Ana səhifə

Icing modeling

Yüklə 16.33 Kb.
ölçüsü16.33 Kb.

Icing modeling

The calculation of atmospheric icing calculation is made in two steps. (i) High-resolution simulations with an atmospheric model (WRF) are used to reproduce the necessary WRF modeling where meteorological data, i.e. of temperature, wind speed and cloud/precipitation particles are derived by simulation. (ii) Modeling of ice accretion using results of the atmospheric WRF simulation as an input to an ice accretion model.

Simulated atmospheric dataWRF modeling

The atmospheric datasets employed in this study are created by dynamically downscaling the ECMWF analysis with the non-hydrostatic mesoscale Advanced Research WRF-model (versions 3.2.1 and 3.0.1version?) (WRFARW, Skamarock et al. 2005), which is a state of the art numerical atmospheric model, used worldwide for operational simulation of weather and in atmospheric research, e.g. of atmospheric icing. With the dynamical downscaling to high horizontal resolution the orography is represented more correctly than at coarser resolutions, e.g. as is typical in the ECMWF-analysis, and hence weather in complex terrain is better reproduced. Furthermore, the relevant microphysics parameterization scheme used is the Thompson bulk scheme (Thompson et al. 2004,2008) which explicitly predicts cloud and precipitation variables used by the icing model. Namely, the mass of cloud water and ice, as well as rain, snow and graupel. As direct measurements are not available, the scheme is using a well-tested and fixed droplet number of 100 droplets/cm3 which may be an overestimate for environments which are mostly maritime.


Gregory Thompson, Roy M. Rasmussen and Kevin Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme, Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132, 519-542.

Gregory Thompson, Paul R. Field, Roy M. Rasmussen and William D. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme, Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 5095-5115.

The Thompson scheme for microphysics parameterization [] is used in the simulations. The cloud/precipitation parameters explicitly predicted by the model are: mass of cloud water, cloud ice, rain, snow, and graupel.

The most extensive main dataset is based on 15 years (1994-2009) of downscaled weather in Iceland at a horizontal resolution of 9 and 3 km, with 50 layers in the vertical. This dataset is a part of the RÁV-project (Rögnvaldsson & Ólafsson (2009)), and is the most detailed high-resolution data-set available to date, describing the weather and climate in the complex terrain of Iceland. This setup is nearly identical to that used in operational numerical simulations by Reiknistofa í veðurfræði (RV) which are used for forecasting of weather at Veðurstofa Íslands ("", the HRAS-system).

In order to better assess the effect of increased horizontal resolution on how well the downscaled weather is reproduced in the steep mountains of East-Iceland, several shorter periods of observed intense icing events were additionally downscaled to a resolution of 1 km and 333 m during, i.e. Dec. 2000 and 1 km during Nov.-Dec. 2006. Furthermore, as a limited number of surface observations are used to correct the atmospheric analysis forcing the dynamical downscaling, a sensitivity test was made where additional surface observations from Egilsstaðir were used as an additional forcing. For the periods downscaled to a reslution of 1 km downscaled to a resolution of 1 km in East-Iceland, another data-set was created where the WRF-model was forced with hourly observations of wind, temperature, pressure and humidity at Egilsstaðir. This forcing is strongest at Egilsstaðir and at the time of each observation, while the forcingits effect decreases quite fast both spatially and temporally with increasing distance and time from the observations. The additional forcing improves the model results for a region near Egilsstaðir and a considerable distance downstream of Egilsstaðir, i.e. in the direction of Hallormsstaðaháls. The model could in theory be forced with observations from other automatic weather stations but here we choose to limit us to only observations from Egilsstaðir. The representation of the orography improves significantly as the resolution improves from 9 km to 3 km to 1 km. The orography near the observation site at Hallormsstaðaháls is reasonably well reproduced at 1 km and the model altitude is approx. 550 m while the true altitude is 575 m. The error is slightly smaller at 333 m model resolution but far worse at 9 and 3 km.

Segja hvernig landlíkan passer við raunveruleika.

Four Different WRF simulations are used in this study and the differences in their setup are, they are givengiven in Table 1. All of the simulations are made with the assumption of droplet number as 100 droplets/cm3.

Table : WRF simulations.


Grid spacing [km]

Forced through observation






Yes, through Egilsstadir







Modeling ice accretion

The time dependent cylindrical ice accretion model described in ISO 12494-2000 is used to predict icing (Makkonen model). It describes how particles that can be either liquid (usually super cooled), solid, or a mixture of water, accumulate to objects. The model gives icing rate described by:

dMdt=α1∙α2∙α3∙w∙V∙A (1)

where V is particle velocity (m/s), A is the cross-sectional area (m2) perpendicular to object, w is water content (kg/m3), 1 is the collision efficiency, 2 is the sticking efficiency and 3 is the accretion efficiency.

Equation (1) is general but the parameters depend on assumptions on icing shape and thus the rotational stiffness of the object. The test spans have conductor with a specific direction and some restriction of rotation at ends. It is assumed that the accretion shape is cylindrical and it is taken into account that the measuring span has specific horizontal direction and does not collect icing as vertical cylinder approach that is often used.

The water phase in the WRF model is classified into: cloud water (Qc), cloud ice (Qi), rain (Qr), snow (Qs) and graupel (Qg). The first three of the phases can lead to icing accumulation but particles in form of ice and graupel are assumed to give no ice accumulation.

Cloud water is treated as the source of in-cloud icing: V is taken as wind velocity, w it taken as liquid cloud water, 1 is described by an empiric formulation as given in [XX]. A fixed droplet number (Nc) 70 droplets/cm3 is used as a base case, it is selected since typical values of Nc are often 75 – 100 droplets/cm3 for maritime air masses. Medium volume droplet is derived using formulas in [Thompson, Bj¢rn Egil etc.]. 2 is taken as 1.0 and 3 is given by formulas in [XX]. Density of in-cloud icing is given by formulas in [xx].

Rain is treated as freezing drizzle/rain when the temperature is below zero. The horizontal wind velocity is taken as V and vertical velocity is taken as 3 m/s. w is taken as Qr, 13 is taken as 0.9, 2 is taken as 1.0.

The snow is treated as wet snow icing when the temperature is between 0 to 2°. The horizontal wind velocity is taken as V and vertical velocity is taken as 1 m/s. w is taken as Qs, 1 is taken as 1. 2 is taken as __ if temperature is in range of 0°C to 2°C, otherwise it is taken as 0. 3 is taken as __. Density of wet snow is assumed to be 700 kg/m3. Nokkur óvissa er með slydduísingar líkanið, sérstaklega a2. Þessi dæmi hafa þó litla slydduísingu svo það skiptir litlu máli.

Ice shedding is not included in this version of the icing model.

Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur © 2016
rəhbərliyinə müraciət