Gulf of Maine Harmful Algal Bloom in summer 2005 - Part 2: Coupled Bio-physical Numerical Modeling

Abstract

A coupled physical/biological modeling system was used to hindcast the 2005 Alexandrium fundyense bloom in the Gulf of Maine and investigate the relative importance of factors governing the bloom's initiation and development. The coupled system consists of a state-of-the-art, free-surface primitive equation Regional Ocean Modeling System (ROMS) tailored for the Gulf of Maine (GOM) using a multi-nested configuration, and a population dynamics model for A. fundyense. The system was forced by realistic momentum and buoyancy fluxes, tides, river runoff, observed A. fundyense benthic cyst abundance, and climatological nutrient fields. Extensive comparisons were made between simulated (both physical and biological) fields and in-situ observations, revealing that the hindcast model is capable of reproducing the temporal evolution and spatial distribution of the 2005 bloom. Sensitivity experiments were then performed to distinguish the roles of three major factors hypothesized to contribute to the bloom: 1) the high abundance of cysts in western GOM sediments; 2) strong northeaster storms with prevailing downwelling-favorable winds; and 3) a large amount of fresh water input due to abundant rainfall and heavy snowmelt. Results suggested that the high abundance of cysts in western GOM was the primary factor of the 2005 bloom. Wind forcing was an important regulator, as episodic bursts of northeast winds caused onshore advection of offshore populations. These downwelling favorable winds accelerated the alongshore flow, resulting in transport of high cell concentrations into Massachusetts Bay. A large regional bloom would still have happened, however, even with normal or typical winds for that period. Anomalously high river runoff in 2005 resulted in stronger buoyant plumes/currents, which facilitated the transport of cell population to the western GOM. While affecting nearshore cell abundance in Massachusetts Bay, the buoyant plumes were confined near to the coast, and had limited impact on the gulf-wide bloom distribution.

Keywords: Bio-physical Numerical Modeling; Coastal circulation; Harmful Algal Bloom.

Figure 1

Multi-nested configuration for the regional GOM circulation and A. fundyense bloom modeling. The outermost model is the data assimilative North Atlantic HYCOM, which provides open boundary conditions (OBCs) for the shelf-scale ROMS model, which in turn provides OBCs for the innermost Gulf of Maine ROMS model.

Figure 2

Improved mortality function used in the A. fundyense population dynamics model.

Figure 3

Comparison of benthic cyst distribution and abundance between 2004 (upper panel) and 1997 (lower panel). The 2004 cyst abundance data was used to produce the central hindcast of 2005 A. f. bloom.

Figure 4

Comparisons of observed (gray) and modeled (thick) subtidal sea levels at 6 coastal gauges in the GOM.

Figure 5

Comparison of observed (left) and simulated (right) time series at GoMOOS mooring B. For both panels, from top to bottom are depths profiles of across-shelf current, along-shelf current, temperature and salinity. Circles in the left hand panels indicate locations of current and temperature/conductivity sensors on mooring B. Three major transport events in May and June are indicated by arrows in both observed and simulated Along-shelf current fields.

Figure 6

Comparison of observed (left panels) and simulated (middle right) surface A. fundyense cell concentration. In-situ cell counts were collected by in-situ ship surveys in May, June 2005. Temporal and spatial evolutions of modeled bloom conditions on March 15^th, April 15^th, May 15^th, and June 15^th. In each panel, both surface and vertical cell distributions are shown. Note for clear visualization, cell concentration is converted in log₁₀ scale.

Figure 7

Temporal and 3-dimentional spatial evolutions of simulated bloom conditions on March 15^th, April 15^th, May 15^th, and June 15^th. In each panel, surface cell concentration map is shown on top of 7 GOM sections showing vertical cell distributions. Note for better visualization, both surface and vertical cell concentrations are scaled by in log₁₀.

Figure 8

Simulated surface cell concentrations from three sensitivity model experiments. Modeled blooms are sampled on the same observation grid as shown in Figure 6. Left columns show the results from twin experiment 1 that using 1997 benthic cyst distribution; Middle columns show the results from twin experiment 2 that using 2004 surface wind forcing; right columns show the results from twin experiment 3 that using 2004 river runoff data.

Figure 9

Artificial benthic cyst abundance data used in sensitivity experiment 4.

Figure 10

Comparison of May monthly-mean cell concentration produced by central hindcast (upper panel), and twin experiment 4(lower panel) that utilized artificial cyst abundance data (Figure 9).