3

I would need to predict wind speed and direction at a wind farm station, using an artificial neural network. Unfortunately, the only historical weather data available are the one at some weather stations close to my target point. Thus, I will be able to predict the short term wind speed and direction at those sites and then I will need to interpolate those predicted value in order to find the forecast for the target site.

Can you please suggest me how to do it?

Should I use ArcGIS?

I read about the use of Inverse distance weighting or Kocriging method but I don't know how to use them.

2
  • Why do you mention neural networks? Do you have to use that?
    – Spacedman
    Commented Sep 18, 2015 at 18:45
  • Well, I do not HAVE to use them. But, according the literature, the neural networks can generate a very reliable prediction (in particular for short term, which I am interested in). So, I decided to reproduce my own neural network on MATLAB suitable for the input data that I got.
    – Ilaria DF
    Commented Sep 28, 2015 at 11:04

3 Answers 3

4

If you are using ArcGIS you would find these tools in the Spatial Analyst toolbar > Interpolation. You will input your featureclass containing the wind data, your Z value will be your wind speed, and then choose your output FC which is where you will save it.

Some choices have to be made as to your interpolation type. IDW is best for a dense sample set, SPLINE is best for smooth data. Quoted below is what ESRI has to say on the matter.

IDW (Inverse Distance Weighted) tool uses a method of interpolation that estimates cell values by averaging the values of sample data points in the neighborhood of each processing cell. The closer a point is to the center of the cell being estimated, the more influence, or weight, it has in the averaging process.

Kriging is an advanced geostatistical procedure that generates an estimated surface from a scattered set of points with z-values. More so than other interpolation methods supported by ArcGIS Spatial Analyst, a thorough investigation of the spatial behavior of the phenomenon represented by the z-values should be done before you select the best estimation method for generating the output surface.

Natural Neighbor interpolation finds the closest subset of input samples to a query point and applies weights to them based on proportionate areas to interpolate a value (Sibson, 1981). It is also known as Sibson or "area-stealing" interpolation.

The Spline tool uses an interpolation method that estimates values using a mathematical function that minimizes overall surface curvature, resulting in a smooth surface that passes exactly through the input points.

Spline with Barriers The Spline with Barriers tool uses a method similar to the technique used in the Spline tool, with the major difference being that this tool honors discontinuities encoded in both the input barriers and the input point data.

The Topo to Raster and Topo to Raster by File tools use an interpolation technique specifically designed to create a surface that more closely represents a natural drainage surface and better preserves both ridgelines and stream networks from input contour data.

The algorithm used is based on that of ANUDEM, developed by Hutchinson et al at the Australian National University.

Trend is a global polynomial interpolation that fits a smooth surface defined by a mathematical function (a polynomial) to the input sample points. The trend surface changes gradually and captures coarse-scale patterns in the data.

Here is a previous article on the matter: How do you decide what interpolation method to use for resampling raster data?

Here is a good paper on choosing the correct method: http://webapps.fundp.ac.be/geotp/SIG/interpolating.pdf

2
  • You can test for a best-method too by iteratively leaving out a single input feature from your historical weather data and seeing which method most closely estimates the actual value of the "missing" point. -- If the data set is large and time is important you can use a sub-set to test on. -- Beware that extrapolating data values can be much worse than interpolating, so take note of where the data points are in relation to each other.
    – user23715
    Commented Sep 18, 2015 at 17:44
  • I would differ to @FelixIP and MappaGnosis answers below. I have never worked with wind data before so their answers are the most useful.
    – ed.hank
    Commented Nov 13, 2016 at 16:19
1

ArcGIS has no suitable solution for interpolation of point observations of wind or other weather/climate phenomenon like preciptation and air temperature.

Additional continious surfaces are needed. Embedding of digital elevation model in interpolation process of XY events/climate stations usually enough for air temperature mapping, less so for rain. Unfortunately wind is the hardest nut to crack. I'd say here you'll want some measure of ridge proximity...

Search web using 'anusplin wind interpolation'. This will give you some pointers.

1

Wind speed interpolation is usually done using specialized software (most of it proprietary). Your calculation must include fluid dynamics and account for surface roughness, terrain topology, air density and many other factors. As FelixP points out, ArGIS is not the tool for the job. You do have many other options though, depending on the exact nature of your available data and project tolerances, including (in no order and not limited to):

  1. WAsP
  2. WindFarmer by Garrard Hassan
  3. WindFarm by ReSoft
  4. WindPro
  5. OpenWind (last time I looked, the basic version was free)
  6. WindStation
  7. Adapt an existing Wind Profile program (this particular example program was aimed at the UK but you may be able to adapt it to your uses or find an alternative).

There are also loads of specialist companies offering software and services to the windfarm industry on wind modelling. For instance, if you are in the UK you could contact the MetOffice and commission a Virtual Met Mast report. I have found these to be extremely cost effective and bankably accurate compared to an actual on-site met mast.

1
  • WindNinja uses less than full fluid dynamics (conservation of mass), but it is freely available. The only required data is a DEM file. firelab.org/windninja
    – user10353
    Commented Oct 5, 2015 at 22:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.