coefficient  should thus have a value equal to that of the drag coefficient (CD). In the MOM model it is tentatively assumed that  equals 210-3.

Maximum oxygen transport to the bottom occurs when the difference O2i - O2bent is
at a maximum, which for a given O2i occurs when O2bent equals O2min, the lowest oxygen concentration that will allow the benthic infauna to survive. For the calculations we use the maximum sedimentation rate F2max (g C m-2 s-1) that occurs beneath the cage centres if there is no mean current, as discussed in section 5. By combining equations (5) and (6) we obtain the maximum acceptable sedimentation on the bottom: The relationship between measured currents and the dimensioning current velocity Ubent used in the model is discussed in section 8. An expression for the maximum potential fish production at a fish farm that does not lead to extinction of the benthic infauna, TPFbentam, can be derived using Eqs. (1), (2), (4) and (7), thus: where FCR is the actual feed conversion ratio, FCRt is the theoretical feed conversion ratio, AF is the total area of the cages in the farm and feed (faeces) the maximum specific loading with excess feed (faeces) accounting for contributions from all cages.

MOM Summary: MOM Applicable to scale A (local). Computes loading of organic matter from a fish farm on the bottom. Computes the maximum loading with benthic animals present beneath the farm. Computes oxygen and ammonium conditions in the fish cages. Requires measurements of currents (time series) and oxygen conditions. MOM model results influences the monitoring program at a site. Exists as a PC program. Programmed in Visual Basic. Some testing of the processes has been performed The model has been used rather much in Norway to compute the holding capacity of sites. The formula for maximal loading allowed with benthic animals does not describe the process (it has not yet been tested thoroughly). (Black Swan)

MOM – model FrordEnv model
Fish farm site Surrounding area Dispersion sub-model Fish sub-model Benthic sub-model Regional water quality model Water quality sub-model

Models needed to estimate holding capacity
Fish model to compute the turnover of matter by fish Dispersion model to compute dispersion of particulate matter emitted by a farm. Benthic model to compute the oxygen transport to the sea bed and from that the maximum loading with organic matter that allows a fauna of benthic animals Fish cage model to compute oxygen and ammonium concentrations in the cages Water quality (Secchi depth) model of surface waters in inshore water Water quality (oxygen concentration) model of basin waters in fjords Model of the natural flow of organic matter into basin waters of fjords The models are based on first order process understanding and kept as simple as possible

Summary of water exchange above sill level
Name Type Forcing Estuarine circulation Baroclinic Local. Freshwater supply to & wind mixing in the fjord Steady outflow from surface & inflow to intermediary layer Isopycnal pumping*) Remote. Density varia-tions due to winds and tides outside the fjord Synoptic inflow and outflow in surface and intermediary layers External sea level pumping Barotropic Remote. Sea level variations due to tides, winds & air pressure outside the fjord Alternating inflow and outflow above sill level. *)In Stigebrandt (2001), “isopycnal pumping” is denoted “intermediary circulation” which might be misleading.

Main processes for the deepwater include
Vertical flux of organic matter and oxygen consumption in deeper layers. The natural supply of organic matter eqauls in the simplest case Energy supply to basin water turbulenceThe mean rate of work against the buoyancy forces in a column of the basin water, W, may be computed from the following expression, see Stigebrandt and Aure (1989). Deepwater circulation Tides play a major role for mixing in the basin water of fjords. Increased tidal velocities across sills usually give increased supply of mixing energy that leads to increased rates of mixing of less dense water into basins. An increased rate of density reduction leads to more frequent exchanges of basin water and by that improved oxygen conditions.

Rate of deepwater circulation.
In Aure and Stigebrandt (1989a) a method was developed to estimate the time lapsed between two consecutive complete exchanges of basin water in fjords. It was demonstrated that the rate of density reduction d/dt in basin water is proportional to W, and the following relationship was suggested The empirical constant CW=2.00.6 and W is obtained from verified theory. One may expect that the basin water is completely exchanged during the period Te defined by Here Re is the mean density reduction in the basin water needed to obtain complete exchange of basin water. Empirical data from fjords in Møre and Romsdal show that one may expect basin water renewal when the mean density reduction since the last exchange is Re≈-4/3 (kg m-3). It should be pointed out that the water higher up in the basin is exchanged more often and the effective Re –value thus decreases upwards.

Water quality in the surface layer – Secchi depth
Due to changes of the nutrient supply to the inshore surface layer, the Secchi depth will change from a known depth. Nutrients coming from local sources, e.g. by local runoff and dissolved from fish farms, may directly be used by phytoplankton in the fjord. It is assumed that the nutrients are mixed into the surface layer. The nutrient supply from fish farming will vary with number and weight of fish, temperature and food composition as described in Stigebrandt (1999b). Water exchange in the surface layer is assumed to be proportional to the total water exchange above sill depth. The constant of proportionality is assumed to be Vs/Vt where Vs is the volume of the surface layer and Vt is the volume of the fjord above sill level. In addition comes the net outflow caused by the freshwater supply and the estuarine circulation. If the latter two are assumed to take place in the surface layer, the water exchange Qs in this layer is The changed concentration of phosphorus in the surface layer, cPf (mmol P m-3), due to changed excretion by fish in fish farms and outlets from human activities is then

The changed supply of phosphorus is Pf (mmol s-1)
The changed supply of phosphorus is Pf (mmol s-1). If it is assumed that all the phosphorus is used for plant production, the concentration of plankton (measured as P) increases by cPf. The Secchi depth then becomes Here D0 is the “normal” Secchi depth (i.e. the depth before the supply is changed by Pf) and Eq. (2.5) has been used. Eq. (5.3) may of course also be used to estimate increases in Secchi depth for cases with decreasing supply of nutrients (negative values of Pf). The computations above were done for P. One may equally well do the computations for N if this is considered to limit the production of organic matter.

Water quality in the basin water
The flux of particulate organic matter into a fjord basin is FC (gC month-1 m-2). In general the flux is composed of both natural marine matter and matter directly and indirectly produced by human activities, e.g. fish farming. If FC is known, the expected volume mean oxygen consumption in the basin water is given by Here  is the amount of oxygen needed to oxidise organic matter measured as carbon. In the computations  is assigned the value 3.5 gO2 (gC)-1. Eq. (6.1) should give the real oxygen consumption but the apparent oxygen consumption may be less because oxygen may be transported into the basin water by turbulent diffusion, see Aure and Stigebrandt (1989b). The natural part of FC, enforced by offshore conditions, may be estimated from Eq. (2.8) with z=Ht, Lm=50 m and FC0 equal to the established regional value. In some fjords with fish farming uneaten food and faeces from the farms may contribute to the flux of organic matter into the basin water.

The time-scale TO to reduce the oxygen concentration in the basin water by the
amount O2 is If O2 is taken equal to the oxygen concentration of new basin water, O2in, one obtains the time-scale for complete oxygen depletion. This measure usually underestimates the observed time scale because i) some oxygen is supplied to the basin water by diffusion as mentioned above and ii) as observed by Aure and Stigebrandt (1989a,b), the rate of oxygen consumption decreases when the oxygen concentration becomes lower than about 2 mlO2/l, the lowest concentration accepted by higher forms of life. It is obvious that the lowest oxygen concentration in the basin water will occur at the end of a stagnation period. The length of stagnation periods Te is determined by the physics, see section 4.5 If Te < TO, the basin water will still contain oxygen when exchanged. However, if TO< Te the basin water will during a stagnation period eventually be depleted in oxygen. Hydrogen sulphide will then appear, first in the deepest parts. The concentration of oxygen at the end of a stagnation period, O2min, is thus dependent on both the initial oxygen concentration (O2in) and the relative lengths of TO and Te and is given by

The relationship O2min and Te/TO is visualised in Fig. 6.1.

FjordEnv Summary: Applicable to predict water quality changes (e.g. Secchi depth or Chl-a concentrations) in the surface layer and (e.g. oxygen consumption and oxygen conditions) in the deepwater in scale B systems (fjords, bays, etc) due to e.g. fish farming. Requires very few measurements. Exists as a stand-alone PC program. Programmed in Visual Basic. Most of the processes used in FjordEnv have been described in scientific publications. Tests of parts of the model have been performed in Norway. The model has been used extensively in Norway since 1992 Indicators e.g. Secchi depth or Chl-a concentrations in the surface layer and oxygen consumption and oxygen conditions in deepwaters. Black Swan: If the model is applied outside its range of applicability. ECASA plans: To couple CSTT and FjordEnv To test the performance of the model for as many ECASA inshore sites as possible