Session 1 - Classification_on_GEE_SOLVED

//+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
//S. Aparício (Solenix), I. Alonso (Solenix)  ☁☁☁\_(ツ)_/☁☁☁😀
// 
//MIT LICENSE CC BY EUMETSAT
//+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
//🔥🛰️
// Session 1 “Hands-on: Classification of fire burns"
//SOLUTIONS

///////////////////////////////////////////////////////////////////////////
//////////////////////   VISUALIZING DEFINED AOI           ////////////////
///////////////////////////////////////////////////////////////////////////
//______________
//_**STEP 1**___

var AOI = AOI_Portugal //AOI_Portugal or AOI_Lassen or AOI_Catalunia
Map.addLayer(AOI, {'color': 'FF0000'}, 'PT', false);    
Map.centerObject(AOI, 10)

///////////////////////////////////////////////////////////////////////////
//////////////    PREPARING SATELLITE IMAGERY     /////////////////////////
///////////////////////////////////////////////////////////////////////////
//______________
//_**STEP 2**___  

//Selecting date range - after the fire
var start = ee.Date('2017-08-10');  //Lassen  '2021-09-15'  // Portugal '2017-08-10' //Catalunia '2022-06-19'  //Afghanistan '2022-05-25' 
var finish = ee.Date('2017-09-10'); //Lassen  '2021-09-17'  // Portugal '2017-09-10' //Catalunia  '2022-06-20' //Afghanistan '2022-06-05' 

//Selecting date range - before the fire
var startpast = ee.Date('2016-08-10' );  //Lassen '2021-07-12'    //Portugal '2016-08-10' //Catalunia '2022-02-28' //Afghanistan '2022-05-01'
var finishpast = ee.Date('2016-09-10'); //Lassen  '2021-07-14'    //Portugal '2016-09-10' //Catalunia '2022-03-02' //Afghanistan '2022-05-07'

////////////// CREATING COLLECTION OF IMAGES///////////////////////////////
//______________
//_**STEP 3**___

//Sentinel-2 collection after the fire
var S2after = ee.ImageCollection('COPERNICUS/S2')
        .filterDate(start, finish)                               
        .filterBounds(AOI)                                         
        .filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 1)); 
        
//Sentinel-2 collection before the fire
var S2before = ee.ImageCollection('COPERNICUS/S2')
        .filterDate(startpast, finishpast)                  
        .filterBounds(AOI)                                     
        .sort('CLOUD_COVER')
        .filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 1)); 

////////////// CLOUD MASK FUNCTION ///////////////////////////////
//______________
//_**STEP 4**___
//Cloud mask S2 function
function maskS2clouds(image) {
  var qa = image.select('QA60');

  // Bits 10 and 11 are clouds and cirrus, respectively.
  var cloudBitMask = ee.Number(2).pow(10).int();  //Math.pow(2,10)
  var cirrusBitMask = ee.Number(2).pow(11).int(); //Math.pow(2,11)

  // Both flags should be set to zero, indicating clear conditions.
  var mask = qa.bitwiseAnd(cloudBitMask).eq(0).and(
             qa.bitwiseAnd(cirrusBitMask).eq(0));

  // Return the masked and scaled data.
  return image.updateMask(mask); // add >>>> .divide(10000) >>>Diving by 10000 converts toa
}

/////////// APPLY CLOUD MASK FUNCTION TO ALL IMAGES IN THE COLLECTION///////////////////////////////
//______________
//_**STEP 5**___

//Applying the function to remove clouds to all the collection
var S2Acloudfree = S2after.map(maskS2clouds);
var S2Bcloudfree = S2before.map(maskS2clouds);

/////////// CREATE A SINGLE IMAGE ///////////////////////////////
//______________
//_**STEP 6**___
//Creating a single image with the median of the collection
var S2A = ee.Image(S2Acloudfree.median());
var S2B = ee.Image(S2Bcloudfree.median());


/////////////////////////////////////////////////////////////////////////////////////////////////////
//////////////    VISUALIZING SATELLITE IMAGERY     /////////////////////////////////////////////////
/////////////////////////////////////////////////////////////////////////////////////////////////////
//______________
//_**STEP 7**___

//Creating visualization settings for false color
var False_vis = {
  min: 260,
  max: 4400,
  gamma: 1.4,
  bands: ['B8', 'B4', 'B3']
};

//Creating visualization settings for true color
var rgb_vis = {
  min: 0,
  max: 2500,
  gamma: 0.89,
  bands: ['B4', 'B3', 'B2']
};


Map.addLayer(S2A.clip(AOI), False_vis, 'S2 After Fire (False Color)')
Map.addLayer(S2B.clip(AOI), False_vis, 'S2 Before Fire (False Color)')

Map.addLayer(S2A.clip(AOI), rgb_vis, 'S2 After Fire (True Color)')
Map.addLayer(S2B.clip(AOI), rgb_vis, 'S2 Before Fire (True Color)')

/////////////////////////////////////////////////////////////////////////////////////////////////////
////////////// CREATING TRAINING DATASET ////////////////////////////////////////////////////////////
/////////////////////////////////////////////////////////////////////////////////////////////////////
//______________
//_**STEP 8**___

//collect more training data on geometry imports>burn/noburn and selecting on the map


//combine the polygons from both classes
var polygons = burn.merge(noburn)

//selectino of the bands to be used
var bands = ['B2', 'B3', 'B4', 'B8']

//______________
//_**STEP 9**___
//collecting training data
var training = S2A.select(bands).sampleRegions({
  collection: polygons,
  properties: ['landcover'],
  scale: 10
});

// Roughly 80% training, 20% testing.
var split = 0.8;
//Spliting with the split value after assigned random values
var trainingRandom = training.randomColumn('random');

var trainingSplit = trainingRandom.filter(ee.Filter.lt('random', split));    //for classification
var testingSplit = trainingRandom.filter(ee.Filter.gte('random', split));    //for validation and accuracy

///////////////////////////////////////////////////////////////////////////////////////////////////////
////////////// TRAINING THE MODEL /////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////////////////////////////
//______________
//_**STEP 10**___

// Training classifiers with training (splited) data
var trainedRF = ee.Classifier.smileRandomForest({
                                    numberOfTrees: 21
                                    }).train(trainingSplit, 'landcover', bands);


/////////////////////////////////////////////////////////////
////////////// RUN CLASSIFICATION  ///////////////////////////
////////////////////////////////////////////////////////////
//______________
//_**STEP 11**___

// Classify the Image with the same bands used for training.
var classified   = S2A.select(bands).classify(trainedRF);



/////////////////////////////////////////////////////////////
////////////// VISUALIZE RESULTING CLASSIFICATION  /////////
////////////////////////////////////////////////////////////
//______________
//_**STEP 12**___

var visClassRFS1S2 = classified.visualize({
                    min:0, 
                    max:2, 
                    palette: ['E6E600','A6FFE6']
});


Map.addLayer(classified.clip(AOI), {min: 0, max: 1, palette: ['green','red']}, 'RandomForest Burnt'); 


//////////////////////////////////////////////////////////////////
////////////// CALCULATING AREAS PER CLASS (OPTIONAL 1)  /////////
//////////////////////////////////////////////////////////////////
//______________
//_**STEP 12**___


var areaImage = ee.Image.pixelArea().addBands(
      classified)
 
var areas = areaImage.reduceRegion({
      reducer: ee.Reducer.sum().group({
      groupField: 1,
      groupName: 'landcover',
    }),
    geometry: AOI_Portugal,
    scale: 500,
    maxPixels: 1e10
    }); 
 
print(areas)

//////////////////////////////////////////////////////////////////
////////////// CALCULATING AREAS PER CLASS (OPTIONAL 2)  /////////
//////////////////////////////////////////////////////////////////

//Select the class from the classified image
var burnt = classified.select('classification').eq(1);//vegetation has 1 value in your case

//Calculate the pixel area in square kilometer
var area_burnt = burnt.multiply(ee.Image.pixelArea()).divide(1000*1000);

//Reducing the statistics for your study area
var stat = area_burnt.reduceRegion ({
  reducer: ee.Reducer.sum(),
  geometry: AOI_Portugal,
  scale: 30,
  maxPixels: 1e9
});

//Get the sq km area for vegetation
print ('Burnt Area (in sq.km)', stat);