Supervised Classifications using Google Earth Engine

Getting Started

We will now begin preparing the image(s) that we will be classifying.

Getting Images

Google Earth Engine has a large selection of remote sensing data to choose from. In this example, I will be using the ' USGS Landsat 8 Collection 1 Tier 1 Raw Scenes' collection. This is a collection of the highest quality images available through Landsat 8.

Define a variable 'L8' so that you can easily reference this entire collection:

var L8 = ee.ImageCollection('LANDSAT/LC08/C01/T1');

Creating a Cloud-Free Mosaic

Now that we have a variable named L8 that is linked to the Landsat 8 collection that we want, we can begin to filter images and choose what we want. For this example, I will create a cloud-free mosaic of Fort McMurray in the month after the major fire that swept through in May of 2016. I have created two functions for this, one that creates a cloud mask and the other that creates a simple composite from a collection of images.

1. By utilizing the 'BQA' band present in Landsat 8 imagery, we can create a cloud mask to remove any cloud pixels that may alter the result. I created a function that will select any pixel that has cloud cover with even moderate confidence. The function will return a mask that removes all of these pixels from an image. Note that I have commented on this one fairly heavily due to it be slightly confusing at first.
   1. I should add again, if you haven't already explored all the bands of your chosen imagery, do. There are plenty of interesting things that can be used.

// Cloud mask for L8 imagery
var L8CloudMask = function(L8) {    
  var qa = L8.select('BQA');  //selects the BQA (quality) band
  var CloudOrShadow = qa.bitwiseAnd(1<<4) //chooses pixels where cloud is present
            .and(qa.bitwiseAnd(1<<6)) //and pixels where cloud confidence ic greater than 34%
            .or(qa.bitwiseAnd(1<<8))  //or pixels where cloud shadow confidence is greater than 34%
            .or(qa.bitwiseAnd(1<<12));  //or pixels where cirrus confidence is greater than 34%
  var masked = L8.mask(); //creates a mask
  return L8.updateMask(CloudOrShadow.not()).updateMask(masked); //updates mask to remove any pixels as described above in 'CloudOrShadow'
};

1. Now onto the composite. This is a much simpler function, it simply intakes a collection of images and uses the built in 'simpleComposite' function to combine them into the best possible image.

// Function to create composite
var createComp = function(collection) {
  var comp = ee.Algorithms.Landsat.simpleComposite({ 
    collection: collection,
    asFloat: true
  });
  return comp;
};

1. With the two main functions created, we can now use a few lines to create our final image.

// Select all images within June 2016 that touch the extent and have the least removed from the cloud mask
var after = ee.ImageCollection('LANDSAT/LC08/C01/T1') 
        .filterDate('2016-06-01', '2016-06-30') 
        .filterBounds(extent)
        .map(L8CloudMask);

// Create composite of images chosen above
var compAfter = createComp(after);
// Clip to extent
var clipAfter = compAfter.clip(extent);

Getting Training Data

Assess Useability of Data

histograms etc

Classification

Test on Validation Data Set

error matrix

Run on Chosen Image

===Calculating Area of Each Class