Google Earth Engine – How to Add Only Certain Types of MODIS Land-Cover

google-earth-engineland-covermodis

I'm trying to create a map on Google Earth Engine with only some Land-Cover class, but I can't find a solution to this problem.

I'm using this MODIS dataset: MODIS/061/MCD12Q1/2002_01_01
I'm using this type of Land-Cover: LC_Type1

From LC_Type1, I just want to use the classes from 1 to 5 (the ones described as Evergreen Needleleaf Forests, Evergreen Broadleaf Forests, Needleleaf Forests, Broadleaf Forest, and Mixed Forests)

How do I filter only these 5 classes?

The final aim is to use these classes to extract NPP values from another MODIS dataset for computing a time-series chart of yearly NPP. This last part is not a problem because is something I just manage to resolve. My only problem is with the "filtering" of these land-cover classes.

I tried to filter these classes but it is always giving me errors.

I'm stuck here:

// MODIS Land-Cover dataset
var landcover = ee.Image("MODIS/061/MCD12Q1/2002_01_01")
                  .select('LC_Type1')
                  .clip(roi)


// visualization palette
var LCVis = {
  min: 1.0,
  max: 17.0,
  palette: [
    '05450a', '086a10', '54a708', '78d203', '009900', 'c6b044', 'dcd159',
    'dade48', 'fbff13', 'b6ff05', '27ff87', 'c24f44', 'a5a5a5', 'ff6d4c',
    '69fff8', 'f9ffa4', '1c0dff'
  ],
};


Map.centerObject(roi, 8)
Map.addLayer(landcover, LCVis, 'MODIS land-cover')

I tried

// MODIS Land-Cover dataset
var landcover = ee.Image("MODIS/061/MCD12Q1/2002_01_01")
                  .select('LC_Type1', [1,5])
                  .clip(roi)

but it's not working.

Is there a solution?

Best Answer

You can mask the other pixels

// create mask with pixels values less than or equal to 5
var mymask = landcover.lte(5);

// Apply mask
var forest = landcover.updateMask(mymask);

// Visualize
Map.addLayer(forest, LCVis, 'MODIS forest')

Related Solutions

[GIS] Extracting land cover type in Google Earth Engine

Here are two ways of doing this. The first is your method of masking with some simplified code. The second is using a grouped reducer:

var s1 = ee.ImageCollection("COPERNICUS/S1_GRD"),
S2 = ee.ImageCollection("COPERNICUS/S2"),
fc = ee.FeatureCollection("ft:1w_ciwxR9FKm0rzMH8oJ3MdyE-1U15PfiOFlGW5Ib"),
fc2 = ee.FeatureCollection("ft:18Xvf7jG2EI4mJMwj0oLV2u-05bAPiwJDaxSRy2ep");

var aoi = fc; 

var bands = ['B2', 'B3', 'B4', 'B8', 'B11', 'B12'];

var S2 = ee.ImageCollection('COPERNICUS/S2')
  .filterDate('2016-07-01', '2016-08-01')
  .filterBounds(aoi);

// You may want to use median here if there's overlapping imagery.
var image = S2.mosaic();

var geometry = ee.Geometry.Rectangle([36.977296602,-15.364411987,37.057018981,-15.447436229]);
Map.centerObject(geometry);

var landcover_roi = image.clip(geometry);

Map.addLayer(landcover_roi, {'bands': 'B12,B8,B4','min': 0,'max': 4000}, 'Sentinel 2A SWIR,NIT,R', 0);

var training = landcover_roi.sampleRegions(fc2, ["info"], 30);
var classifier = ee.Classifier.cart()
    .train(training, "info", image.bandNames());
var out = landcover_roi.classify(classifier);
Map.addLayer(out, {palette: [
  'ff0000',// cleared red
  'ffcc66',// cultivated
  '387242',// forest 
  '99ff33',// grass 
  'e0e0d1',// rock 
  'e0e0d1'// rock grass
], min:1, max:6});

Map.addLayer(ee.Image().paint(fc, 1, 3));
var fc = fc2;
Map.addLayer(fc2);

var names = ['cleared', 'cultivated1', 'cultivated2', 'grass', 'rock', 'rockgrass'];
var total = out.eq([1, 2, 3, 4, 5, 6]).rename(names);

var count = total.multiply(ee.Image.pixelArea());

// The grouped reducer ignores the weights.  Unweight here for comparison.
var area = count.reduceRegion(ee.Reducer.sum().unweighted(), geometry, 30);
print(area);

var stats = ee.Image.pixelArea().addBands(out).reduceRegion({
  reducer: ee.Reducer.sum().group(1), 
  geometry: geometry, 
  scale: 30,
});
print(stats);

Google Earth Engine – How to Reduce Land Cover Data Sets in Earth Engine

You can get the "histogram" property (resulting from the reducer) as a Dictionary which can be used to set the landcover:frequency pairs as new properties for a county feature. Then just map that function over all counties in the FeatureCollection.

// Print out the frequency of landcover occurrence for each county
//...

// Optional for debugging: Drop other properties and geometry > simpler looking csv.
frequency = frequency.select(['histogram'], null, false)

var frequency = frequency.map(function(feature){
  var dict = ee.Dictionary(feature.toDictionary().get('histogram'))
  feature = feature.set(dict)
  return feature
})
print('per class landcover frequency properties', frequency)

//Exports table to drive
Export.table.toDrive(frequency, "MaineLandCoverData");

Related Question