Emanuel Coelho

Combining satellite ocean color and hydrodynamic model uncertainties in bio-optical forecasts

Abstract We describe an approach to produce short-term (1- to 3-day) forecasts of bio-optical properties by coupling moderate-resolution imaging spectroradiometer satellite (MODIS) ocean color imagery with a hydrodynamic model. The bio-optical property (chlorophyll in this case) is treated as a conservative tracer; the satellite distribution is advected forward in time using the current field from the hydrodynamic model. Uncertainties in both the satellite chlorophyll values and the currents from the circulation model impact the final forecast; we apply ensemble techniques to quantify the errors separately and in combination. For the ocean color imagery, we further apply ensemble techniques to partition the chlorophyll uncertainties into components due to atmospheric correction and bio-optical inversion, by applying noise to the near-infrared and visible band sets separately. The standard deviation for each ensemble suite provides an indication of uncertainty, or confidence in the satellite chlorophyll values and the hydrodynamic model current fields. By combining the two ensemble sets, we produce a final chlorophyll forecast field and associated uncertainty map that include both sets of uncertainties. We examine mean and individual forecast ensemble members (spread-skill statistics, RMS differences) to assess predictive value. This work represents a significant advancement in representing errors associated with satellite ocean color imagery and bio-optical forecasts.