Environmental Engineering Reference
In-Depth Information
The most rigorous procedure is to derive empirical relationships by fitting variables
and functions to the “training” data and then determine the error with respect to the
“validation” data that have been withheld. Many sites lack sufficient data with which
to conduct such a validation. In such cases where the validation data must be included
in the training data set, there is a tendency to underestimate the errors.
Nevertheless, statistical models are a valid approach when proper procedures are
followed. Statistical methods can also be combined with other approaches, such as
numerical wind flow models. A good example of this is the ruggedness index (RIX)
correction that is sometimes used with the Wind Atlas Statistical Package (WAsP)
model (described below). RIX is a parameter that has been found through statistical
modeling to be a good predictor of WAsP errors in some circumstances (1).
13.1.4 Numerical Wind Flow Models
The most popular methods of spatial modeling rely mainly on numerical wind flow
models. There are numerous wind flow models in use by the wind industry today,
which are based on a variety of theoretical approaches. All models attempt to solve at
least some of the physical equations governing motions of the atmosphere, with varying
degrees of complexity. The models fall into four general categories: mass-consistent,
Jackson-Hunt, computational fluid dynamics (CFD), and mesoscale NWP models.
Mass-Consistent Models. The first generation of wind flow models devel-
oped in the 1970s and 1980s (e.g., NOABL (2), MINERVE) were mass-consistent
models, so called because they solve just one of the physical equations of motion,
that governing mass conservation. When applied to the atmosphere (assuming it is
incompressible, a good assumption within the boundary layer), the principle of mass
conservation implies that wind forced over higher terrain must accelerate so that the
same volume of air passes through the region in a given time. As a result, these models
predict stronger winds on hilltops and ridgetops and weaker winds in valleys. They
cannot handle thermally driven wind patterns, such as sea breezes and mountain-valley
circulations, and flow separations on the lee side of hills or mountains.
The solution offered by a mass-conserving model is not unique: the governing
equation actually permits an infinite number of solutions. Instead, most models are
designed to depart by the smallest possible amount from an initial wind field “guess”
derived from observations or another model (e.g., a NWP model run at a coarser
resolution). Such a characteristic sets this type of model somewhat apart from other
numerical models, which make no such assumption. It also means that mass-consistent
models are able to take advantage of data from additional meteorological towers in a
natural way, by modifying the initial guess.
Jackson-Hunt Models. The next generation of models (e.g., WAsP (3, 4), MS-
Micro or MS3DJH (5, 6), Raptor (7), Raptor NL (8)) were originally developed in the
1980s and 1990s based on a theory advanced by Jackson and Hunt (9). They go beyond
mass conservation to include momentum conservation by solving a linearized form of
the Navier-Stokes equations governing fluid flow. The most important simplification
Search WWH ::




Custom Search