Vortex uses a supercomputer cluster to run a non-linear flow model (WRF) that scales large atmospheric patterns (NCAR-NCEP, ECMWF and NASA) down to fine spatial resolutions (SRTM), generating modelled wind resource data suitable to be used where and when no measurements are as yet available.
Current wind modellers on the market offer either pre-calculated products or specific studies on a consultancy basis. Pre-calculated modelled wind resource data need to cover the whole world and are therefore simpler than Vortex ones: lower resolutions, linear flow results, etc. Instead, Vortex has automated a full non-linear modelling chain down to the microscale. Each calculation is performed on-demand, only on the client’s area of interest and without a need of a slow human interaction. Vortex is initialised by users, allowing direct interaction between the clients PC and our computer cluster over the Internet. This approach is much faster (and its price more competitive) than consultancy and much more powerful than pre-calculated results.
Because the calculations are non-linear, and non-linear computing is not immediate. Pre-calculating the whole world at high resolution using a non-linear model would require a huge amount of computing power and would also result in thousands of non-interesting areas being calculated. As a beneficial side effect, running each wind data simulation on-demand allows Vortex to incorporate cutting edge technology or lessons from a former case through an on-going learning process.
Vortex’s system core, WRF, is a sophisticated code that has been effectively employed to describe the physics and dynamics of atmospheric circulations with a significant degree of realism at a wide range of scales. WRF is neither a microscale nor large-scale model but is a one-in-all. WRF accuracy is based on the ability to portray the different mechanisms that interact at each one of the relevant atmospheric scales. The WRF model is the result of years of development by the atmospheric research community combined with additional experience from the extended use for weather forecast applications. The usual truncation, or simplification, of the equations that control atmospheric movements of air masses, is definitively small in WRF when compared to other lighter atmospheric codes. However, equally as relevant as this non-linear approach, is how the model includes other relevant factors such as radiation, thermal effects, air-sea-land interactions. In this sense, WRF is a modular model that can be adapted for different applications depending on the scale of the atmospheric movements, surface boundaries and thermal characteristic of the air masses, etc. In Vortex’s implementation of the WRF code in our automated modelling chain, it is especially relevant the use of WRF-LES (Large Eddy Simulation) module in the, up to our knowledge, unique product in the market Vortex LES.
You may find dozens of validation studies in our Knowledge Center. Current Vortex system accuracy values are presented in different studies done in sites around the world. These values change periodically thanks to the feedback of our users, who compare Vortex simulation results against their good-quality data. Our approach is risky but honest. Vortex is currently the only company on the market that allows for a systematical comparison of our model results against your measurements for you to gain confidence in the reliability of our simulations first hand.
No, it’s not mandatory. However, there are some Vortex products, such as SERIES, FARM, LES or EXTREMES, which results can be enhanced by calibrating with measurements.
Yes, Vortex SERIES provide Mesoscale reference time series: 10, 20 or 30 years of hourly data at any location worldwide. Before purchasing them, Vortex offers the possibility of downloading a free 6 months sample for you to check that the modelled series correlate with your measurements.
Since SERIES (like all Vortex products) are computed on demand, their calculation domain is centred on your selected point of interest so “distance to nearest calculation node” is always 0.
Vortex products are delivered at any height between 50 and 300m a.g.l.
No, sorry. One of the Vortex’s main aims is to provide our customers with fast results. This is only possible by automating the calculation process. Accepting your own surface data as an input for the simulation would imply a certain degree of manual work (revisions, corrections, etc.) that would slow down the whole process. If you have finer resolution input data (and time to process it) our suggestion is that you locally run your own Microscale simulations with, for example, WAsP or WindSim.
Vortex results have been compared fairly successfully against user measurements taken inside or near a forest after certain precautions have been taken: Vortex land use data comes from satellite information. Such information can distinguish between different types of trees but says nothing about the trees’ height. Depending on the forest, it is advisable to transport the calculation height down to take into account the canopy of the forest. For example, if trees are 15m high, measurements at 60m will better match model results at 45m. This kind of corrections must be performed by the client when information about the forest is available.
For microscale applications, the WRF model poses a CFD-like algorithm based on the Large Eddy Simulation (LES) approach. When WRF runs in LES mode the result is commonly known as WRF-LES model. This means that the simulation is run as usual but the turbulence parameterisation is replaced by the LES model and hence turbulent eddies are explicitly resolved. Vortex has improved the source code for its implementation for real simulations. This modified version of the model is the so-called Vortex-LES.
LES can mitigate uncertainty in offshore project wind resource assessment analysis by placing a virtual mast at any position on the wind turbines layout. Moreover, and in the context of the European offshore development, LES outstanding accurate representation of neutral stability conditions ensure more realistic estimation of vertical profiles in sites across North and Baltic seas.
Vortex FARM remodelling is based on a calibration process dependent on time. This can be done thanks to the fact that model WRF generates output information which is time-dependent. In the calibration process, we are able to obtain a calibrated time series at each grid node of the model. Once obtained, the adjusted time series are collapsed/averaged into commonly used file formats such as WRG or Tiff files which are not time-dependent and are dramatically size reduced when compared to the original calibrated time series. After the files have been delivered, we perform maintenance tasks that delete both the original and calibrated time series. This is a necessary process in order to avoid a massive storage system that would add a not affordable cost to our hardware structure. It’s much more sustainable and efficient to re-run the case with new observations if necessary.
Vortex ERA5 SERIES is now simulated within 10 days of real-time thanks to the preliminary daily updates of ERA5T reanalysis dataset provided by the ECMWF. As detailed in the documentation. ERA5T dataset will automatically become ERA5 after 3 months if no modifications are performed on the dataset. Potential differences between ERA5T and ERA5 might be due to later assimilation of meteorological observations that are not available with only a few days delay and might require a slower data filtering procedure resulting in a change in ERA5 dataset compared to the preliminary ERA5T dataset. In order to guarantee full consistency in our WRF-ERA5 SERIES, Vortex uses ERA5 final dataset if possible and ERA5T for the last 3 months. As a mode of example, see the case below: -Product Request: Vortex ERA5 20 year -Date of Request: 10 April 2020 -Vortex will produce hourly values based on ERA5 final dataset from 2000-01-01 up to 2019-12-31 -Vortex will produce hourly values based on ERA5T preliminary dataset from 2020-01-01 up to 2020-03-31 For potential updates of the mentioned 20-year dataset above, see below how the system will perform the extension: -Date of Update: 10 May 2020 -Vortex will produce hourly values based on ERA5 final dataset from 2020-01-01 up to 2020-01-31 -Vortex will product hourly values based on ERA5T preliminary dataset from 2020-02-01 up to 2020-04-30 Please notice that, even though the month of interest is April 2020, January of 2020 has been re-run in the update request by using ERA5 instead of ERA5T. This is to ensure that any potential modifications between the 2 datasets are taken into account at Vortex even though this might have a retrospective effect on updates.
MERRA2 Vortex SERIES have a 2 month + 5 days delay. MERRA2 reanalysis is released with a delay of 2 months and 5 days from the current date.
ERA5 dataset from C3S ECMWF Long-term averaged period 1991-2020 Wind mean speed averaged at 100 m height New monthly anomaly every month 5th Last month anomaly: percentage of monthly mean wind speed over same monthly mean wind speed average for all period 1991-2020. Accumulated monthly anomaly: percentage of January to current month mean wind speed over mean wind speed on the same period for 1991-2020. Quarter anomaly: percentage of quarter mean wind speed over mean wind speed on same quarter for period 1991-2020. Q1, 1st quarter: January, February, and March (JFM) Q2, 2nd quarter: April, May, and June (AMJ) Q3, 3rd quarter: July, August, and September (JAS) Q4, 4th quarter: October, November, and December (OND) Year anomaly: percentage of yearly mean wind speed over mean wind speed for period 1991-2020.
Dataset SEAS5 on single levels from C3S ECMWF. Ensemble mean of all 51 ensembles available. Long-term averaged period 1993-2016. New monthly forecast launch at 1st month day and release at 14th. Actual month anomaly: Mean wind speed anomaly for current month (release on 14th). Anomaly mean wind speed is calculated over the 1993-2016 re-forecast period of SEAS5. Next 2 months: Mean wind speed anomaly for next 2 months. Anomaly mean wind speed is calculated over the 1993-2016 re-forecast period of SEAS5.
Here you can find a summary of the variables included in each product.
The timestamps on the Vortex SERIES are for the end of the period, i.e wind speed at 9PM is th average wind speed from 8 to 9PM.
The pricing information for Vortex data can be found on our pricing page. You can visit the pricing page for detailed information on our pricing plans and subscription options
It depends on the product. Here you can find more information about the delivery time for each product.
Our Richardson number is computed as a “simple” bulk Richardson
approximation by using the results of the model at 10 meter below and
10m above the selected output height by the user. This is the formula we
use for this estimation:
Ri = 9.81*(h2-h0).*(T2-T0) / (T1*((u2-u0)2.+(v2-v0)2.))
-where u -> component U-wind
-where v -> component V-wind
-where T -> potential temperature
-where h -> height above ground level
-0 corresponds to “$yourseriesheight-10”
-1 corresponds to “$yourseriesheight”.
-2 corresponds to “$yourseriesheight+10”
We use the following formula for calculating the Potential Temperature:
T = t + 0.0098*$height + 273.15
where t (lower case) is the temperature in degrees celsius and T (upper
case) is the potential temperature. It’s just a height correction
(ground level height).
It is obvious that we could be using a larger interval, such as +-30 or
+-40 meter, for the Ri bulk estimation but we considered that using
+-10m represented a nice first order approximation and can give a good
idea of the stability at different heights.
Since you can extract Vortex SERIES at different heights, the RI number
will show slight differences at each height. This might be interesting
for offshore projects where temperature inversions and stability is
quite an issue.
Regarding stability conditions, it can be accepted that:
1) Ri < 0 -> Thermal & dynamic unstability. The more the negative, the
more the unstability. That’s very characteristic for daytime hot
conditions.
2) 0.25 > Ri > 0 -> Dynamically unstable. Even though the temperature
profile shows stability, turbulence may occur due to shear. Typical
situation is nocturnal low level jets.
3) 1 > Ri > 0.25 -> Latent turbulence, both in terms of temperature &
wind. The flow is apparently laminar but it will become turbulent with
just a slight perturbation.
4) Ri > 1 -> Laminar flow very stable.
The coordinates system we use is WGS-84.
The wind speed values that appear in the extremes report are the maximum values of 10-minute averages. To explain, the wind speed generated by the model is averaged every 10 minutes, from which we calculate the mean and maximum (gust) values. This is done for each 10-minute interval, for the windiest 6 days of the last 30 years. For each year, we select the maximum of these averages, but it is a maximum of averaged values, not a maximum gust value. We then create a graph with all the years and calculate the extrapolation of Vref50 using a Gumbel fit.
The global map in the interface, although obtained with the same model, has a slightly different regionalization methodology up to 1 km resolution compared to what we use in our products. The Vortex MAP at 1 km should be more accurate due to the type of downscaling used. Additionally, you can download the Vortex MAP in TIFF or KMZ formats to analyze in your software, whereas, with the global map, you can only obtain the approximate average speed value by clicking on the map.
