Skip to content
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

GeoHackWeek 2016 #Renewables Mapping Project

A hack week project to locate optimal sites for renewable energy power plants using random forest classification and Google Earth Engine (GEE).

Run the Project

To run the project:

  1. Create a GEE account.
  2. Copy and paste the code from gee_scripts/prospect.js into the GEE code editor and run.


  1. Load satellite and ground based datasets of features considered useful for determining renewable energy potential (eg. solar irradiance, wind speed).

  2. Locate sites of existing renewable energy power plants (solar, wind and geothermal). Power plant locations (U.S.A) were sourced from: The columns retained for this study are station id, lat, long, solar_MW, wind_MW and geo_MW and a new column powerplant_MW that lists the maximum MW (Mega-Watt) value of the three and so shows where there is a powerplant of this type and the MW. The csv file was converted to a shapefile (.shp) using QGIS and interpolated using IDW in ArcGIS to produce a raster (.tif).

    To upload to GEE, the shp file was read into Google Earth Pro, and save as a kml file. This was read into Google Drive as a Fusion Table. NOTE: in order to work correctly in GEE, columns with 'location' data type that are not lat or lon (e.g. State, City) are removed from the Fusion Table. This Fusion Table can then be correctly read by GEE.

  3. Extract a list of feature values at pixels where renewable energy plants are located to form a training dataset.

    • GEE prospector.js:
      var training = predictors.sampleRegions(allRenewables, ['nameindex'], 30);
  4. Feed the training data to a random forest classifier.

    • GEE prospector.js:
      var bands_more = ['vs', 'srad', 'elevation', 'aspect', 'slope', 'b1', 'b1_1' ];
    • GEE prospector.js:
      var trained_more = ee.Classifier.randomForest({numberOfTrees:500, bagFraction:0.63, outOfBagMode:true}).train(training, 'nameindex', bands_more);
  5. Use the trained classifier to predict which of the 3 types of renewable energy would be best suited to land across the US.

    • GEE prospector.js:
      var classified_more =;
  6. Mask U.S. and areas of importance or impracticality (eg. forest canopy, water, steep slopes, indian reserves, impervious surface (i.e. cities)

    • U.S. boundaries:
      GEE prospector.js:
      var country = ee.FeatureCollection('ft:1N2LBk4JHwWpOY4d9fobIn27lfnZ5MDy-NoqqRpk', 'geometry');
      var usBoundary = country.filter(ee.Filter.eq('ISO_2DIGIT', 'US'));
      var mask = ee.Image.constant(0).int32();
      mask = mask.paint(usBoundary, 1);
      classified_more = classified_more.updateMask(mask);
    • Forest canopy, water, impervious surfaces:
      from USGS/NLCD/NLCD2011
      GEE prospector.js examples:
      Map.addLayer(slope.mask(, {palette:'gray'}, 'Slope Mask');
      Map.addLayer(forestmask,{palette:'088d00'},'Forest Mask 50%',false);
      Map.addLayer(watermask,{palette:'94BFFF'},'Water Mask');
      Map.addLayer(reserveland,{color:'purple',opacity:0},'Reserve Mask',false);

Intial Project Development

Renewable Prospecting Hackpad


No description, website, or topics provided.



No releases published


No packages published

Contributors 4

You can’t perform that action at this time.