Vision transformers to model land surface phenology

Description

Climate change is modifying different aspects of life on the planet Earth. The continuous rise in global temperatures and weather patterns affects the distribution of plants and the timing of their main biological events (e.g., leafing and blooming). Phenology is the science that studies the timing of these events and their deriving forces in association with environmental phenomena. This project aims to exploit the power of state-of-the-art Deep Learning (DL) algorithms to facilitate the process of understanding phenological events from geospatial data gathered by Earth Observation (EO) satellites.

You will develop models to learn the relationships between phenological events and open-access EO data time series. Considering the complexity and diverse structure of phenological data, the direct adaptation of the existing DL models for analyzing such data is challenging. This topic aims at carefully selecting and modifying the state-of-the-art Vision Transformers (ViT) [1] and applying a suitable data representation. Your model will be checked against convolutional neural networks (CNNs) as the de-facto DL algorithm for EO-based information extraction [1]. You will also compare your model's output with the land surface phenology maps produced based on secondary EO products [2].

By estimating the timing of phenological events over large areas, you will contribute to a better understanding of (future) phenological changes and analyze the impact of climate change on plants.

Dataset

The following data sources will be used in this research

• EO data
• Phenological observations from http://www.pep725.eu/

Workflow

The workflow of the MSc thesis will be as follows.

1. A literature review
2. Dataset generation
3. Model development, validation and tuning
4. Evaluation