Air-sea heat fluxes in climate and numerical weather models, and data products

Mapping fluxes: satellite gridded products

Satellites measure air-sea fluxes through a variety of sensors, including microwave radiometers, scatterometers, and altimeters. Microwave radiometers measure the microwave radiation emitted by the ocean and atmosphere, which can be used to estimate the surface temperature and humidity. Scatterometers measure the roughness of the ocean surface, which is related to the wind speed and direction. Altimeters measure the height of the ocean surface, which is related to the sea level and can be used to estimate the ocean currents.

Using satellite measurements to estimate air-sea fluxes requires the use of mathematical models that relate the satellite measurements to the fluxes of interest. Some key variables estimated from satellite measurements for air-sea fluxes are:

• Sea surface temperature (SST)
• Sea surface salinity (SSS)
• Sea surface height (SSH), the measure of the height of the sea surface relative to a reference level. It can be used to estimate ocean currents and the transport of heat and freshwater across the ocean surface.
• Wind speed and direction, satellite-based scatterometers and radiometers can measure the wind speed and direction.
• Sea surface roughness, the sea surface roughness is a measure of the small-scale waves and ripples that form on the ocean surface.

Examples of satellite gridded products

• J-OFURO3 (Japanese Ocean Flux Data Sets with Use of Remote Sensing Observations), generated by the Japan Meteorological Agency (JMA), provides global estimates of surface heat fluxes over the ocean using a combination of in situ and satellite observations.
• OAFLUX (Objectively Analyzed Air-Sea Fluxes) is a global air-sea flux dataset that combines satellite and in situ observations to estimate surface heat fluxes over the ocean.
• HOAPS (Hamburg Ocean Atmosphere Parameters and Fluxes from Satellite Data) offers global fields of precipitation, evaporation, and latent heat flux over the oceans, derived from satellite data.

Table 1. Examples of satellite gridded products

Biases in data products

There is still considerable uncertainty that exists between data products. This is due to the different methods used to estimate the fluxes, the different satellite and in situ observations used as input, and the different assumptions and parameterizations used in the flux algorithms. There is growing research investigating these biases (e.g. BPG17, THK+18, CGE+19).

Climate models

Climate models are always run for a historical period and (often) also for the future to get an idea of the climate change that we can expect. The latest iteration of climate model simulation from the Coupled Model Intercomparison Project is CMIP6. CMIP6 simulate air-sea fluxes at grid scales of ~50–100 km.

In the CMIP6 and other climate model simulation, the air-sea fluxes are calculated using bulk flux algorithms. These algorithms are based on the assumption that the fluxes are proportional to the difference between the air and sea surface temperatures and the wind speed. Therefore, these algorithms require inputs of:

• Surface properties (e.g., sea surface temperature, ocean currents).
• Atmospheric boundary layer conditions.
• Wind speed.

Challenges

Models that resolve ocean mesoscale eddies (~0.1 degree ocean grid) have much better representation of high eddy rich regions. The mesoscale eddies are important for the air-sea fluxes because they can transport heat and salt, and they can also affect the wind stress. However, these models are computationally expensive and not yet feasible for long-term climate simulations. The lower resolution (~1 degree ocean grid) climate models can’t resolve these eddies. Therefore, they underestimate the variability of the sea surface temperature and misrepresent the variability of the fluxes.

High-resolution models capture ocean-forced flux variability but not climate scale models. As we have seen, the dominant variability of turbulent heat fluxes comes from the western boundary currents and tropical Pacific, but models cannot represent the WBCs as they are narrow and fast ocean currents, often only 50–100 km wide. Because WBCs have very sharp sea surface temperature (SST) fronts, coarse climate models appear too broad, too weak, mislocated, and less variable than in reality (SBPT19).

These SST fronts drive strong atmospheric responses (e.g., storm tracks, precipitation bands). Coarse models smooth out these SST gradients, weakening the:

• Surface wind convergence/divergence patterns
• Evaporation and precipitation patterns
• Atmospheric pressure adjustments

Numerical weather prediction

Numerical Weather Prediction (NWP) models are used to forecast weather conditions. They are typically run at higher resolutions than climate models, allowing better representation of mesoscale oceanic features and tropical cyclone-ocean interactions, such as the Kuroshio and Gulf Stream.

Some key elements of numerical weather prediction models:

• High-resolution grids (e.g., 10-20 km).
• Mesoscale oceanic features.
• Tropical cyclone-ocean interactions.
• Coupled models (e.g., ocean-atmosphere models) are increasingly used to improve forecasts.

Challenges

NWP models cannot directly simulate turbulence at meter-scales near the ocean — so they parameterise these small-scale fluxes based on larger-scale variables, using well-tested physics. Data assimilation of SSTs and atmospheric observations is critical to bringing models closer to reality – reanalysis.

Numerical weather prediction models have known biases in the ocean surface fluxes, which can be attributed to:

• Biases in the SSTs used to force the models.
• Biases in the atmospheric model physics.
• Biases in the ocean model physics.

This has put large importance of observing in-situ air-sea fluxes. Some efforts include: