EMD-WRF On-Demand ICING

Example of a cold-climate wind farm.

Heated cup anemometer at iced mast.

Icing Map in windPRO with IEA Loss Percentage AEP as Legend.

Histograms during meteorological icing (example results from report).

Seasonal variation of modelled instrumental and meteorological icing (example results from report).

Introduction

EMD-WRF OD ICING is the name of EMD’s icing model; it was first released in September 2022 with the availability of windPRO 3.6. It is available as a time-series product, similar to the well known EMD-WRF OD service. The model is fully validated: A technical note with results from recent validation study on EMD-WRF OD ICING is available [14], see further below for the download link.

The EMD-WRF OD ICING model is configured with the following setup:

• Driven by an icing configuration of the standard EMD WRF [1] On-Demand service [2].
• Run with a spatial resolution of 3x3 km and an hourly temporal resolution.
• Using the ERA-5 reanalysis data from ECMWF as global boundary data [3], see Table 2 below.
• Microphysics from Thompson scheme is used for parameterization of the cloud physics and the MYJ scheme for the planetary boundary layer physics [4], [5].
• The median volume diameter (MVD) by [6] is used, with a constant droplet concentration (Nc) of (default) 100 cm-3 and the liquid water content (LWC) in kg/m3 [7].
• Atmospheric data feeds into the standard cylinder-based model [8], [9] including melting and shedding [10].
• The WRF grid point (latitude, longitude) closest to the at mast location or site location is used as a default.
• The grid point holds a certain elevation above sea level and icing is modelled as a default for 15 heights in the vertical direction above ground level (agl.).
• The modelled ice load (kg) is used to identify hours of instrumental icing based on the industry standard thresholds of 10 g [11]. And similar from the modelled ice accretion rate (g/h), hours of meteorological icing [12] is found using the threshold of 10 g/h [9].
• The modelled icing results (IEA loss/class and meteorological icing hours) are downscaled to the terrain elevation height of the Copernicus DEM’s [15] for making icing maps.

The final step of EMD’s modelling chain, is an estimate of the expected production loss of a site which is found by using the IEA Ice Classification system seen in Table 1 below. A wide range of climate parameters will be available from a model-run; the complete list of parameters are seen in the Table 2 further below.

Table 1: IEA Task 19 Ice Classes: Production Loss Estimate as a Percentage of the Annual Energy Production. From [12].

IEA Ice-Class	Meteorological Icing (% of year)	Instrumental Icing (% of year)	Production loss (% of AEP)
5	> 10.0	> 20.0	> 20.0
4	5.0 - 10.0	10.0 - 30.0	10.0 - 25.0
3	3.0 - 5.0	6.0 - 15.0	3.0 - 12.0
2	0.5 - 3.0	1.0 - 9.0	0.5 - 5.0
1	0.0 - 0.5	< 1.5	0.0 - 0.5

Usage Notes

• Interested in the performance of the EMD-WRF On-Demand ICING modelling chain and a validation study?
  A validation paper by Petersen et al is available. It was presented at the International Workshop on Atmospheric Icing of Structures 2022:
  On the Modelling Chain for Production Loss Assessment for Wind Turbines in Cold Climates. The paper is available as a pdf-file here.
• Doing EMD-WRF OD ICING evaluations in very complex terrain/orography?
  EMD-WRF OD ICING relies on a downscaling approach with coupled-modelling at both mesoscale (3km grid resolution) and microscale (30m grid resolution). However, when in complex terrain, these resolutions are not always fine enough to ensure sufficient modelling accuracy. If the differences between meso- and micro terrain elevations are large (some hundreds of meters), then caution is advised, and users should expect increased uncertainty in results. If you are an EMD-WRF OD ICING user, feel free to contact the modelling team through the windPRO support hotline for further guidance and recommendations (support@emd.dk).

What you get and how to order?

To get pointwise-timeseries data, you need meso-credits (two credits are worth one month of mesoscale icing data). Credits can be ordered here: http://www.emd.dk/windpro/online-ordering/

Please note, that a time period of 10 seasons will include the complete icing analysis, whereas shorter time periods include only raw timeseries. Complete icing analysis includes:

• Icing reports as pdfs at three hub-heights - 100m, 150m and 200m
  • Including predicted AEP loss
• Icing maps at three hub-heights - 100m, 150m and 200m
  • IEA ice class, IEA Ice loss (% AEP) and modelled meteorological icing (icing rate > 10 g/h)
• The possibility to use your icing results and timeseries directly in your windPRO project
• Monthly, yearly and seasonal icing analysis and bin-sector analysis as csv-files
• Timeseries of raw WRF data and modelled icing

windPRO 3.6 Sample File and Reports
If you are a windPRO 3.6 user and interested in seeing a sample icing-calculation with reports, data-sheets (excel) and downscaled maps, then we have the Stor Rotliden sample available for download:

• windPRO 3.6 export file - here.
  Just download, inspect the downscaled icing maps (result-layers) and find the icing pdf-reports within the meteo-object (see the .mesores tab).
• pdf-sample report - 100m hub height - here. (More heights in the windPRO project file)
• time series data - excel - 100m hub height -here (More heights in the windPRO project file)

Data Availability

All EMD-WRF OD ICING is available with global spatial coverage. The temporal availability and update frequency depends on a number of factors such as availability from the boundary data providers, bandwidth and download times, as well as availability on EMD high-performance computing and storage systems. EMD-WRF OD ICING is availability with the ERA5 only, see table below.

Table 2: Data Source for the EMD-WRF OD Icing Configuration

Dataset	First date	Most recent date
EMD-WRF OD (ERA5)	1999.01.01	7-10 days from end of last month

Set of Standard Dataset Parameters for EMD-WRF OD ICING

A large quantity of useful parameters are available directly in windPRO to aid in your analysis. The different parameters in each On-Demand ICING dataset is shown in the list below.
Unless other specification, x is the vertical heights of: 10m, 25m, 50m, 75m, 100m, 150m, 200m, 300m, 400m, 500m, 600m, 1000m.

Table 3: EMD-WRF On-Demand ICING dataset. Including raw WRF data and modelled icing parameters; iceInten.x, MIce.x, MeteoSignal.x, InstruSignal.x, MVD.x, LWC.x.

Parameter	Unit	Description	Type
time		UTC time stamp
psfc	Pa	Pressure at site	Instantaneous
msl	Pa	Pressure at mean sea level	Instantaneous
wSpeed.x	m/s	Wind speeds	Instantaneous
wDir.x	deg	Wind direction	Instantaneous
wSpeed.0-30mb	m/s	Wind speeds at pressure level 0-30mb.	Instantaneous
wDir.0-30mb	deg	Wind speeds at pressure levels 0-30mb.	Instantaneous
wSpeed.850hpa	m/s	Wind speeds at pressure level 850hPa.	Instantaneous
wDir.850hpa	deg	Wind speeds at pressure levels 850hPa.	Instantaneous
temperature.2	celcius	Temperatures at height 2m	Instantaneous
waterTemp	celcius	Water temperature	Instantaneous
soilTemp.0-10cm	celcius	The temperature in the upper 10cm of the soil	Instantaneous
relHumidity.2	%	Relative humidity in height 2m above ground level	Instantaneous
snowDepth	m	Snow depth (if present)	Instantaneous
vis.s	m	Visibility at surface	Instantaneous
sensHeatFlux.s	w/m2	Sensible Heat Flux at surface	Instantaneous
totPrecip.s	kg/m^2	Total Precipitation at surface	1h Accumulated
downShortWaveFlux.s	w/m^2	Downward shortwave irradiance at surface	1h Average
totalCloudCover.a	%	Total cloud cover in atmosphere	1h Average
convCloudCover.a	%	Convective cloud cover in atmosphere	1h Average
4LFTX	K	N/A
rmol	1/m	Inverse Monin-Obukhov-Length
znt	m	Rougnhess length	Instantaneous
u*	m/s	U-star (friction velocity)	Instantaneous
swdDir.s	W/m^2	Direct shortwave irradiance at surface	1h Average
swdDni.s	W/m^2	Direct normal shortwave irradiance at surface	1h Average
swdDif.s	W/m^2	Diffusive shortwave irradiance at surface	1h Average
sqrtTKE.x	m/s	Wind speed given as standard deviation in m/s. Derived from the turbulent kinetic energy (TKE)	Instantaneous
cloudWater.x	mg/kg	Cloud water content	Instantaneous
cloudIce.x	mg/kg	Cloud ice content	Instantaneous
press.x	Pa	Pressure	Instantaneous
temperature.x	celcius	Temperatures	Instantaneous
rh.x	%	Relative humidity	Instantaneous
vis.x	m	Visibility	Instantaneous
cloudBottom	m	Distance from ground level to the bottom of cloud	Instantaneous
cloudTop	m	Distance from ground level to the top of cloud	Instantaneous
waterTemp	celcius	Water temperature	Instantaneous
Ice model output parameters.
iceInten.x	g/h	Ice accretion rate (intensity) [8][9]	Instantaneous
MIce.x	kg	Ice load on a standard cylinder [13]	Instantaneous
MeteoSignal.x		Meteorological icing, iceInten.x > 10g/h [9]	Instantaneous
InstruSignal.x		Instrumental Icing, MIce.x > 10g [11]	Instantaneous
MVD.x	m	Median Volume Diameter [6]	Instantaneous
LWC.x	kg/m^3	Liquid Water Content [10]	Instantaneous

References

1. W. C. Skamarock, J. B. Klemp, J. G. D. O. Dudhia, D. M. Barker, W. Wang and J. G. Powers, “A description of the advanced research WRF version 2,” NCAR Technical Note, Boulder, Colorado, USA, 20005.
2. M. L. Thøgersen, “help.emd.dk,” EMD International A/S, 2019. [Online]. Available: https://help.emd.dk/mediawiki/index.php?title=EMD-WRF_On-Demand_and_Custom-Area. [Accessed 30 November 2021].
3. ECMWF, “Advancing global NWP through international collaboration,” ECMWF, [Online]. Available: https://www.ecmwf.int/. [Accessed 23 March 2022].
4. G. Thompson, P. R. Field, R. M. Rasmussen and W. D. Hall, “Explicit Forecasts of Winter Precipitation Using an Improved Bulk Microphysics Scheme. Part II: Implementation of a New Snow Parameterization,” American Meteorological Society, vol. 136, no. Monthly Weather review, pp. 5095-5115, 2008.
5. Z. I. Janjic, “Nonsingular Implementation of the Mellor-Yamada Level 2.5 Scheme in the NCEP Meso model,” National Centers for Environmental Prediction, Washington, 2001.
6. K. Finstad, E. Lozowski and L. Makkonen, “On the median volume diameter approximation for droplet collision efficiency,” Journal of Atmospheric Sciences, vol. 45, pp. 4008-4012, 1988.
7. G. Thompson, B. E. Nygaard, L. Makkonen and S. Dierer, “Using the Weather Research and Forecasting (WRF) model to predict ground/structural icing,” in International Workshop on Atmospheric Icing on Structures (IWAIS), 2009.
8. L. Makkonen, "Models for the Growth of Rime Glaze Icicles and Wet Snow on Structures," Royal Society, vol. 1776, no. Ice and Snow Accretion on Structures, pp. 2913 - 2939, 2000.
9. ISO, "DS/ISO 12494:2017 Atmospheric icing on structures," Danish Standard Association, København, 2017.
10. K. Harstveit, “Using Metar-data to Calculate In-cloud Icing on a Mountain Site Near by the airport,” in 13th International Workshop on Atmospheric Icing on Structures (IWAIS), Andermat, Switzerland, 2009.
11. K. Hämäläinen and S. Niemelä, “Production of a Numerical Icing Atlas for Finland,” Wind Energy, vol. 20, pp. 171-189, 2017.
12. I. Baring-Gould, R. Cattin, M. Durstewitz, M. Hulkkonen, A. Krenn, T. Laakso, A. Lacroix, E. Peltola, G. Ronsten, L. Tallhaug and T. Wallenius, "13 Wind Energy Projects in Cold Climate 1st edition," IEA Wind Task 19, 2011.
13. L. Makkonen, "Modelling of Ice Accretion on Wires," Climate Appl. Meteor., vol. 23, pp. 929-939, 1984.
14. M.C. Pedersen, T. Ahsbahs, W. Langreder, M.L. Thøgersen: On the Modelling Chain for Production Loss Assessment for Wind Turbines in Cold Climates, Proceedings – Int. Workshop on Atmospheric Icing of Structures, IWAIS 2022 - Montreal, Canada, June 19-23 (pdf-here)
15. M. L. Thøgersen, “help.emd.dk,” EMD International A/S, 2021. [Online]. Available: https://help.emd.dk/mediawiki/index.php?title=Copernicus_DEM. [Accessed 09 August 2022].