National Aeronautics and Space Administration

Part 1: Using Google Earth Engine for Land

Monitoring Applications
Zach Bengtsson, Juan Torres-Pérez, and Amber McCullum

June 16, 2021

Course Structure and Materials

• Three 2-hour sessions on June 16, 23, & 30

• Sessions will be presented once in English 12:00-
14:00 EDT
• Webinar recordings, PowerPoint presentations,
and the homework assignment can be found
after each session at:
– https://appliedsciences.nasa.gov/join-
mission/training/english/arset-using-google-
earth-engine-land-monitoring-applications
– Q&A following each lecture and/or by email
at:
• [email protected]
• [email protected]
• [email protected]

NASA’s Applied Remote Sensing Training Program 2

Webinar Agenda

Part 1: Google Earth Part 2: Land Cover

Part 3: Time Series Analysis
Engine Basics & General Classification & Accuracy
& Change Detection
Applications Assessment

NASA’s Applied Remote Sensing Training Program 3

Webinar Agenda

Part 1: Google Earth

Engine Basics & General
Applications
Zach Bengtsson Juan Torres-Pérez Amber McCullum

NASA’s Applied Remote Sensing Training Program 4

Google Earth Engine Account Reminder

• Make sure you sign up for a

Google Earth Engine account
as soon as possible, if you
haven’t already, using the link
below:
– https://signup.earthengine.g
oogle.com/#!/
– A Gmail address is not
required. It is recommended
that you use your
work/institutional email.

NASA’s Applied Remote Sensing Training Program 5

Session Outline

• Introduction to Google Earth Engine (GEE)

– Advantages of cloud-based raster computing and remote sensing analysis
functionalities
– Available data from land monitoring satellite sensors
– Applications of GEE for land management
• GEE JavaScript API Activity
– API overview and data catalog search
– Importing and filtering Landsat 8 data
– Simple vegetation indices calculation
• Python API Demonstration
– Brief review of Python API available for GEE

NASA’s Applied Remote Sensing Training Program 6

 Introduction to Google Earth Engine
Functionalities and Available Data Types
Cloud-Based Raster Computing for Remote Sensing Analysis

• Cloud-based raster
computing removes barriers
and limitations related to…
– Data hosting and storage
– Imagery access and
availability
– Personal computing
capabilities
• GEE is also free for scientists,
researchers, and developers.
Google Earth Engine code editor interface using the JavaScript API,
displaying Landsat 8 surface reflectance true color imagery for the U.S.
Credit: Google Earth Engine Developers

NASA’s Applied Remote Sensing Training Program 8

The Google Earth Engine Platform

• Google Earth Engine (GEE) takes

advantage of cloud computing
capabilities to provide users with
a single place for accessing
satellite data, applying remote
sensing methodologies, and
displaying analysis results.
• GEE’s application programming
interface (API) allows users to
easily apply land cover
monitoring algorithms and
classifications with coded
commands.

Image Credit: Google Earth Engine

NASA’s Applied Remote Sensing Training Program 9

Application Programming Interface (API)

• The Earth Engine JavaScript API

is currently the most widely used
method of working with GEE.
• A Python API through Google
Colaboratory (Colab) is also
available for those interested in
using Python.
– This is a bit more
complicated than working
directly in the GEE code
editor with JavaScript.
Google Colab notebook using a coded section to display elevation in an output
cell. Credit: Google Colab

NASA’s Applied Remote Sensing Training Program 10

Google Earth Engine Functionality

• Uses of GEE for satellite imagery

analysis include:
– Automation of data
processing and display
– Near real-time monitoring
(limited by the availability of
data in the catalog)
– Machine learning algorithm
application
– Graphical User Interface
implementation

Simple Classification and Regression Trees (CART) classifier

implemented in the GEE API to identify three classes urban, forest, and
water in the San Francisco Bay Area for May 2021 using Landsat 8
imagery. Credit: Google Earth Engine Developers
NASA’s Applied Remote Sensing Training Program 11
GEE Land Applications

• Long-term monitoring of landscape

change and land cover type
• Computation of indices relevant to land
management such as normalized
difference indices for vegetation, water,
snow, soil, and urban areas
• Landscape time series analysis and
change detection
• Summary statistics
• Validation and accuracy assessment
methods
• Visualization and presentation of results
Time series of MODIS NDVI displayed using Google Earth
Engine. Image Credit: Google Earth Engine Developers

NASA’s Applied Remote Sensing Training Program 12

Available Satellite Sensor Data in GEE: Landsat Series

• Data from the entire Landsat series is

available for use in GEE, including:
– Landsat 1-5
• Multispectral Sensor Radiance
– Landsat 4-8
• Raw Images
• Top of Atmosphere
• Surface Reflectance
• The full archive includes data from 1972 to
present day, with data from 1982 onward
at 30m resolution (Landsat 4-8). Available Landsat series
• GEE Data Catalog Link: collections in the GEE data
catalog. Image Credit: Earth
– https://developers.google.com/earth- Engine Data Catalog

engine/datasets/catalog/landsat

NASA’s Applied Remote Sensing Training Program 13

Available Satellite Sensor Data in GEE: Sentinel-2

• Available Sentinel-2 data

includes:
– Top of Atmosphere
– Surface Reflectance
• Sentinel-2 land applications
are very similar to those of
Landsat, but differences
include:
– 10-20m spatial resolution
– 5-day revisit
– Less temporal coverage Available Sentinel-2 data products in GEE. Note the temporal coverage starts in 2015 and
• GEE Data Catalog Link: 2017. Image Credit: Earth Engine Data Catalog

– https://developers.google.
com/earth-
engine/datasets/catalog/
sentinel-2

NASA’s Applied Remote Sensing Training Program 14

Available Satellite Sensor Data in GEE: MODIS

• MODIS reflectance data products

include:
– Daily, 500m Nadir BRDF-
Adjusted Reflectance
– Daily, Global, 250m Surface
Reflectance
– 8-Day, Global, 250m Surface
Reflectance
• Pre-processed products include:
– Daily, Global, 500m Snow Cover
– 16-Day, Global, 250m
Vegetation Indices False color image of the Terra Surface Reflectance Daily Global 250m projected in the
– Yearly, Global, 500m Land GEE JavaScript API. Credit: Google Earth Engine Developers

Cover Type
MODIS GEE Catalog Link:
https://developers.google.com/earth-engine/datasets
NASA’s Applied Remote Sensing Training Program /catalog/modis 15
Available Satellite Sensor Data in GEE: Sentinel-1 SAR

• Pre-processed SAR data from

Sentinel-1 is available to work with
in GEE.
• Previous ARSET training:
– Forest Mapping and Monitoring
with SAR Data
• GEE Data Catalog:
– https://developers.google.com
/earth-
engine/datasets/catalog/COP
ERNICUS_S1_GRD

A previous ARSET training details the use of SAR data for land
management, particularly in vegetated ecosystems.
Link: ARSET

NASA’s Applied Remote Sensing Training Program 16

Available Satellite Sensor Data in GEE: Land Cover Products

• A variety of land cover data

products are available in GEE,
including:
– Copernicus Global Land Cover
Layers
– MODIS Land Cover Type Yearly
Global 500m
– Global PALSAR-2/PALSAR
Forest/Non-Forest Map
– USGS National Land Cover
Database
Copernicus Global Land Cover Layers: CGLS-LC100 collection 3 displayed globally in
• GEE Data Catalog: GEE. Credit: Earth Engine Data Catalog

– https://developers.google.com/
earth-
engine/datasets/tags/landcover

NASA’s Applied Remote Sensing Training Program 17

Applications of GEE for Land Management: Burn Severity

• Burn severity mapping completed in GEE

manipulates pre-loaded Sentinel-2 or
Landsat 8 data and uses the GEE platform
as a means to quality control and filter
data.
• Normalized Burn Ratio (NBR) and
differenced NBR (dNBR) are calculated.
• Thresholding rates the severity of wildfire
burning to complete a full burn severity
assessment.
• Refer to the step-by-step UN-SPIDER burn
severity in GEE training

Example of burn severity mapping using Sentinel-2 data

in Empedrado, Chile in February 2017. This map was
produced using the UN-SPIDER Burn Severity with GEE
script. Credit: UN-SPIDER
NASA’s Applied Remote Sensing Training Program 18
Applications of GEE for Land Management: Mangrove Mapping

• ARSET training using GEE to map

mangroves:
– Remote Sensing for Mangroves in
Support of the UN Sustainable
Development Goals
• Random Forest Classification to create
a time series for mangrove extent
change
• Creation of apps as management and
Example of mangrove extent mapping over a time series to track how
communication tools mangrove presence has change over a 20-year period (from the exercises
completed during the ARSET Mangrove Mapping series).
• The previously mentioned ARSET SAR Image Credit: NASA ARSET
training also shows methods of using
Visit the Data Explorer and Comparison Apps
SAR data in GEE to map mangroves.
for more information on mangrove mapping
and apps in GEE.

NASA’s Applied Remote Sensing Training Program 19

Applications of GEE for Land Management: Land Cover
COUNTIES
• Public health officials and vector biologists
at the Maine Medical Center were Cumberland
interested in identifying lands within All other
Cumberland County, Maine that counties
represent forest edge habitat.
• These habitats are targeted for vector
borne disease mitigation since they
represent a higher risk of tick-human
encounters and thus exposure to illnesses
like Lyme Disease.
• The DEVELOP Massachusetts team
completed a supervised land cover
classification for the the county and then
identified borders between forested and
developed cover types to identify edge 0 40 80 160 240 320
habitat. Miles
Map of the State of Maine, highlighting the study
area of Cumberland County.
Image Credit: NASA DEVELOP
NASA’s Applied Remote Sensing Training Program 20
Applications of GEE for Land Management: Land Cover

Urban Development
Urban Edge
Forest Cover Forest Cover
Forest Edge Urban Cover
Cultivated Forest – Urban Edge
0 5 10 20 30 40
Miles Forest – Urban Edge
NASA’s Applied Remote Sensing Training Program 21
Applications of GEE for Land Management: Vegetation Indices

• Biodiversity conservation
organizations working in the
Visayan Islands of the Philippines
are concerned about the impact Philippines
of habitat degradation on
endangered species.
• Vegetation decline is associated
with habitat loss of two particular
species: the Visayan warty pig and Negros Island

the Visayan spotted deer. Panay Island

Visayan Islands
• The DEVELOP Georgia team used
the Terra MODIS archive in GEE
from 2000-2019 to map changes in 0 25 50 75 100
km
vegetation and identify areas with
critical vegetation loss.

NASA’s Applied Remote Sensing Training Program 22

Applications of GEE for Land Management: Vegetation Indices
NDVI Change EVI Change

NDVI EVI
Increase Increase

No Change No Change

NDVI EVI
Decrease Decrease

N
0 10 20 30 40
Kilometers
NASA’s Applied Remote Sensing Training Program 23
 Basics Functions in GEE JavaScript API Activity
https://code.earthengine.google.com/00b566d9ba821f7fb3011468533bccc9
 Python API Demonstration
https://colab.research.google.com/github/google/earthengine-
api/blob/master/python/examples/ipynb/ee-api-colab-setup.ipynb
Summary
• The GEE platform provides users with cloud-based computing resources
that can decrease barriers like data storage space and personal
computing power.
• The capabilities of GEE are similar to those of many GIS platforms used
to manipulate satellite data for key land-related remote sensing
processes, including algorithm application and land cover classification.
• GEE hosts many datasets relevant to land monitoring.
– Landsat Series, MODIS, Sentinel-2, and Sentinel-1 SAR
• The JavaScript API enables coding and automation of basic remote
sensing functions like imagery filtering, and vegetation index
calculation.
• Users interested in the Python API can explore Google Colab.
• Session 2: Land Cover Classification & Accuracy Assessment

NASA’s Applied Remote Sensing Training Program 26

Contacts
• Trainers:
– Zach Bengtsson: [email protected] Follow us on Twitter
– Juan Torres-Pérez: [email protected] @NASAARSET

– Amber McCullum: [email protected]

• Training Webpage:
– https://appliedsciences.nasa.gov/join-mission/training/english/arset-
using-google-earth-engine-land-monitoring-applications
• ARSET Website:
– https://appliedsciences.nasa.gov/what-we-do/capacity-
building/arset

NASA’s Applied Remote Sensing Training Program 27

 Thank You!

NASA’s Applied Remote Sensing Training Program 28