Abstract
The effective monitoring of dynamic marine phenomena, such as phytoplankton blooms, across multiple spatial and temporal scales remains challenging. However, emerging closed-loop observation systems which integrate adaptive, multi-platform sensors with operational ocean models offer substantial potential to enhance accuracy and responsiveness. An operational closed-loop state estimation system was developed and tested in near real-time during a two-month field campaign in Frohavet, Norway. This closed-loop system integrated an Ensemble Kalman Filter with a coupled physical–chemical–biological ocean model across nested domains from the North Atlantic shelf to local coastal regions. Observations from the agile CubeSat HYPSO-1 nano-satellite and an uncrewed surface vehicle (USV, AutoNaut) were assimilated, dynamically informing the USV’s navigation and demonstrating the feasibility of adaptive, multi-tiered monitoring. Incorporating HYPSO-1 chlorophyll-a observations improved phytoplankton estimates at regional scales, while assimilating USV-based chlorophyll-a data further refined the predictions locally. The campaign highlighted operational challenges, including communication delays, software constraints, persistent cloud coverage, and solar storms. Post-campaign analyses identified and mitigated ecosystem model biases related to silicate dynamics and fixed carbon:nitrogen:chlorophyll-a conversion factors, further improving the model accuracy. Addressing these limitations through greater automation, tighter integration, and robust contingency planning is critical to scaling future closed-loop ocean monitoring systems.