Optimizing plankton survey strategies using Observing System Simulation Experiments

A major problem in biological oceanography is sparseness of data. Ocean observing systems are being developed to fill this need. Design of these systems can benefit from optimal sampling analysis applied to output of biological–physical models. This study describes a method for optimizing plankton surveys through the use of Observing System Simulation Experiments (OSSEs). Using a coupled physical–biological hindcast simulation for 1999 as “truth”, we applied the variance quadtree (VQT) algorithm to determine the locations of a fixed number of samples to reduce error. The optimized sampling process derived by the VQT algorithm is significantly better (at 95% confidence level) than simple random sampling. The relationship between root mean square error (RMSE) and number of samples allows one to balance the number of samples and the expected error, aiding the design of ocean observing systems. Compared with an existing sampling strategy used in the Gulf of Maine region, a fixed VQT-derived sampling strategy for Pseudocalanus alone can reduce expected errors by 20% on the annual average (range from 15% to 23%). While sampling for combination of two variables (adult Pseudocalanus and phytoplankton), the errors are modestly reduced by an annual average of 7% (range from −4% to 14%), suggesting that the ongoing operational observing strategy is close to optimal for multi-constituent sampling. The VQT-derived sampling stations are more densely (sparsely) spaced in areas having larger (lower) variance in the quantity of interest. Sampling strategies differ for Pseudocalanus and phytoplankton, reflecting differences in the statistics of their distributions. This kind of information can be used for directed sampling in the real ocean, which must grapple with multiple physical and biological properties that vary on different scales. This methodology can contribute to optimal sampling of biological–physical properties by ocean observing systems.