Using AXBTs to improve the performance of coupled hurricane-ocean models

Hurricanes develop and are maintained by heat energy they receive from the sea surface. The warmer the sea surface temperature (SST) is below the hurricane, the more energy is available to the hurricane. Wind-induced mixing of the upper ocean by a hurricane can cool the sea surface via entrainment of cooler water into the oceanic mixed layer (OML) from below. Therefore, the future intensity (and perhaps track) of a given hurricane depends not only on the initial temperature of the sea surface below the hurricane, but also on the magnitude of the wind-induced sea surface cooling in the region providing heat energy to the hurricane. The magnitude of the wind-induced cooling depends on the magnitude of the surface wind stress, the depth of the OML, and the temperature gradient at the base of the OML.

Scientists at the Hurricane Research Division (HRD) measure ocean temperature profiles in the pre-hurricane environment and in the wind-induced cold wake via AXBT instruments dropped from aircraft. Such ocean temperature profiles can also be derived from the GFDL/URI coupled hurricane-ocean model (hereafter GFDL model), which has been run operationally at NCEP to forecast hurricane track and intensity since 2001. If initial and predicted temperature profiles in the GFDL model are inconsistent with AXBT observations, then steps should be taken to improve the GFDL model's representation of the 3D ocean temperature field.

The ocean component of the 2005 operational version of the GFDL model (hereafter OP05) is the Princeton Ocean Model (POM). Before the GFDL model is run to produce a hurricane forecast, POM is run twice to “spin-up” the ocean. During the first preliminary POM run (hereafter phase 1), which operationally forecasts out 48 hours, the GDEM monthly ocean climatology is assimilated with NCEP real-time SST, and ocean fronts are imposed as necessary. During the second preliminary POM run (hereafter phase 2), which operationally forecasts out 72 hours, the cold wake is created by assimilating the wind speed information from the hurricane message file (hereafter MSG) provided by TPC.

19 AXBTs were dropped in the Gulf of Mexico (GoM) on 15 September 2005. One of the AXBTs dropped at 27.895°N, 88.623°W yields a temperature profile that is consistent with the analogous OP05 phase 1 profile. A second AXBT dropped at 25.579°N, 87.174°W (hereafter A2) yields a temperature profile that is significantly warmer than the analogous OP05 profile. This difference is attributed to OP05's inability to accurately represent the spatial structure and temporal variation of the Loop Current (LC).

In research mode, a new type of feature-based modeling is being developed that represents the LC more accurately. With the new LC initialization, the GDEM climatology can be assimilated with real-time sea surface height data (SSH) in addition to SST. Based on AXBT A2, the SSH/SST-assimilated temperature profile is more accurate than the SST-assimilated temperature profile (i.e. the OP05 profile) at the location of A2.

The GDEM climatology used in OP05 has 1/2° grid spacing. Two other monthly ocean climatologies currently available are GDEM with 1/4° grid spacing and Levitus, which also has 1/4° grid spacing. Simulations with SSH/SST assimilation applied to these alternative climatologies do not yield temperature profiles that are more consistent with the available AXBTs than the SSH/SST-assimilated GDEM 1/2° simulation does.

24 AXBTs were dropped in the GoM during 22-23 September 2005. Some of these AXBTs were dropped in advance of Hurricane Rita, while others were dropped in Rita's wake. By utilizing one or more of the AXBTs dropped in Rita's wake, it is possible to assess the accuracy of the cold wake generated during phase 2. After assimilating both SSH and SST, two phase 2 simulations are performed: (1) using the MSG wind as in phase 2 of OP05, and (2) using the MSG wind with a new wind stress parameterization. Comparisons with an AXBT dropped in Rita's wake at 24.217°N, 87.459°W reveal that the OML temperature is more realistic using the original wind stress, but the upper thermocline temperature is more realistic using the new wind stress.

Research in progress includes AXBT comparisons with improved versions of the LC initialization and with simulations from other hurricanes. Eventually, AXBT profiles may be directly assimilated into future versions of the GFDL model and/or newer generation hurricane forecast models (e.g. Hurricane WRF) to obtain a more accurate initial ocean condition in real-time. By improving the initial ocean condition, hurricane-ocean model forecasts may improve as well.