Arizona State University researchers and scientists have teamed with Google to create a new layer in Google Earth 5.0. The layer lets anyone, anywhere, recommend places on Mars to photograph with ASU's THEMIS camera on NASA's Mars Odyssey orbiter.
THEMIS is the Thermal Emission Imaging System, a multiband infrared and visual camera designed at ASU by Dr. Philip Christensen. A Regents Professor of Geological Sciences in the School of Earth and Space Exploration, Christensen is THEMIS' principal investigator and also director of the Mars Space Flight Facility on the Tempe campus.
Late on a Wednesday afternoon, Dr. Christensen approached software engineers Christian Yates and Eric Engle to develop an application that would let the public participate in Mars exploration by enabling them to suggest places for THEMIS to photograph.
By Friday morning, the team had a working application, with the help of Drupal.
"We wanted to give the general public a way to suggest places on Mars for THEMIS to photograph," says Christensen. "Using the feature, people can recommend sites, and these recommendations go to mission scientists who will decide what areas THEMIS images. If a public suggestion matches what the researchers choose, we'll notify the person who suggested the site and let them see the image as soon as we do."
The challenge to the Mars Space Flight Facility engineers was to create a user-friendly application accessible from within Google Earth that engages the public and encourages interaction.
Additionally, the app had to withstand a potential onslaught of traffic from being embedded into millions of copies of Google Earth.
| View from Google Earth |
|---|
|
Most users will first encounter the "Suggest an Image" feature through the Google Earth application itself, where it resides in the Mars Gallery section. A click of a button loads the feature, and draws tracks on the planet's surface indicating where THEMIS will be able to take photos during the coming week.
Clicking on the track opens a "balloon" that lets the user add a comment, and their email address if they'd like to receive responses.
The Suggest an Image web site allows users a second method of using the layer. By registering with the site first, they can download a customized KML file to load into Google Earth that will let them enter suggestions without having to type their email address each time.
| The Web Site |
|---|
|
For users that started directly from the Mars Gallery, the web site allows access to their past suggestions, images that have been successfully captured, and their preferences.
The site is also responsible for updating users when the status of their place suggestions change. Users can receive notifications via email when their suggestions are uploaded to the Mars Odyssey spacecraft, when the data is downlinked back to Earth, and when an image is fully processed and available for viewing on the web site.
With the release of version 5 of Google Earth, a number of new opportunities opened up. Most importantly for us, Google introduced a Mars layer, where users can explore images from instruments aboard a variety of past and present Mars spacecraft, including THEMIS.
Google Earth 5 also brings flexibility to developers by allowing greater use of HTML, CSS and Javascript within the application, which opens up doors for interacting with external services.
| The Process |
|---|
|
Drupal sits at the heart of the Suggest an Image application, and interacts with a variety of services:
Store the data downloaded from the spacecraft, as well as geometry defining where Odyssey will be orbiting in the near future.
Query the databases to assemble KML files containing orbital tracks, and also determine when a user's suggestion has intersected with an image taken by THEMIS, returning data files to Drupal.
| A processed image |
|---|
|
A python web service written for the Mars Image Explorer (also a Drupal application) serves finished images in a variety of formats and sizes that users can download and share.
Fulfills several roles in the Suggest an Image application:
Drupal 6.10 was chosen as the core of the Suggest an Image service, and several contributed modules were used to complete the application a tight timeframe.
| Module | |
|---|---|
| Mars Suggest | A proprietary module created for this project that manages the user interface, modifies and serves KML files and interacts with other modules to perform notifications, etc. |
| CCK | Adds a Suggestion content type with fields for image status, ID, latitude and longitude. |
| Profile | Adds opt-in buttons to user pages for notifications. |
| Notifications & Messaging frameworks | Used to deliver notifications on suggestion status changes to users. |
| Lightbox2 | Allows users to zoom into full-size versions of finished THEMIS images. |
| Login Destination and LoginToboggan | Redirect users directly to their account pages on login, and allow immediate registration, respectively. |
| Panels2 | Customizes user's account pages, allowing for inclusion of instructions, embedding of custom views, and custom KML download links. |
| Views2 | Used to generate lists of finished images, users past suggestions, and to power generation of the front page "heat map" of user suggestion activity. |
| Service Links | Creates links for sharing finished images via a number of social networks. |
| Quota by Role | Allows limiting of the number of submissions from each user per week. Note that some functions of Quota by role had to be migrated to our Mars Suggest module in order to set specific date ranges on quotas. |
| Infrastructure |
|---|
|
The Suggest an Image application had to be able to perform under the load imposed by direct inclusion into the Google Earth Mars Gallery layer. To accomplish this, both hardware and software architecture was carefully designed.
On the hardware side, a multi-layered architecture was devised, using layer-7 hardware load balancers from Coyote Point systems to direct traffic to a number of Squid cache servers and Lighttpd web servers. Both the Squid and Lighttpd layers scale well horizontally. PHP performance is augmented with the Alternative PHP Cache (APC).
While PHP code (Drupal core, modules, themes, etc.) is kept local to the web servers for best performance and synced with Rsync, the Drupal /files directory is served from our NetApp filers via NFS. THEMIS images themselves are served via a proprietary Python web service from our large AFS file system.
On the software side it was necessary to devise a way to minimize the load to both Drupal's MySQL server and our Postgres data servers. To do this, Drupal was utilized to generate and package HTML, CSS and images into a KMZ file (essentially, a ZIP file) which is served up by the web server layer and cached by our Squid farm for a week at a time.
Encapsulating all of the components this way consolidates the number of connections necessary to render the application's interface in Google Earth to one per session. Additional calls to Drupal are only made when a user actually submits a suggestion.
While Google Earth 5.0 supports far more HTML, Javascript and CSS than its predecessor, some changes still need to be made to the application to optimize performance.
Since Google Earth does not support cookies, or even the POST method, an alternative way of submitting new suggestion nodes was devised, relying on keys rather than cookies or session ids.
Graphic design was also changed to reflect the limits of what types of Javascript and CSS could be implemented from within a KML file.
With the recent announcement of the Google Earth 5.0 browser plugin by our colleagues at Google, we'll be able to bring the Suggest an Image experience directly into Drupal.
We also have plans to increase the interactivity of the application by enabling a heatmap layer within Google Earth, and capturing the most popular areas each week with the THEMIS camera.
JAVA programming, KML generation and orbital geometry:
Eric Engle - eengle [at] mars [dot] asu [dot] edu
Drupal/Python development, architecture and UX design:
Christian Yates - cyates [at] mars [dot] asu [dot] edu
Graphic Design:
Chris "Scooby" Capages - ccapages [at] mars [dot] asu [dot] edu
