EO4EU's Environmental Pest Monitoring Use Case Showcased at the ESA EO4Africa Symposium
The latest developments for Use Case 6 - Environmental Pest Monitoring, specifically for the prediction of Desert Locust suitable ecosystem, were presented at the ESA EO4Africa Symposium held at ESRIN from September 21st to 24th 2024.
The Desert Locust is one of the most destructive insect species, posing a severe threat to food security in regions where agriculture is already fragile. Early action is essential to prevent locusts from reaching their final developmental stages, where they can form devastating swarms. Our model is designed to serve as an early warning system, offering the critical ability to predict future suitable ecosystems for these pests, allowing for timely and targeted intervention.
Model Overview
Our innovative approach combines a Recurrent Neural Network (RNN) with Maxent modeling, enabling it to analyze the temporal evolution of environmental variables and produce a probability of locust presence. The model is lightweight, efficient, and delivers fast inference results.
The input data used for the model are:
- Soil water content (ERA5-Land)
- Normalized Difference Vegetation Index (NDVI) from MODIS satellite data
- Skin temperature (ERA5-Land)
- Precipitation (ERA5-Land)
To capture the dynamic changes in locust-prone environments, the model processes a time series covering 50 days, with each time-step representing a five-day average. This setup allows the model to continuously track and analyze the evolving conditions that influence locust habitats.
Performance and Impact
The presented model is capable of generating high-probability forecasts, helping predict areas at risk of locust infestation. This information is valuable for governments, NGOs, and local communities to act early and mitigate the potentially devastating effects on agriculture and food security.
The participation in the EO4Africa Symposium has allowed us to showcase the promising capabilities of this model and its potential to transform pest monitoring and management in vulnerable regions.
