NASA Ames Crisis Mapping Toolkit

The Crisis Mapping Toolkit is a collection of algorithms and utilities for creating maps in response to crisis. The CMT relies on Google Earth Engine (EE) for much of its data processing. The CMT is released under the Apache 2 license.

The CMT is developed by the NASA Ames Intelligent Robotics Group, with generous support from the Google Crisis Response Team and the Google Earth Engine Team.

The CMT currently provides:

• Algorithms to determine flood extent from MODIS data, such as multiple thresholding techniques, learned approaches, Dynamic Nearest Neighbor Search, and more.
• Algorithms to determine flood extent from SAR data, such as histogram thresholding and active contour.
• Algorithms to detect water and clouds in LANDSAT images.
• Various helpful utilities, such as:
  • An improved visualization UI, with a drop down menu of layers similar to the EE javascript playground.
  • Local EE image download and processing, for the occasional operation that cannot be done efficiently in EE.
  • A configurable domain specification to define problem domains and data sources in XML.
  • Functions for searching image catalogs.

The CMT is still under development, so expect changes and additional features. Please contact Brian Coltin (brian.j.coltin at nasa dot gov) with any questions.

Installation

• Follow the instructions for installing Google Earth Engine for Python.
• Download the CMT source code from Github.
• Install PyQt4.
• Install the CMT with

  python setup.py install
  

Documentation

Before calling any CMT function, you must initialize EE, either by calling ee.Initialize, or by using the cmt ee_authenticate package:

from cmt import ee_authenticate
ee_authenticate.initialize()

Using the CMT UI

To use the CMT UI, replace your import of the EE map client:

from ee.mapclient import centerMap, addToMap

with

from cmt.mapclient_qt import centerMap, addToMap

then use the centerMap and addToMap functions exactly as before. That's it!

Using the CMT LocalEEImage

See the documentation in local_ee_image.py. When you construct a LocalEEImage, the image is downloaded from EE with the specified scale and bounding box using getDownloadURL. You can then access individual pixels or bands as PIL Images. Images are cached locally so if you are testing on the same image you do not need to wait to download every time. We recommend using LocalEEImage sparingly, only for operations which cannot be performed through EE, as downloading the entire image is expensive in both time and bandwidth.

Data Access

Data used in our examples has been uploaded as Assets in Earth Engine and should be accessible without any special effort. Unfortunately some datasets, such as TerraSAR-X, cannot be uploaded. If you find any missing data sets, contact the project maintainers to see if we can get it uploaded.