Google Earth Engine Training
Gennadii Donchyts, Corine ten Velden, Arjen Haag, Jaap Schellekens, Hessel Winsemius
Agenda
Time Description Source
09:00 Start
09:05 Introduction of the course
09:10 Google Earth Engine and its possibilities 1. [Link]
09:30 Tutorials 2. [Link]
GEE_Background_and_Exercises
Playground; images and image
collections; image composites; spectral
indices (NDVI, NDWI);
10:30 - 10:45 Coffee break
10:45 Tutorial 2. [Link]
GEE_Background_and_Exercises
Importing and exporting rasters and
vectors; exporting videos; elevation
datasets;
12:30 - 13:15 Lunchbreak
Agenda
13:15 Optical Image Analysis 3. Optical Image [Link]
GEE_Advanced_Optical
DN, TOA, SR; cloud masking;
topographic analysis; supervised
classification; image processing;
harmonic fit
14:15 Surface Water Detection 4. Water [Link]
GEE_Advanced_Water
spectral water indices;
thresholding; commission/omission
errors
15:15 Break
15:30 SAR Analysis 5. SAR [Link]
GEE_Advanced_SAR
GRD;polarization modes; speckle
noise and filters; water detection
17:00 Wrap-up Feedback
Earth Engine:
Google's Cloud Platform for
Big Earth Data Analytics
Google Earth Earth Engine
3-D Globe Geospatial
Visualization Analysis
Requests Results
add mosaic
Geospatial filter
join
distance
Algorithmic
Datasets focal_min
reduce
convolve Primitives
Storage and Compute
The Earth Engine Public Data Catalog
Landsat 4, 5, 7, 8 MODIS Terrain Land Cover Atmospheric
Raw, TOA, SR, ... Daily, NBAR, LST, ... SRTM, GTOPO, NED, ... GlobCover, NLCD, ... NOAA NCEP, OMI, ...
... and many more, updating daily!
> 200 public datasets > 4000 new images every day
> 5 million images > 5 petabytes of data
Earth Engine: Origins
Global Forest Watch
JavaScript & Python API
// Make a median composite from two years of Landsat 7.
// Load the image collection.
var collection = [Link]('LANDSAT/LE7_L1T');
// Filter it down to 2011 and 2012.
var filtered = [Link]('2011-01-01', '2012-12-31');
// For each pixel, for each band, calculate the median and make an image
// of the result. The median tends to remove clouds, shadows, data gaps.
var medianImage = [Link]();
// Add the image to a map.
[Link](medianImage, {bands:['B3', 'B2', 'B1'], min:35, max:170});
Code Editor link:
[Link]
Timelapses
Iranian Bay
Colorado River
Pacullpa, Peru
Singapore
Flood Extent Mapping
Focus Area
Mekong river and tributaries from Vientiane, Lao PDR, to
the South China Sea. (With adjustments the tool can easily Initial Results
be used elsewhere.)
Thailand Example: Modified Normalized Difference Index
(8th percentile) over the Landsat 8 period.
This normally represents flood water.
Water detection
Thailand Example: Modified Normalized Difference Index
(40th percentile) over the Landsat 8 period.
This normally represents permanent water.
Contact: [Link]@[Link]
Surface water mapping for Australia
Comparison of detected water mask to OpenStreetMap water
[Link] – water mask for Murray & Darling Basin, Australia
Contact: [Link]@[Link] Donchyts et. al. 2016, [Link]
[Link]
(Arsen [Link]., 2014)
Area
get cloud-free images
extract boundary
interpolate
Reconstruction of reservoir surface area using optical satellite imagery
Water mask with water occurrence image
Surface area time series estimation (2013-2015)
Coastline Extraction – Zandmotor
Contact: [Link]@[Link]
Participatory DEM generation
Contact: [Link]@[Link]
Height above the nearest drainage (HAND)
Contact: [Link]@[Link]
Height above the nearest drainage (HAND)
Contact: [Link]@[Link]
Height above the nearest drainage (HAND)
Contact: [Link]@[Link]
Aqua Monitor
[Link]
Tehri Dam
Contact: [Link]@[Link]
Aqua Monitor
Contact: [Link]@[Link] [Link]
Aqua Monitor
Contact: [Link]@[Link] [Link]
Aqua Monitor
Contact: [Link]@[Link] [Link]
Aqua Monitor
Contact: [Link]@[Link] [Link]
Aqua Monitor
Contact: [Link]@[Link] [Link]
Aqua Monitor
Contact: [Link]@[Link] [Link]
Aqua Monitor
Contact: [Link]@[Link] [Link]
Questions?