Schedule Optimization for a Heterogeneous Earth Observation Constellation

Spacecraft operations require finding the right compromise between utilizing and preserving their available resources in the pursuit of contributing as efficiently as possible to their declared mission goals. To achieve this balance, key resources such as data storage and downlink, power, thermal range, payload, and stakeholder interests need to be carefully managed. Managing the approach to the scheduling of operations for a single spacecraft has shown to exceed the capacity of human operators. Moreover, the challenges increase significantly when multiple assets are involved in a constellation, leading to new use cases and requirements. The goal of this paper is to define and implement an optimization approach to the scheduling of a heterogeneous Earth observation satellite constellation accounting for both mission success criteria as well as cost-effectiveness of the operations, while also taking into account all other relevant resource constraints. This approach was developed and implemented in preparation for the integration of the FOREST-2 satellite mission developed by OroraTech into an expanding constellation of satellites used for detecting wildfires. The algorithmic approach is based on the strategy used during the operational phase of the FOREST-1 mission, which served as a predecessor to FOREST-2. Special consideration was given to how changes in the number of satellites and orbital planes, together with the manner in which they are placed on them, impact overall constellation performance, especially with regard to measures such as continuous coverage and maximum revisit times. The state of the art of commonly used approaches for the scheduling of constellations of Earth observation satellites is presented and evaluated with respect to their suitability for the task at hand, and compared to the proposed approach. The model description and implementation into the scheduling workflow are introduced and the scheduling results for a selection of constellation design scenarios are presented in order to assess and validate the optimization efforts of this work. These test cases range from two single satellites to one, two, and four planes with eight satellites each, to the case of eighteen satellites on as many orbital planes. The average resource allocation has been proven to show a full allocation of the available data budget for each case apart from the two-satellite example. Further, the data budget could be identified as the limiting factor, with average power budget allocations in the range of 20-35%. It has been demonstrated that the fulfilment rate of imaging orders is significantly improved by first sizing each plane for continuous ground coverage, and then adding more orbital planes with the same minimum number of satellites in each, compared to randomly assembling a constellation of individual satellites.