Random forest (RF) Image Classification Using Google Earth Engine (GEE):
Here we will be doing satellite image processing by Google Earth Engine (GEE), Here we will classify random forest (RF) images using Google Earth engine (GEE).
We've put the whole process of this RF classification on YouTube, and the programming code is given below to make the process easier.
YouTube Link: https://youtu.be/moFOIpm3JcI
CODE:
-------------------------------------------------------------------------------------------------------------
Map.centerObject(studyarea,11);
var SFCCvis = {bands: ['B8', 'B4', 'B3'], max: 3000};
////////Sentinel-2 Data 2020////////
var S2020 = ee.ImageCollection('COPERNICUS/S2')
.filterDate('2021-03-01', '2021-03-31')
.filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 20))
.filterBounds(studyarea);
print(S2020, 'S2 Data 2020');
var listS2020dataset=S2020.toList(S2020.size());
print(listS2020dataset, 'Data list_2020');
var img1 = ee.Image(listS2020dataset.get(5)).clip(studyarea);
Map.addLayer(img1, SFCCvis, 'SFCC Image1');
//Cloudfree scenes: 0,2,4,5
var imgcollection=ee.ImageCollection([ee.Image(listS2020dataset.get(0)),
ee.Image(listS2020dataset.get(2)),ee.Image(listS2020dataset.get(4)),
ee.Image(listS2020dataset.get(5))])
.select(['B2','B3','B4','B5','B6','B7','B8','B8A','B11','B12']);
print(imgcollection, 'imgcollection');
////Image stacking
var stackCollection = function(imgcollection) {
var first = ee.Image(imgcollection.first()).select([]);
var appendBands = function(image, previous) {
return ee.Image(previous).addBands(image);
};
return ee.Image(imgcollection.iterate(appendBands, first));
};
var image_stack = stackCollection(imgcollection);
print(image_stack, 'Image Stack');
////// Image Classification
// Class Merging
var polygons=Forest.merge(Cropland).merge(Fallow).merge(Waterbody).merge(Sand);
print(polygons, 'Training data');
var sample = polygons.randomColumn();
var trainingsample = sample.filter('random <= 0.75');
var validationsample = sample.filter('random > 0.75');
print(trainingsample, 'Training sample');
print(validationsample, 'Validation sample');
var training = image_stack.sampleRegions({
collection: trainingsample,
properties: ['Class'],
scale: 20
});
print(training, 'Training data Band values');
var validation = image_stack.sampleRegions({
collection: validationsample,
properties: ['Class'],
scale: 20
});
//RF Classifier Model Building
//ee.Classifier.smileRandomForest(numberOfTrees,
//variablesPerSplit, minLeafPopulation, bagFraction, maxNodes, seed)
var RFclassifier = ee.Classifier.smileRandomForest(200).train(training, 'Class');
var Classified = image_stack.classify(RFclassifier).clip(studyarea);
//print(Classified, 'Classified');
var Palette = [
'green', //Forest
'yellow', //Cropland
'#a89e32', //Fallow
'blue', //Waterbody
'grey', //Sand
];
Map.addLayer(Classified, {palette: Palette, min: 1, max: 5}, 'Classified map');
//// Accuracy Assessment
// Get a confusion matrix and overall accuracy for the training sample.
var trainAccuracy = RFclassifier.confusionMatrix();
print('Training error matrix', trainAccuracy);
print('Training overall accuracy', trainAccuracy.accuracy());
// Get a confusion matrix and overall accuracy for the validation sample.
validation = validation.classify(RFclassifier);
var validationAccuracy = validation.errorMatrix('Class', 'classification');
print('Validation error matrix', validationAccuracy);
print('Validation accuracy', validationAccuracy.accuracy());
///// Variable Importance
var explain = RFclassifier.explain();
print(explain, 'Explain');
//Variable Importance of RF Classifier
var variable_importance = ee.Feature(null, ee.Dictionary(explain).get('importance'));
// Chart of Variable Importance of RF Classifier
var chartTitle = 'Random Forest: Bands Variable Importance';
var chart =
ui.Chart.feature.byProperty(variable_importance)
.setChartType('BarChart')
.setOptions({
title: chartTitle,
legend: {position: 'none'},
hAxis: {title: 'Importance'},
vAxis: {title: 'Bands'}
});
// Chart: Location and Plot
chart.style().set({
position: 'bottom-left',
width: '400px',
height: '400px'
});
Map.add(chart);
////// Export
Export.image.toDrive({
image: Classified,
description: 'RF_Classified_map',
region: studyarea,
scale: 20,
fileFormat: 'GeoTIFF',
maxPixels: 1e9,
});
//https://code.earthengine.google.com/73aa2cc676bdca3f0c866472b3097d60
-------------------------------------------------------------------------------------------------------------
We hope you find this code useful. Stay tuned with us to find out and learn new things like this. If there is any problem you can let us know by commenting below or contacting us.