Stability classes for pollution dispersion modeling from reanalysis data

Atmospheric dispersion models used for regulatory purposes are commonly driven by single-point measurement timeseries. The fundamental input quantities are time of day, wind direction, speed and atmospheric stability. In this case, stability is usually expressed in terms of stability classes, as formulated by Pasquill and Gifford or Klug and Manier, rather than in terms of a continuous stability measure, such as Monin-Obukhov stability.

In the last decade, a number of high-resolution atmospheric reanalyses that provide hourly output fields, including ERA5 (31 km) and CERRA (5.5 km), have become available. It is therefore possible, at least in principle, to derive the input for pollution models from these reanalyses. Commercial environmental consulting institutes have presented attempts at this, albeit without disclosing the exact method.

And this choice is far from trivial since the number of possible procedures is large: Measurements from operational weather stations or semi-mobile weather stations ususally do not permit the direct calculation of atmospheric stability measures. Consequently, the stability is derived via various schemes that yield the aforementioned stability classes. Reanalyses, in contrast, yield both the quantities that would be measured by a weather station as well as stability measure, such as the Obukhov length. The latter can be used to either determine a stability class, or alternatively, it may be fed directly into the dispersion model.

The present study aims to compare the various methods used to derive stability classes within the two most relevant classifications, and to assess the levels of agreement, both with each other and with weather station data. The comparison was performed for locations in different regions, with varying terrain complexity. For two locations where long-term eddy-covariance measurements are performed, the Obukhov-length inferred via the stabilty classes is compared to the values measured in situ.