Animal response to natural disasters can change over time. For instance, the shelters of some species may be removed or modified through human action or have artificial barriers imposed that hinder their escape from fires or floods. Nowadays, we have access to animal tracking and satellite based weather data that can help us to correlate and understand the evolution of animal response to these events.
We have developed a web application that takes two representative scenarios for studying the aforementioned correlation: high precipitation and fire intensity events. This weather based data taken from NASA repositories is combined with Movebank tracking data from different species of birds.
These two scenarios are examples of the possibilities this kind of study gives and can be considered an initial prototype of a larger project which may include: historical data, a wider variety of both land and marine animals tracks, and consideration of other extreme natural events. With this information it would be possible to protect biodiversity as well as taking advantage of animal movement and senses to detect certain kinds of danger.
Our web application consists of a single page that allows the user to select between high precipitation and fire scenarios. Once the scenario is selected through the radio-button, it is necessary to Submit the change to the server to upload the information on the screen.
The high precipitation scenario shows a map with an example of the evolution of intensity precipitation from June to November 2015 near the Gulf of Mexico. In order to visualize this evolution, the user has the possibility of selecting different days through a temporal slider located on the bottom of the map. Apart from the precipitation information, the map overlays the tracking data of different species of birds for the selected date. The information related to the precipitation data is plotted in a few tones of blue, with the lighter blue giving the lower precipitation intensity and the darker blue the higher intensity. The bird tracking data is presented as crosses (‘x’) with a different color for each of the individual birds tracked. Additional information about each bird (ID, date) can be found by hovering over it’s cross.
Changing the selection with the radio-button, the user can access the fire scenario. The intention was to present a similar scenario to the precipitation example, combining MODIS14A2 FireMask satellite data with animal tracking data available for the same period. However, due to the lack of time and complications relating to the coordinate transformation of the sinusoidal projection data, it was only possible to include the precipitation example in the application.
As mentioned before, the idea of this project is to provide a tool to investigate potential correlations from a wide variety of extreme weather data and land/marine animal tracking data, in order to find behavioral patterns that could aid in the protection of both animal and human lives.
We were inspired to choose this challenge due to a shared interest in animal conservation and the possibility of applying space data to this field.
The application was developed using Python (including Flask, Bokeh, netCDF4, pyhdf, pandas, xarray and other libraries), JavaScript and HTML and is deployed on Heroku.
We started by exploring the Movebank datasets in Jupyter notebooks using pandas, Bokeh and matplotlib to present and manipulate the data. After finding a study in an interesting geographical area with a suitable number of animals tracked, we developed the code to process the GPM_3IMERGDF Precipitation data before combining them in Bokeh and deploying with Flask. We then attempted to add the FireMask layer from the MODIS data, but faced difficulties mapping the Sinusoidal data with Bokeh after transforming to Mercator format and were unfortunately not able to finish this part of the development in time. However we were able to prepare the functions necessary to extract the MODIS data from it’s HDF format and transform to Mercator. We then developed the slider functionality and improved the user interface before deploying the app to Heroku.
The public link to our web application is: https://space4life.herokuapp.com/
We used NASA GPM_3IMERGDF Precipitation data and NASA MODIS14A2 FireMask data for the weather layers in the application.
The satellite based data was fundamental for our project giving us the possibility of performing the correlation between events such as high precipitations and fires which can be detected through Earth observation satellites and animal movements on the land.
Link to presentation: https://drive.google.com/file/d/1xVzHC0e0jawx_a7hnd79AS-aU2eR7sQR/view?usp=sharing
We have used the following two datasets from NASA DAACS to get the data related to the precipitation and fire intensities:
Animal tracking data was taken from the following study, accessed via Movebank. Note that the data from this study was used only as an example of the tool and should not be repurposed without consulting the authors and reviewing the licensing information for the study on Movebank: