which the information contained in the observations is
projected onto the (unobserved) model state estimate.
Advanced DA techniques use time-variable model dynam-
ics to actively interpolate information from observations
up- and downstream and forward and backward in time.
Observations are assimilated over a time interval, given
the temporal evolution of the circulation (e.g., Moore et al.,
2020). Identifying observations that best constrain an
ocean model can drive improved observing system design
for more accurate and more cost-effective prediction.
Observation impact studies aim to quantify how specific
observation types, locations, and observing frequencies
impact model estimates (e.g., Oke et al., 2015).
In this article, we assess observation impact in a dynamic
western boundary current (WBC). WBCs are swift, pole-
ward-flowing currents that exist on the western sides of
subtropical ocean gyres. They transport warm water from
the tropics toward the poles, redistributing heat and mod-
ulating global climate. Mesoscale eddies form due to insta-
bilities in the strong boundary current flow, making WBC
extension regions hotspots of high eddy variability (Imawaki
et al., 2013; Li et al., 2022a). WBCs typically exhibit the high-
est errors in ocean forecasts (e.g., Brassington et al., 2023)
due to their strong flows, the complexities of eddy shedding
and evolution (e.g., Kang and Curchitser, 2013; Pilo et al.,
2015; Yang et al., 2018), and their complex vertical struc-
tures (e.g., Sun et al., 2017; Pilo et al., 2018; Brokaw et al.,
2020; Rykova and Oke, 2022). Understanding the interplay
of observing system design and modeling approaches is
crucial to improving prediction in highly dynamic, eddy-rich
oceanographic environments.
The East Australian Current (EAC) is the WBC of the
South Pacific subtropical gyre, and its eddies dominate
the circulation along the southeastern coast of Australia
(Figure 2a; Oke et al., 2019). The southward-flowing current
is most coherent off 28°S (Sloyan et al., 2016) and intensi-
fies around 29°–31°S (Kerry and Roughan, 2020). The cur-
rent typically separates from the coast between 31°S and
32.5°S, turning eastward and shedding large warm-core
eddies in the Tasman Sea (Cetina Heredia et al., 2014). The
EAC is a relatively well-observed WBC system, with obser-
vations collected as part of Australia’s Integrated Marine
Observing System (IMOS; Figure 2b–d) spanning from the
coherent jet to the eddy field (e.g., Roughan et al., 2015).
The EAC therefore provides an ideal testbed for assessing
observation impact across differing dynamical regimes.
Observing networks, numerical models, and DA schemes
make up the key components of ocean prediction systems.
Data-assimilating models are useful for evaluating and
designing observing networks. Here we synthesize the
results from three different model-based approaches in
order to assess observation impact across a common sys-
tem (the EAC). We use three methods for studying observa-
tion impact: an adjoint-based approach to directly quantify
FIGURE 1. Conceptual schematic showing sequential time-dependent data assimilation and a summary of the three methods presented in this study
for assessing observation impact.