OSSE impact analysis of airborne ocean surveys for improving upper-ocean dynamical and thermodynamical forecasts in the Gulf of Mexico

摘要

A prototype, rigorously validated ocean Observing System Simulation Experiment (OSSE) system is used to evaluate the impact of different sampling strategies for rapid-response airborne ocean profile surveys in the eastern interior Gulf of Mexico. Impacts are assessed with respect to improving ocean analyses, and forecasts initialized from those analyses, for two applications: improving oil spill forecasts and improving the ocean model response to tropical cyclone (TC) forcing. Rapid model error growth in this region requires that repeat surveys be conducted frequently in time, with separation of less than 4 days required to approach maximum error reduction in model analyses. Substantial additional error reduction in model dynamical fields is achieved by deploying deep (1000 m) AXCTDs instead of shallow (400 m) AXBTs. Shallow AXBTs constrain the ocean thermal field over the upper 400 m nearly as well as deep AXCTDs. However, in addition to constraining ocean fields over a greater depth range, AXCTDs also measure salinity profiles and more accurately constrain upper-ocean density than AXBTs, leading to a more accurate representation of upper ocean pressure and velocity fields. Sampling AXCTD profiles over a one-half degree array compared to one degree leads to substantial additional error reduction by constraining variability with horizontal scales too small to be corrected by satellite altimetry assimilation. A 2-day lag in availability of airborne profiles does not increase errors in dynamical ocean fields, but it does increase errors in upper-ocean thermal field including Tropical Cyclone Heat Potential (TCHP), demonstrating that these profiles must be rapidly made available for assimilation to improve TC forecasts. The additional error reduction in ocean analyses achieved by assimilation of airborne surveys translates into significantly improved forecasts persisting over time intervals ranging between 1 and 2 weeks for most model variables but several weeks for TCHP. In particular, upper-ocean temperature forecasts can be significantly improved for an extended interval of time by conducting airborne profile surveys.