Using Artificial Intelligence to forecast harmful algal blooms in Lake Atitlan, Guatemala
Abstract
As the demand for freshwater intensifies, and the degradation of water bodies increases, projects that provide actionable information to water authorities when local water resources are threatened are needed. One such situation is in Lake Atitlan, where Harmful Algae Bloom (HAB) events drastically impact the availability of drinkable water by producing toxic or harmful effects on people and other living organisms. In this project we aimed to develop a prototype HAB Early Warning system which can be used to inform Guatemala authorities about upcoming HAB events. Lake Atitlan is a landmark of Guatemala's biodiversity and culture that has experienced recurring HABs over the last decade. Each HAB event last from several days to weeks in length leading to the loss of safe drinking water to local inhabitants who depend on the lake's water. This presentation shows the preliminary results of our novel research that leverages the power of artificial intelligence and was awarded one of the eleven National Geographic and Microsoft Artificial Intelligence (AI) for Earth innovation grants. The machine learning model with the largest diagnostic and predictive skill will be discussed, and examples of how a HAB event can be predicted will be given. Use of up to 25 weather and environmental variables (e.g., local wind, rainfall, runoff estimates, solar insolation, temperatures) will be discussed, along with physical explanations on relevance of predictive variables to identifying and predicting HAB events. This project is the first of its kind to provide forecasting capabilities for HABs in a tropical freshwater body experiencing intense anthropogenic pressures and rapid degradation of the natural ecosystem. Overall project goals include: (a) developed a semi-operational HAB Early Warning system that can be used by local water authorities and the public, and (b) expand such a system to other tropical lakes which regularly suffer from HAB events.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMIN51C..02F
- Keywords:
-
- 1640 Remote sensing;
- GLOBAL CHANGE;
- 1942 Machine learning;
- INFORMATICS;
- 4329 Sustainable development;
- NATURAL HAZARDS;
- 6620 Science policy;
- POLICY SCIENCES & PUBLIC ISSUES