Impact of bio‐optical data assimilation on short‐term coupled physical, bio‐optical model predictions

Abstract

[1] Data assimilation experiments with the coupled physical, bio-optical model of Monterey Bay are presented. The objective of this study is to investigate whether the assimilation of satellite-derived bio-optical properties can improve the model predictions (phytoplankton population, chlorophyll) in a coastal ocean on time scales of 1-5 days. The Monterey Bay model consists of a physical model based on the Navy Coastal Ocean Model and a biochemical model which includes three nutrients, two phytoplankton groups (diatoms and small phytoplankton), two groups of zooplankton grazers, and two detrital pools. The Navy Coupled Ocean Data Assimilation system is used for the assimilation of physical observations. For the assimilation of bio-optical observations, we used reduced-order Kalman filter with a stationary forecast error covariance. The forecast error covariance is specified in the subspace of the multivariate (bio-optical, physical) empirical orthogonal functions estimated from a monthlong model run. With the assimilation of satellite-derived bio-optical properties (chlorophyll a or absorption due to phytoplankton), the model was able to reproduce intensity and tendencies in subsurface chlorophyll distributions observed at water sample locations in the Monterey Bay, CA. Data assimilation also improved agreement between the observed and model-predicted ratios between diatoms and small phytoplankton populations. Model runs with or without assimilation of satellite-derived bio-optical observations show underestimated values of nitrate as compared to the water sample observations. We found that an instantaneous update of nitrate based on statistical relations between temperature and nitrate corrected the model underestimation of the nitrate fields during the multivariate update.

Assimilation of remotely-sensed optical properties to improve marine biogeochemistry modelling

“…Where CV is low, a negative correlation skill indicates poor performance in simulating weekly fluctuations, rather than seasonal cycles. Skilled simulation of short-term fluctuations is still a challenging task for complex marine ecosystem models (see, e.g., Allen et al, 2007;Stow et al, 2009;Shulman et al, 2013).…”

“…However, the assimilation of satellite-derived optical data into ecosystem models is still in its infancy. Shulman et al (2013) demonstrated that the assimilation of phytoplankton absorption can improve the short-term predictions of chlorophyll and PFT ratios, in a five-day assimilative simulation in the Monterey Bay. However, to the authors' knowledge, the benefits of assimilating satellite-derived optical data for long-term biogeochemical simulations have yet to be tested.…”

Use of remote-sensing reflectance to constrain a data assimilating marine biogeochemical model of the Great Barrier Reef

“…Ciavatta et al, 2014) or phytoplankton absorption (e.g. Shulman et al, 2013). For complex coastal regions that are dominated by case 2 waters, an explicit spectrally resolved in-water optics model opens the possibility of directly assimilating RSRs and avoids the costly requirement of calibrating an empirical IOP algorithm that is regionally specific.…”

“…There are now examples of operational and pre-operational global systems that routinely assimilate Chl a products (Ford et al, 2012). Additionally, there has been further experimentation with assimilating alternative remotely sensed apparent optical properties (AOPs) such as the vertical attenuation coefficient at 443 nm, Kd 443 (Ciavatta et al, 2014) and inherent optical properties (IOPs) such as phytoplankton absorption (a ph ), as described in Shulman et al (2013). A map of the Great Barrier Reef region, with the colour bar denoting the water depth, markers denoting the population centres (red triangles), IMOS NRS sites (yellow triangles), GBRMPA MMP Water Quality Meters (WQMs; yellow circles) and points of interest referred to in the text (red circles), with the glider track (white line adjacent to Lizard Island).…”

Can we do better than the grid survey: Optimal synoptic surveys in presence of variable uncertainty and decorrelation scales

“…The extent of the domain was chosen to correspond to the location of the 2010 BIOSPACE field experiment [Shulman et al, 2013]. It is likely that the spatial and temporal scales in this domain are correlated.…”