4.5: Calculations of Wind

Satellites, ships, and buoys measure winds at various locations and times of the day. If you wish to use the observations to calculate monthly averaged winds over the sea, then the observations can be averaged and gridded. If you wish to use wind data in numerical models of the ocean’s currents, then the data will be less useful. You are faced with a very common problem: How to take all observations made in a six-hour period and determine the winds over the ocean on a fixed grid?

One source of gridded winds over the ocean is the surface analysis calculated by numerical weather models. The strategy used to produce the six-hourly gridded winds is called sequential estimation techniques or data assimilation. “Measurements are used to prepare initial conditions for the model, which is then integrated forward in time until further measurements are available. The model is thereupon re-initialized” (Bennett, 1992: 67). The initial condition is called the analysis.

Usually, all available measurements are used in the analysis, including observations from weather stations on land, pressure and temperature reported by ships and buoys, winds from scatterometers in space, and data from meteorological satellites. The model interpolates the measurements to produce an analysis consistent with previous and present observations. Daley (1991) describes the techniques in considerable detail.

Surface Analysis from Numerical Weather Models

Perhaps the most widely used weather model is that run by the European Centre for Mediumrange Weather Forecasts ecmwf. It calculates a surface analysis, including surface winds and heat fluxes (see Chapter 5) every six hours on a 1\(^{\circ} \times\) 1\(^{\circ}\) grid from an explicit boundary-layer model. Calculated values are archived on a 2.5\(^{\circ}\) grid. Thus the wind maps from the numerical weather models lack the detail seen in maps from scatterometer data, which have a 1/4\(^{\circ}\) grid.

ECMWF calculations of winds have relatively good accuracy. Freilich and Dunbar (1999) estimated that the accuracy for wind speed at 10 meters is ±1.5 m/s, and ±18\(^{\circ}\) for direction.

Accuracy in the southern hemisphere is probably as good as in the northern hemisphere because continents do not disrupt the flow as much as in the northern hemisphere, and because scatterometers give accurate positions of storms and fronts over the ocean.

The NOAA National Centers for Environmental Prediction and the US Navy also produce global analyses and forecasts every six hours.

Reanalyzed Data from Numerical Weather Models

Surface analyses of weather over some regions have been produced for more than a hundred years, and over the whole Earth since about 1950. Surface analyses calculated by numerical models of the atmospheric circulation have been available for decades. Throughout this period, the methods for calculating surface analyses have constantly changed as meteorologists worked to make ever more accurate forecasts. Fluxes calculated from the analyses are therefore not consistent in time. The changes can be larger than the interannual variability of the fluxes (White, 1996). To minimize this problem, meteorological agencies have taken all archived weather data and reanalyzed them using the best numerical models to produce a uniform, internally-consistent, surface analysis.

The reanalyzed data are used to study oceanic and atmospheric processes in the past. Surface analyses issued every six hours from weather agencies are used only for problems that require up-to-date information. For example, if you are designing an offshore structure, you will probably use decades of reanalyzed data. If you are operating an offshore structure, you will watch the surface analysis and forecasts put out every six hours by meteorological agencies.

Sources of Reanalyzed Data

Reanalyzed surface flux data are available from national meteorological centers operating numerical weather prediction models.

1. The U.S. National Centers for Environmental Predictions, working with the National Center for Atmospheric Research have produced the NCEP/NCAR reanalysis based on 51 years of weather data from 1948 to 2005 using the 25 January 1995 version of their forecast model. The reanalysis period is being extended forward to include all dateS up to the present with about a three-day delay in producing data sets. The reanalysis uses surface and ship observations plus sounder data from satellites. Reanalysis products are available every six hours on a T62 grid having 192 × 94 grid points with a spatial resolution of 209 km and with 28 vertical levels. Important subsets of the reanalysis, including surface fluxes, are available on CD-ROM (Kalnay et al. 1996; Kistler et al. 2000).
2. The European Centre for Medium-range Weather Forecasts, ECMWF, has reanalyzed 45 years of weather data from September 1957 to August 2002 (ERA-40) using their forecast model of 2001 (Uppala et al. 2005). The reanalysis uses mostly the same surface and ship data used by the NCEP/NCAR reanalysis plus data from the ERS-1 and ERS-2 satellites and SSM/I. The ERA-40 full-resolution products are available every six hours on a N80 grid having 160 × 320 grid points with a spatial resolution of 1.125\(^{\circ}\) and with 60 vertical levels. The ERA-40 basic-resolution products are available every six hours with a spatial resolution of 2.5\(^{\circ}\) and with 23 vertical levels. The reanalysis includes an ocean-wave model that calculates ocean wave heights and wave spectra every six hours on a 1.5\(^{\circ}\) grid.