Optimization of a micronekton model with acoustic data

In the pelagic foodweb, micronekton at the mid-trophic level (MTL) are one of the lesser known components of the ocean ecosystem despite being a major driver of the spatial dynamics of their predators, of which many are exploited species (e.g. tunas). The Spatial Ecosystem and Population Dynamics Model is one modelling approach that includes a representation of the spatial dynamics of several epi- and mesopelagic MTL functional groups. The dynamics of these groups are driven by physical (temperature and currents) and biogeochemical (primary production, euphotic depth) variables. A key issue to address is the parameterization of the energy transfer from the primary production to these functional groups. We present a method using in situ acoustic data to estimate the parameters with a maximum likelihood estimation approach. A series of twin experiments conducted to test the behaviour of the model suggested that in the ideal case, that is, with an environmental forcing perfectly simulated and biomass estimates directly correlated with the acoustic signal, a minimum of 200 observations over several time steps at the resolution of the model is needed to estimate the parameter values with a minimum error. A transect of acoustic backscatter at 38 kHz collected during scientific cruises north of Hawaii allowed a first illustration of the approach with actual data. A discussion followed regarding the various sources of uncertainties associated with the use of acoustic data in micronekton biomass.