WCD-07. Using a Linear Inverse Model to predict Ocean Biogeochemistry

Abstract
Oceanic biogeochemical quantities such as Chlorophyll, Net Primary Production (NPP), Oxygen, CO2, among others, are fundamental to the health of the oceans and the dynamics of marine ecosystem, making the ability to understand and predict their evolution in response to climate variability and change a key priority of high societal relevance. Forecast systems based on global climate models that include ocean biogeochemistry are becoming available, but they are very expensive to run, complex to understand, and their results may be hindered by climate model biases and limitations. Linear Inverse Models (LIMs) can provide a complementary approach to the above forecast systems. They are inexpensive to run, simpler to understand, and, being derived from observations, are free of the biases that may hamper the dynamical climate models. However, while successful in modeling and predicting physical quantities, LIMs have not yet been used for the prediction of ocean biogeochemistry, whose evolution may follow nonlinear trajectories, and be hard to represent with a linear approach. In this study, we use a LIM that is based on sea surface temperature, sea surface height and NPP information to explore the ability of the LIM methodology to “simulate” NPP, and then use this LIM to examine the potential predictability of NPP in the tropical Pacific Ocean. The LIM methodology allows us to detect the growth of the NPP anomalies is association with the development of ENSO events. The largest NPP anomalies are found in the central-western Pacific, near the ocean front that separates the western Pacific “Warm Pool” from the eastern Pacific “Cold Tongue”. Cross-validated anomaly correlations show statistically significant skill for NPP prediction at lead times up to 24 months, exceeding the SST prediction skill at both 12 and 24 months lead times.