FAQ

Long-term virtual weather data time series refers to model generated data sets that capture the historical records of wind conditions, and related weather, over an extended period at a given time interval such as, for example, every 10-minutes for a given latitude and longitude and height above ground. These time series provide information about the variation in wind speed and direction, and other weather variables, at a specific location over months, years, or even decades.

The basic wind variables provided in a standard wind VMT are:

• wind speed (m/s)
• wind direction (degrees relative to true north)
• air density (kg/m 3 )
• temperature (degrees C)
• barometric pressure (mb also known as hPa)
• relative humidity (%)

Extra variables that can be added to a given VMT data time series, available for additional cost are:

• precipitation (mm)
• precipitation Type (snow/rain)
• 3-second wind gust (m/s)
• TI (turbulence intensity)
• vertical wind speed (m/s)
• Delta T per 100 meters (°C/100-m)
• wind veer in degrees per 100 meters (°/100-m)
• shear exponent across 100 meters
• Bulk Ri (Richardson number)

Long-term wind time series data is typically collected by weather monitoring stations equipped with anemometers and wind vanes, which are instruments used to measure wind speed and direction. These stations are strategically placed in various geographical locations, to support societal infrastructure such as aviation, and to support the weather observation network used for daily weather forecasting. Data are recorded at regular intervals, such as every hour or every few minutes

A common use of long-term wind time series is as a “long-term reference” for a wind project site being considered for development. Typically, on-site measurements are gathered for a short period (1-5 years), which provides a view into the hourly to seasonal variability of the wind, but not the interannual variability and long-term mean over say 10 to 60 years. The long-term reference data provides that interannual to decadal context for the on-site measurements, but is only useful for that purpose if it correlates well with the on-site data at the daily time scale. If the reference site is too far away or in too different of a geography, correlations with on-site measurements will be low. Sometimes there simply have never been observations of the weather anywhere near or that are representative of a given project location. This is where VMTpro provides a valuable alternative. While model generated, it is drawn from simulations of the atmosphere at the precise location of interest rather than from a distant observation site, so significant improvements over distant off-site references can be obtained.

Another now widely used choice for a long-term reference is the raw reanalysis data itself, either from NASA’s MERRA2 dataset, or from ERA5. However, with very coarse grid spacings of roughly 55 km (for MERRA2) or 28 km (for ERA5), important local terrain or coastal features, and associated wind flow responses, are not well captured. VMTpro improves on these options by essentially downscaling the reanalysis with a highly sophisticated and high-resolution weather prediction model (WRF), leading to better representation of local topography and wind flow, and higher correlations to measurements.

Configuring and running the Weather Research and Forecasting (WRF) model for a local area involves several steps. Here's a general overview of the process:

• Define and select the model do main size and location, grid resolution, physics parameterizations, and other model settings.
• Prepare the initial and boundary conditions data required for the simulation from the ERA5 Reanalysis source data for the full length of the desired time series.
• On a supercomputing cluster with several thousand processors, run the WRF model.
• Extract from the model results from the desired variables at the desired horizontal locations and vertical heights, for a continuous time series at the chosen length of time and time interval. Time periods that are available range from 5 to 60 years. Time interval choices are 1 minute, 2 minutes, 5 minutes, 10 minutes, 15 minutes, 20 minutes, 30 minutes or 60 minutes.
• It's important to note that running the WRF model requires computational resources and expertise in atmospheric modeling. You may need to consult with ArcVera’s experts to ensure proper interpretation of the model results for your local area.

ERA5 (Fifth Generation of the European Re-Analysis) is a global atmospheric reanalysis dataset produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). Reanalysis refers to the process of combining historical observations with numerical weather prediction models to create a consistent and comprehensive record of the Earth's climate system.

ERA5 provides high-resolution atmospheric, land surface, and oceanic variables covering the entire globe from 1959 to the present. It is based on ECMWF's Integrated Forecasting System (IFS), a state-of-the-art numerical weather prediction model covering the globe at coarse tile size compared to VMT Pro. The dataset incorporates a vast amount of observations from various sources, including surface observations, radiosondes, satellites, and remote sensing instruments.

Some key features of ERA5 include:

• Temporal Resolution: ERA5 provides hourly data, allowing for detailed analysis of diurnal variations and short-term weather phenomena.
• Spatial Resolution: The dataset offers a horizontal grid spacing of 0.25 degrees latitude and longitude (approximately 28 km), providing a high level of detail for global atmospheric conditions.
• Variables: ERA5 includes a wide range of atmospheric variables, such as temperature, humidity, wind speed and direction, precipitation, surface pressure, solar radiation, and many others. It also provides information on land surface variables like soil moisture, snow depth, and sea ice cover.
• Vertical Levels: ERA5 offers data at multiple vertical levels in the atmosphere, allowing for analysis and research across the vertical structure of the atmosphere.
• Assimilation Methods: ERA5 utilizes advanced data assimilation techniques to merge observations with the IFS model, ensuring that the reanalysis captures the best estimate of the atmospheric state at each time step.

ERA5 has been widely used in climate research, weather forecasting, and various applications across different fields. It supports studies on climate variability, trend analysis, extreme events, climate model evaluation, and more. The dataset is freely available to the public, and users can access ERA5 data through the ECMWF Climate Data Store (CDS) or other data portals that provide access to ECMWF's data holdings.

Model Output Statistics (MOS) bias correction is used for the improvement of numerical weather prediction (NWP) outputs. In the context of long time series for wind energy development sites, it's crucial to have accurate wind speed and direction forecasts, even slight inaccuracies in wind prediction can have significant financial implications.

MOS stems from the idea that numerical models (like those used for weather forecasting) often have systematic biases. These biases might be due to model physics, resolution, or other factors.

MOS provides a way to statistically correct these biases by comparing model outputs with observed data. To develop MOS corrections, you'd typically use a historical dataset where both the model predictions and actual observations are known.

Using statistical techniques, relationships are derived between the model outputs and observations.

This relationship then serves as the correction factor for future model outputs. For wind energy development, long time series of wind data (often spanning multiple decades) are used to assess the feasibility, expected energy production, and potential financial returns of a site. If the historical data comes from NWP models, it might carry biases.

Applying MOS bias correction can make these long time series more representative of the actual wind conditions at the site.

Grid spacing (or tile size) for wind time series refers to the distance between adjacent grid points in the spatial grid on which the WRF model is computed. ArcVera’s model domain covers not just the requested point itself, but a surrounding area large enough to capture the relevant meteorology that affects the variables of interest at the point. Effects such as terrain-induced thermal flows, land-sea breezes, gap flows, downslope gravity wave effects, flow modulation due to land surface contrasts, and other effects impact the point variables. All of these effects are influenced by local variations in topography, land surface type, and coastlines, so the more highly resolved these variations are by the model grid, the more accurately the model will simulate these effects, thereby improving the accuracy of the time series. Available grid spacing (or tile size) choices are 3 km, 2 km, 1 km, and 600 m.

The VMTpro product is updated when research indicates material improvement is likely. This often corresponds to updates to the WRF model. Customers will be informed when improvements are implemented. Numerous additional features are planned and will be released to VMTpro over time based on customer feedback, changing market requirements, and as improvements are made.

Our data sets are flexible and offer a choice of grid spacing ranging from 3 km to 600 meters granularity, and a choice of time intervals from 60 minutes down to 1 minute. The user may choose a period for the time series spanning up to 60 years.

VMTpro covers from -58º (south) latitude to 70º (north) latitude. And also replace this:

Yes. Each request kicks off a client-specific VMTpro simulation based on the user-specified choices entered into the VMTpro order page. The order number, time stamp, and email generated by the VMT Pro order interface is used to track the computing simulation, and the processed output, which is sent to the customer using the same number.

Orders typically complete in less than 6 hours with 99.5% of orders completing within 48 hours. If MOS-correction is requested, turnaround time is between 5 and 10 days.

No, time series are complete, without gaps. For example, if a customer orders a 1-minute data set for 60 years, there will be at least 31,536,000 individual minute rows in the data set. This is calculated as 60 minutes per hour times 8760 hours per year times 60 years (60 x 8760 x 60 = 31,536,000 minutes). Leap years and partial most-recent-year data will increase the amount of data supplied to the customer.