Design of a 2-D Lattice Flower Constellation for Earth observation applying the twin satellite concept

Abstract: Events such as forest fires or floods are a danger to our Earth’s environment and the people living in it. The sooner they can be detected, the less damage they can cause. An idea arises: use satellites to monitor the Earth and relay information to prevention and rescue organizations in a very short time, regardless of accessibility from ground. PandionAI is a Swedish start-up that aims to create a constellation of satellites that will carry built-in machine learning models that analyze the Earth images and detect early on such environmental catastrophes. The aim of this thesis is to design a constellation for such purpose. In parallel, a target list that includes potential places of interest for such events will be developed and the constellation performance will be evaluated with it by defining Key Performance Indicators (KPIs). The target list has a dynamic nature as weather changes and climate evolves, so events of interest will be at different places in the globe as time passes. Because of this, four different target lists were developed, one per season. The events covered by the target lists are the environmental hazards: fires, floods, deforestation and CO2 footprint from factories. Constellation design is a highly complex problem due to its infinite possible solutions for a given purpose. Nevertheless, one can optimize for specific design parameters and assumptions. This thesis proposes a model in which for a given number of total satellites, multiple constellations are compared, and the search for the optimal design is performed given the defined KPIs. Constellations in the optimization process are designed applying the 2-D Lattice Flower Constellation method and adding the twin satellite concept.