Dali Llama| Automated Detection of Hazards
Awards & Nominations
Dali Llama has received the following awards and nominations. Way to go!
Automated Detection of Hazards
LLAMA - Leveled Detection for the Living Aquaculture and Monitoring of Harmful Algal Blooms
Summary
Government units grapple with the early detection of harmful algal blooms (HABs) leading to massive fish kill and lack of access to clean water for communities highly dependent on rivers and lakes for drinking or livelihood.While there are tools and satellite data available, currently there’s a lack of open-source data due to the insufficient attention it gets compared to other natural phenomena such as earthquakes or volcanic eruptions. Hence, we are developing a real-time web app readily available for the public to encourage a proactive approach among concerned individuals towards sustained clean water sanitation and life below water (SDGs 6 and 14).
How We Addressed This Challenge
LLAMA is an open-source web app that uses satellite data to detect and analyze harmful algal blooms (HABs) as well as the predicted movements of suspected algal blooms with computer vision and machine learning.
According to the Census of Fisheries, the vast majority of municipal fishing operations in the Philippines (1.752 million or 98.4%) were individual operations. The coastal areas that are able to identify and prevent its adverse implications give a better likelihood for the community to thrive and flourish.
How it Works
We use combined near-real-time (NRT) data of Fluorescence Line Height (FLH), Chlorophyll-a (Chl-a) levels, Colored Dissolved Organic Matter (CDOM), and other essential data for detecting HABs, from MODIS-Aqua, and SNPP-VIIRS.
By viewing the satellite map, local and health authorities can be informed and alerted that certain bodies of water might be already polluted by algae. To maintain a healthy aquatic ecosystem, ensuring the cleanliness of the water is essential.
Impact
Without early detection, or more so, treatment, the algae will grow which can lead to the disruption of an ecosystem due to eutrophication. Fish and other aquatic resources will die underneath the ocean. This occurrence has severe consequences to humans such as loss of livelihood, lack of food to eat, and lack of a clean source of water.
Societies extract vast quantities of water from rivers, lakes, wetlands, and oceans to supply the requirements of cities, farms, and industries. The information gathered by our monitoring program will let the users stay updated about the present condition of the bodies of water.
The goals of this project are:
- Mitigation of fish kill
- Avoidance of water contamination
- Protection of wildlife
In order to ensure:
- Sustainable living
- Livelihood of future generations
How We Developed This Project
Since the Philippines is surrounded by bodies of water, our team aims to create a satellite map to detect different areas where algae blooms might likely occur. One might think that growth of the algae blooms has little environmental impact to focus our attention to. But it is not, if we ought not to take action, it can disrupt our ecosystem.
According to Aqua Culture Alliance Org, a densely concentrated algal bloom can deplete oxygen in the water due to the high respiration rate of the algae (eutrophication), or by bacterial respiration during their decay. In effect, the fishes suffocate.
In addition, algae threaten humans who ingest filter-feeding shellfish like oysters and mussels. Biotoxins concentrate within the shellfish flesh, causing illnesses like paralytic shellfish poisoning (PSP), diarrhetic shellfish poisoning (DSP), and amnesic shellfish poisoning (ASP). Crabs that feed on shellfish can also become toxic.
While there are tools and satellite data available for HABs, currently there’s a lack of open-source data due to the insufficient attention it gets compared to other natural phenomenons such as earthquakes, flooding, or volcanic eruptions.
Thus, we are developing, LLAMA - an open-source web app that uses satellite data to detect and analyze HABs as well as the try to predict movements of suspected algal blooms with computer vision and machine learning. The coastal areas that are able to identify and prevent its adverse implications give a better likelihood for the community to thrive and flourish.
With our web app, we try to visualize where HABs might take place by combining all the necessary data readings that contributes to algal growth. We used fluorescent line height for phytoplanktons, chlorophyll-a concentration in lakes and coastal areas, particulate organic and inorganic carbon readings, as well as other data from NASA's OceanColor Web to try to determine if we can identify them. We used Google Cloud Vision for the detection with our custom-made HAB training data from various sources on the internet. Initially, we wanted to try using raw binary data to create a custom detection algorithm, but with our team's lack of experience with large scale datasets, we decided to stick with optical-based detection.
During the first hours of the hackathon, we first tried looking at the data with QGIS, Panoply and SeaDAS. With those images in mind, we then tried collecting more data that might influence the growth of algae in brackish and coastal waters, thus we were directed to the OceanColor Web Level-3 Browser, where almost all of our needed data for accurate visualization can be found.
We processed a sample image of Lake Erie, where algal blooms are regularly occurring, with computer vision using OpenCV and PyPlot.
Our initial webmap visualization can be seen at https://dalillama.co/environment, where all of our accumulated data were applied as datasets and tilesets to Mapbox.
How We Used Space Agency Data in This Project
We used NASA's ocean-related data from OceanColor Web. We also used the data from Worldview to help with our visualization. Landsat-8 imagery was also used to test our early model with Cloud Vision.
Project Demo
Presentation for LLAMA :https://docs.google.com/presentation/d/15L78Npx7gTz6aDtdVvPsZw0koiB8p7UE9NffdbrXrP8/
Project Code
Data & Resources
Data Used:
- Chlorophyll-a (https://oceancolor.gsfc.nasa.gov/)
- Normalized Fluorescence Line Height (https://oceancolor.gsfc.nasa.gov/)
- Sea Surface Temperature (https://oceancolor.gsfc.nasa.gov/)
- Organic and Inorganic Carbon Particulate (POC and PIC) (https://oceancolor.gsfc.nasa.gov/)
- Phytoplankton absorption at 443nm wavelength (https://oceancolor.gsfc.nasa.gov/)
- Remote surface reflectance at the green wavelength (https://oceancolor.gsfc.nasa.gov/)
Satellites Used:
- MODIS Aqua and Terra for near real-time (NRT) readings. (https://earthdata.nasa.gov/earth-observation-data/near-real-time)
- Suomi NPP - VIIRS for higher resolution data and cross referencing. (https://earthdata.nasa.gov/earth-observation-data/near-real-time)
- Landsat-8 for testing (https://landsat.gsfc.nasa.gov/landsat-data-continuity-mission/)
Socio-Economic Data:
- https://tribune.net.ph/index.php/2020/09/19/red-tide-spreads-to-e-visayas-tahong-capital/
- https://thefishsite.com/articles/algal-toxins-in-pond-aquaculture
- http://www.fao.org/fi/oldsite/FCP/en/phl/profile.htm