Modeling Studies of Circulation and Marine Fish Early Life History in the Coastal Gulf of Alaska

A.J. Hermann, PMEL/JISAO & S. Hinckley, AFSC

P.J. Stabeno, PMEL, B.A. Megrey, AFSC, J.M. Napp, AFSC


Accurately model coastal circulation in the Gulf of Alaska and its effects on commercially important fish and their prey.


Development of coupled physical and biological models

Circulation (Hermann, Stabeno)
  • Semispectral Primitive Equation Model (SPEM) of Haidvogel et al.
  • three-dimensional, eddy resolving (4 km resolution)
  • driven by 12-hour winds and monthly runoff
Prey (Hinckley, Napp) - Nutrient- Phytoplankton- Zooplankton model (NPZ)
  • spatially explicit; fixed spatial grid (Eulerian)
  • zooplankton segregated by stage and/or species
  • includes relevant prey items for pollock
Arrows indicate typical paths of pre-spawning adults, larvae, and early juvenile starges. Inset map shows regional circulation. Hatched area indicates region of spawning. Pollock (Hinckley) - Individual Based Model (IBM)
  • spatially explicit; follow representative individuals through space and time (pseudo-Lagrangian)
  • stochastic
  • includes egg, larval, and juvenile stages
Stored output from the physical model is used to drive biological models.

Hindcasting of circulation and biology in the coastal Gulf of Alaska

  • After tuning the physical model with hydrographic, current meter, and drogued drifter data, we ran hindcasts of circulation and biology for 1978, 1987, 1988, 1989, 1991, and 1994 (movie). These years span a broad range of wind, runoff and recruitment conditions.
Salinity (psu) and surface velocity (m s-1) from physical model formid-July, 1987
  • Mesoscale statistics of the circulation field (frequency and rotational sense of eddies) and mean currents are captured by the physical model.
Depth-integrated of larvae (#/m2) in mid-May of 1987 and 1989.
  • Measured interannual differences in pollock larval density are captured by the individual-based model. Weaker advection in 1989 led to a higher concentration of larvae near the exit of Shelikof Strait in mid-May, relative to 1987.
  • Measured spatial pattern of surface chlorophyll-a (spring climatology from Coastal Zone Color Scanner archives) is similar to spatial pattern from NPZ model (spring 1987). Higher densities of phytoplankton (and zooplankton) occur near the southwest exit of Shelikof Strait.
Cross section of the mean alongshore current (cm s-1) off Cape Kekurnoi for the period April 8-September 30, 1991. Current meter locations are indicated at the top. Surface chlorophyll (ug/l) from the NPZ model,versus CZCS data.

Exploring the effects of spawning location on early life history

  • Why does a population of walleye pollock spawn at the exit of Shelikof Strait each spring? One hypothesis is that this population evolved to optimize physical transport to the juvenile nursery area near the Shumagin Islands 375 km to the southwest. Alternatively, factors other than physical transport (e.g. density of prey) may be significant.
  • We addressed these hypotheses with the coupled suite of physical and biological models, driven by winds and runoff appropriate to two years of good recruitment, 1978 and 1994. Five regions (1-5) and four spawning times (Early, Middle, Late, Very Late) were considered. "1-Middle" represents typical observed spawning.
Initial locations of spawned populations in the experiment. Transport of fish to the juvenile nursery area promotes successful recruitment for a given year.
  • Animated results show that fish spawned to the south of Kodiak Island (3-Middle) or much earlier or later than the observed spawning period (e.g. 1-Very Late) do not reach the Shumagin Island nursery area as juveniles by early September. However, the region and time of spawning which did allow successful transport to the nursery area (e.g. 4-Late) was much broader than the observed region and time. Hence factors other than physical transport alone must be considered to explain the spawning location and timing of this stock.
Final locations of spawned
populations for 1994.

Future Directions:

  • Extend coupled model approach to the Bering Sea for southeast Bering Sea Carrying Capacity (SEBSCC) program and to the wider Gulf of Alaska for West Coast Global Ecosystem Dynamics (GLOBEC) program. Extend physics to include tidal and wind mixing.
  • Embed regional models in a global Spectral Element Ocean Model (SEOM) (with Dale Haidvogel, Rutgers University). Global simulations include tides and wind forcing; regional simulation include tides, winds, and runoff.
  • Apply coupled model approach to Gulf of Alaska salmon early life history (with Peter Rand, N.C. State University), for GLOBEC.
Pacific region of the global SEOM grid. Regions of new SEBSCC and GLOBEC models are indicated. Flow chart of models in development for the SEBSCC program.

For more information, please refer to:

Funding support provided by NOAA Fisheries Oceanography Coordinated Investigations and Coastal Ocean Programs