Bourassa-Vincent-Wood Flux and Sea State Model

The BVW (Bourassa-Vincent-Wood) model is a fully coupled flux and sea state model. It determines internally consistent fluxes (momentum, heat, and moisture), atmospheric stabilities, and sea states.

Current Version: 1998.5, Updated 2 Nov. 1998: Update History

Review of Relevant Concepts

Winds and Stresses

Energy Fluxes

Atmospheric Stability

Sea State

Description

The Bourassa-Vincent-Wood (hereafter BVW) coupled flux and sea state model (Bourassa et al. 1998) includes a host of new physical considerations that have been shown to be significant for a wide range of applications. These considerations include turbulent transport due to capillary waves (surface ripples), directional sea state (including traditional sea state parameters such as phase speed and wave age), wave-wave interaction between dominant waves and ripples, a capillary cutoff, and convection. The model differs from others in that the physical premises are applicable to low and moderate wind-speeds (U₁₀ < 7 m s^-1), as well as higher wind-speeds. Furthermore, for low wind speeds (U₁₀ < 2 m s^-1) the capillary cutoff distinguishes between aerodynamically smooth and rough surfaces.

The mean wind direction relative to the direction of wave propagation has been observed to result in large differences in drag coefficients (Geernaert 1988; Dobson et al. 1994; Rieder 1994; Donelan et al. 1997). The influence of the directional sea state are included in an anisotropic roughness length, which allows a cross wind component of the stress to be modeled. To date, BVW is the only model that predicts different stresses for waves propagating with the mean wind versus waves moving in any other direction. For light and moderate wind speeds over the open ocean, dependence on the directional aspects of sea state are likely to overwhelm dependency on wave age. The considerations of capillary waves and swell are critical to modeling fluxes in areas of low and moderate wind speeds, such as the tropics and the belts of sub-tropical high pressure systems. Most models are not accurate at low and moderate wind-speeds (U₁₀ < 7 m s^-1; approximately 50% of the global winds over water), and tend to underestimate fluxes in these regimes.

Most BVW improvements over previous models are incorporated in the momentum roughness-length parameterization. The roughness length is treated as root-mean-square sum of weighted contributions from gravity waves, capillary waves, and an aerodynamically smooth surface. In addition, the dimensionless constant in the relationship between the capillary wave component of momentum roughness length and friction velocity was re-evaluated using both wave tank data and field data (Bourassa et al. 1998). The new value was found to be 0.06, a factor of three smaller than the original estimate (Wu 1968, 1994) of 0.18. Modeling the influence of capillary waves has been shown to improve the accuracy of modeled surface fluxes and drag coefficients determined from observations. Several sets of observations (Bradley et al. 1991, 1993; Fairall et al. 1996a, Donelan et al. 1997; Dupuis et al. 1997) were used to examine mean increases in modeled fluxes due to capillary waves. The average changes in latent heat fluxes (based on three sets of tropical observations) were compared to modeled increases due to convection (sometimes called 'gustiness'; Godfrey and Beljaars 1991; Fairall et al. 1996a,b). The mean impact of capillary waves was found to be larger than that of convection by a factor of four. This relatively large impact of capillary waves is partially due to a reduction in the impact of convection when capillary waves dominate the surface stress. For U₁₀ < 7 m s^-1, capillary waves have been found to increase the mean modeled tropical fluxes of momentum and latent heat by 0.004 N m^-2 and 6 W m^-2.

The use of a simple sea state model (considering swell at one frequency and wind driven waves, with the direction of the swell not required to match the wind direction) made the model much more computationally efficient than spectral wave models, and allowed complete physical compatibility with the flux model. One of a variety of sea state observations (phase speed or period of the dominant waves, wave age, significant wave height, or significant slope) can be combined with a modeled or observed wind forcing parameter (wind velocity, pseudostress, friction velocity, stress, or equivalent neutral winds [observed by satellite based scatterometers]) to determine sea state dependent fluxes (momentum, sensible heat, and latent heat). Details of the model input and output can be found on a web site: http://www.coaps.fsu.edu/~bourassa/ BVW_html/bvw_docs.html. One aspect of the sea state model is the HEXOS dependency of the gravity wave component of roughness length on sea state (Smith et al. 1992). There are several questionable aspects of this parameterization (Yelland et al. 1998; Bourassa et al. 1998) which reduce model accuracy for high wind speeds (for U₁₀ > 10 or 12 m s^-1). These errors can be mitigated by choosing appropriate sea state input for these conditions, such as an equivalence to neutral drag coefficients based on observations (Smith 1988; Anderson 1993; Dobson et al. 1994).

Another aspect of the sea state model is wave-wave interaction between the orbital velocity of the dominant waves and the 'velocity frame of reference' of the capillary waves. Roughness length in the log-wind equation is exponentially dependent on the velocity frame of reference, hence it is critical that each roughness length component (gravity waves, capillary waves, and smooth surfaces) be in the same frame of reference. For U₁₀ < 7 m s^-1, this wave-wave interaction component of the sea state model has much more influence than the HEXOS contribution to sea state.

The BVW and Clayson, Fairall, and Curry (Clayson et al. 1996) flux models also differ slightly from most previous models in the choice of atmospheric stability parameterizations. They use Benoit's (1977) parameterizations for an unstable atmosphere, which is a slightly more detailed solution than the usual Businger-Dyer parameterizations (Dyer 1974). For a stable boundary layer, the Beljaars and Holtslag (1991) parameterization is applied. These changes result in systematic changes in fluxes: reduced flux magnitudes due to relatively stable atmospheric stratification. The differences are small for near neutral stratification, and increase with increasing departures from neutral conditions. The increases for stable air are usually small (<5 W m^-2); however, they can be much larger for extremely stable air. The changes for unstable stratifications are larger (<10 W m^-2), and can exceed 25 W m^-2 for extreme stratifications. The Monin-Obukhov scale length is calculated as described by Liu et al. (1979). An advantage of the Liu et al. parameterization over traditional Monin-Obukhov parameterizations is the consideration of moisture stratification, which can dominate atmospheric stability considerations for relatively dry air over warm waters (Smith 1988). The increases in flux magnitudes due to these improved stability parameterizations usually >0.001 N m^-2 and >1 W m^-2, however, they can be greater for extreme stability conditions.

Examples of calls to these routines are given for C, FORTRAN, and IDL in the links for these compilers in the section 'Required Files and Examples?' Examples of variable declarations are also provided.

pmix:

ht_adj:

find_q:

There are many methods of measuring the moisture content of air. When data sets are combined, it is likely that several methods have been used. In most boundary layer applications, these measurements must be converted to specific humidity. This routine converts all the common humidity measurements to specific humidity.

Routines have been developed to test the installation of pmix. Tests can be made on C, FORTRAN, and IDL versions of the code. Note that each of these versions calls the C routines: mixed language compilers are required. The code is designed for SGIs, for which the same source code can easily be used produce C and FORTRAN routines. The test programs are ctest_bvw98.c, ctest_ht_adj98.c, ftest_bvw98.f, ftest_ht_adj98.f, test_bvw98.pro, and test_ht_adj98.pro. Input data and the expected output are given in testdata98.dat.

These files can be downloaded through this web site or taken from our public FTP site (anonymous FTP to coaps.fsu.edu, change directories to pub/bourassa/bvw98). The linked pages (below) provided examples of variable declarations and calls to the routines.

C users: bvw98.c, bvw98.h, ctest_bvw98.c, ctest_ht_adj98.c, testdata98.dat, Makefile (for SGIs)

FORTRAN users: bvw98.c, bvw98.h, ftest_bvw98.f, ftest_ht_adj98.f, testdata98.dat, Makefile (for SGIs)