How do you decide where to install met sites?
Planning an efficient met campaign can be a challenging task. It’s hard to balance the need for quality data with budgetary restraints. From a resource assessment perspective, the more met sites the better since the model uncertainty is linked to the quality and quantity of the met data. From a budgetary point of view, however, unnecessary or redundant met sites should be avoided since, with every 60-m met mast another ~$40 – $75k USD is added to the cost of the project.
The optimized met campaign is therefore one that minimizes both the cost and the wind flow model uncertainty. But how does one achieve this balance?
The key is having the ability to objectively select the minimum required met tower locations while simultaneously minimizing the wind flow model uncertainty. Below is a step-by-step demonstration of how to use Continuum wind flow modeling software to select met sites and how to monitor the wind flow model uncertainty as more met sites are added to the project.
1) Define project area of interest (AOI)
For this demonstration, an area was randomly chosen. It is ~25 km east-west and ~17 km north-south and is located in northern Texas. (Figure 1)
2) Get digital elevation data and land cover data
Following the steps in this tutorial, GeoTIFFs of the elevation and land cover data were downloaded from the USGS.
3) Create a ‘dummy’ met site
The wind speeds modeled in this exercise are used only in a relative sense. It is important, however, to have an accurate representation of the wind rose. For this reason, one could use data from a near-by project site, airport data or re-analysis data (e.g. MERRA) to form a dummy met site. For this demo, the TAB file from a near-by project in Texas was used and the coordinates were changed to be close to the center of the AOI on high ground. (Figure 1)
4) Create Continuum model
Next, the elevation, land cover and met data were imported into Continuum to form a model. Figure 2 shows the Input tab of Continuum where the required data is uploaded and the elevation map is displayed.
5) Create wind speed map in Continuum
On the ‘Maps’ tab in Continuum, a wind speed map can be easily generated; you specify the bounds of the map, the grid resolution and click “Generate Map”. For this exercise, the entire AOI was mapped and a grid resolution of 500 m was used. Once complete, the wind speed map will be displayed and may be exported as a CSV or WRG file (Figure 3).
6) Analyze modeled wind speeds and identify low/medium/high wind speed areas
Continuum (patent pending) creates relationships between wind speed and terrain exposure and, using a self-learning algorithm, modifies the model coefficients to minimize the met cross-prediction error. The accuracy of the model is very good as demonstrated in this paper, in this comparison and in this study. The uncertainty in the wind flow model is minimized when the full range of wind speeds are represented by the met sites.
The wind speed map generated in Continuum was exported as a CSV and analyzed in Excel. The low land and valleys were omitted and the remaining data was sorted by wind speed. Based on the range of modeled wind speeds, map nodes with low, moderate and high wind speed were identified (Figure 4).
7) Select met site locations
To kick off the met campaign at this proposed site, three locations were selected which represent low, medium and high wind speed areas and are spread across the project area (Figure 5).
The selected met sites are shown on Google Earth in Figure 6 below.
Step 8: Monitor wind flow model uncertainty
After the first three met sites have been installed and data has been collected, a wind flow model can be created and the model uncertainty can be assessed. Depending on the model error, additional met sites may be desired to decrease the uncertainty.
Using the built-in Round Robin analysis tool in Continuum, you can analyze how the model uncertainty changes as more met sites are added to the project. In the Round Robin, met sites are systematically omitted from the model creation and are then predicted by the remaining met sites.
Continuum generates Round Robin analyses using a range of met subset sizes (i.e. the number of met sites used in each model) and plots the error as a function of the subset size. Once the Round Robin error plateaus, the addition of met sites is no longer decreasing the model error and the met campaign can be deemed complete. To learn more, check out this past blog post: When are enough met sites enough?
Continuum provides an objective method of selecting the met locations and takes the guesswork out of planning the met campaign. This technique and the Round Robin analysis ensures that the cost of the met campaign is minimized along with the wind flow model uncertainty.