Parameterizing the seasonal–diurnal wind climate of Rome: Fiumicino and Ciampino

Abstract

This study uses the Offset Elliptical Normal (OEN) mixture model to parameterize the seasonal–diurnal wind climate of Rome, Italy, for engineering design purposes, and also to reveal the physical meteorology of this “mixed” wind climate. The joint probability density function (PDF) of the wind vectors in the zonal–meridional plane are represented by a mixture of disjoint bivariate‐Normal ellipses. Each OEN ellipse is tentatively attributed to a physical wind mechanism: sea/land breezes; transient coastal jets; katabatic flows; the migrating atmospheric tide; and (non‐diurnal) geostrophic winds. The model enables a fuzzy probability of each wind mechanism to be assigned to each observation, allowing the progression of the various diurnal cycles to be tracked through time. This seasonal–diurnal parametric model is directly applicable to wind energy resource estimation, pollution studies, pedestrian comfort, and heating and ventilation design. The model is also an essential starting point in methods to simulate realistic time series of wind speed and direction.

Automated Probabilistic Analysis and Parametric Modelling of the Seasonal-Diurnal Wind Vector

“…The original OEN mixture model is fully described in [9] and the later refinements and extensions in [10,11] and its application to whole-year data in [8]. Here, for the convenience of the reader, the representation of the model on the zonal-meridional plane is shown in Figure 5 and the defining equations are reproduced in Appendix A.…”

“…Step 8 implements the new fuzzy OEN procedures described in [10,11] (see §A.3) to remove the deterministic seasonal/diurnal variation of the location and scale parameters by a process of demodulation, so revealing the random perturbations and estimating their PDFs, autocovariances and spectra.…”

“…The principal difference between this implementation and the original in [9] is that the ellipses attributed to large scale weather (LSW) systems are no longer forced to be non-diurnal by averaging over all H. The improved fitting process no longer needs this precaution to be able to resolve these ellipses. In any case, [10] recognised that the relative frequencies of the LSW ellipses must vary diurnally to maintain Σf i = 1 in response to the varying diurnal components.…”

“…Some residual diurnal variation is evident in most of the cases but decays with lag, τ, indicating that this is likely to be a "leakage" error since any real diurnal cycle must show a constant amplitude. The fitted curves are the bandwidth limited pink noise (BLPN) model introduced in [10] and defined by Equation A.10.…”

“…The ability of OEN to parameterise seasonal-diurnal variation was first demonstrated in 2015 on the well-studied, mixed-mechanism wind climate of Adelaide [9] where it was demonstrated that each disjoint model component corresponded to the known physical mechanisms. It was then applied to the wind climate of Rome [10], using observations from the Fiumicino and Ciampino airports, and again showed correspondence between model components and mechanisms, and with the differences between sites matching the differences in topography. The method has since been refined and its application has been extended with additional analyses that were previously thought impractical [11], but which now allow the underlying predictions of the model to be validated.…”