Google Earth Engine (GEE) compositing

Make a Best Available Pixel (BAP) composite using GEE's Python API in a User Interface within a Jupyter Notebook.

Requirements

1. Google Earth Engine account
2. Python 3+ (3.6+ preferred)
3. Jupyter:
   • In Windows you can use Anaconda
   • In Linux it will be installed with the python packages needed

Install

• Windows

  • Install Anaconda
  • Open Anaconda Prompt (search for it in Programs Menu)
  • (OPTIONAL) Create a new environment: link
  • Use pip to install this package:

    pip install geecomposite

  • If this is the first time you use GEE's Python API you have to authenticate:

    earthengine authenticate

  • (JUST IN CASE) Enable ipyleaflet:

    jupyter nbextension enable --py --sys-prefix ipyleaflet

  • open a notebook

    jupyter notebook

  • open this file

• Linux

  • Open a terminal
  • (OPTIONAL) Create a new environment: link
  • Use pip to install this package:

    pip install geecomposite

  • If this is the first time you use GEE's Python API you have to authenticate:

    earthengine authenticate

  • (JUST IN CASE) Enable ipyleaflet:

    jupyter nbextension enable --py --sys-prefix ipyleaflet

  • open a notebook

    jupyter notebook

  • open this file

Usage

Running all cells in bap.ipynb will get you to the User Interface:

import ee
ee.Initialize()
import ipygee as ui
import geecomposite

bapwidget = geecomposite.Bap()
bapwidget

• File Manager
  • Once you have set the parameters for creating a composite, you can save them using the Save Tab. And if you have already saved one, you can open it using the Open Tab. Configuration files are stored in json format
• Site
  • Here you can choose an EE Asset using its path. Once you write the assetPath click on Fetch Properties, then choose the property that segregates your sites, click on Fetch Options and finally select the site.
• Collections
  • choose one from: Landsat 8, 7, 5, 4 and/or Sentinel 2
• Season
  • Choose the day and month for the start and end of the season
• Year
  • if the season goes over the end of the year, then:
    • startday-startmonth-year to endday-endmonth-year+1
    • example: 15-11-2010 to 15-02-2011
  • if not:
    • startday-startmonth-year to endday-endmonth-year
    • example: 15-06-2010 to 15-11-2010
  • Back & Forth: if these parameters are greater than zero then it will take images from back years back and forth years forth
• Clouds
  • Choose a cloud cover percentage to filter the collection
  • Choose whether to mask the clouds out or not
• Scores
  • Doy
    • Maximum score for a given "day of year"
  • Satellite
    • Maximum score for the best satellite in the given period
  • Cloud distance
    • Minimum score for the pixels next to clouds (mask in general)
  • Mask cover
    • Maximum score to the image with less masked pixels
  • Index
    • Maximum score for pixels with a given vegetation index (momentarily set to 0.8)
  • Outliers
    • Maximum score for pixels with values between mean+-std for the chosen bands
  • Medoid
    • Maximum score for pixels with the minimum euclidean distance to the median for the chosen bands
    • NOTE: to make a "medoid" composite use only this score
• Export
  • Section for exporting the result to an EE asset (to Drive not currently available directly)
• Map
  • Visualize in a map the composite resulted from the set parameters. You can use the chosen site (Add using site) or the map's bounds (Add using map bounds)

Alternatively you can get the BAP object

bap = bapwidget.BAP()

and also get the resulting composite

composite = bapwidget.composite()