Ecosystem Modeling II Kate Hedstrom, ARSC November 2006 With help from Sarah Hinckley, Katja Fennel, Georgina Gibson, and Hal Batchelder

Outline Simple NPZ model Mid-Atlantic Bight model Gulf of Alaska model GLOBEC Northeast Pacific ROMS options

Annual Phytoplankton Cycle Strong vertical mixing in winter, low sun angle keep phytoplankton numbers low Spring sun and reduced winds contribute to stratification, lead to spring bloom Stratification prevents mixing from bringing up fresh nutrients, plants become nutrient limited, also zooplankton eat down the plants

Annual Cycle Continued In the fall, the grazing animals have declined or gone into winter dormancy, early storms bring in nutrients, get a smaller fall bloom Winter storms and reduced sun lead to reduced numbers of plants in spite of ample nutrients We want to model these processes as simply as possible

Franks et al. (1986) Model Simple NPZ (nutrient, phytoplankton, zooplankton) model Three equations in the three state variables Closed system - total is conserved Units?

The Equations P Change = nutrient uptake - mortality(P) - grazing Z Change = growth efficiency * grazing - mortality(Z) N Change = - nutrient uptake + mortality(P) + mortality(Z) + (1 - growth efficiency) * grazing

Goal Initial conditions are low P, Z, high N The goal is to reproduce a strong spring bloom, followed by reduced numbers for both P and N Hope to find a steady balance between growth of P and grazing by Z after the bloom Note: at 1000 steps per day, they can get away with the Euler timestep

Original Franks Model

Different Grazing Function

Half the Initial N

Half the Initial Value for All

2.5 Times the Z Mortality

NPZ Results There’s more than one way to damp the oscillations –Change initial conditions –Change grazing function –Change mortality constant It’s not clear if any of these methods is “right” Current practice is to add detritus falling out of the mixed layer (NPZD model)

Simple NPZD Attempt

NPZD Summary The Franks model is attempting to describe the ocean mixed layer at one point Next is a 1-D vertical profile of the biology, including light input at the surface Final goal is a full 3-D model with the physical model providing temperature, currents, seasonal cycle, etc. Each component is advected and diffused in the same manner as temperature

More Modern Models Seven, ten, or more variables –Large and small zooplankton –Large and small phytoplankton –More (specific) nutrients –Detritus Still missing some fields such as gelatinous zooplankton (salmon food) Models are tuned for specific regions, specific questions

NO 3 Chlorophyll Large detritus Organic matter N2N2 NH 4 NO 3 Water column Sediment Phytoplankton NH 4 Mineralization Uptake Nitrification Grazing Mortality Zooplankton Susp. particles Aerobic mineralization Denitrification Katja Fennel’s Model

Fennel’s Model Fasham-type model In shallow coastal waters, the sinking particles hit the bottom where nutrient remineralization can occur Add a benthos which contributes a flux of NH4 as a bottom boundary condition Her goal is to track the nitrogen fluxes and extrapolate to carbon

Nested physical- biological model

North Atlantic ROMS Climatolo- gical heat/fresh- water fluxes 3-day average NCEP winds SSH (m)

Middle Atlantic Bight (MAB) Cape Hatteras Hudson Delaware Chesapeake Georges Bank Nantucket Shoals 50, 100, 200, m isobaths: dashed lines

3.3 DIN 0.9 PON 2.9 PON 2.5 DIN DNF: 5.3 TN Rivers: 1.8 TN cross-isobath export of PON, import of DINcross-isobath export of PON, import of DIN x mol N y TN 4.2 TN denitrification removes 90% of all N entering MAB denitrification removes 90% of all N entering MAB Fennel et al. (submitted to Global Biogeochemical Cycles)

Gulf of Alaska Sarah Hinckley and Liz Dobbins built a full NPZD model for the shelf and tested it in 1-D They then ran a 3-D Gulf of Alaska with that model (3 km resolution) The shelf waters bloomed, as did the offshore waters

The Iron Story They believe that offshore waters don’t bloom due to iron limitation Iron is brought to the coastal ocean via river sediments Not enough iron data to do a full model of it Trace quantities of iron are difficult to measure on steel ships

More Iron People have done experiments at sea, seeding the water with iron - the ocean does bloom after Adding an iron component to their NPZD model allowed Sarah and Liz to obtain more realistic looking results It’s not a full iron model, containing nudging to a climatology

Method (Iron Limitation) FeLim: a multiplicative factor affecting growth of both small and large Phytoplankton. Affects large phytoplankton more strongly Michaelis-Menton function (as in Fennel et al. 2003) Iron depletion and nudging back to climatology Does not follow iron through whole ecosystem Sarah Hinckley

Iron Climatology

No Iron Limitation – NO3 Kodiak I. PWS Alaska Peninsula

Iron Limitation – NO3 Alaska Peninsula PWS

Northeast Pacific GLOBEC Goal is one model that runs off California and Gulf of Alaska Get the fish prey right, then move on to IBM’s –Track individuals as Lagrangian particles - euphausiids and juvenile coho salmon –Include behavior - swimming vertically –Growth depends on food availability, temperature –Use stored physical and NPZ fields from ROMS

ROMS Biology Options NPZD Fasham-type Two phytoplankton-class model (Lima and Doney) –Small –Diatoms –Silica and carbon Multiple phytoplankton-class model (ECOSIM) –Four phytoplankton classes –Internal carbon and nitrate for each –No zooplankton

Computer Issues Adding biology doubles or triples the computer time needed for the physical model These models need more time for more grid points (480x480x42 for CGOA) and also a shorter timestep if the grid spacing is small They can take weeks of computer time, using many processors in parallel

Conclusions Biological modeling is an area of active research Coupled biogeochemical systems are hot too Still things to figure out, including the rest of the annual cycle, interannual variability and other changes