The field of Earth Observation is moving towards the development of more robust and efficient models for predicting scenarios with missing data. Recent studies have highlighted the importance of understanding the factors that influence the effectiveness of multi-source models, such as the nature of the task, the complementarity among data sources, and the model design. This has led to a shift towards more streamlined approaches that prioritize model complexity and the necessity of all collected data sources. Additionally, there is a growing trend towards the use of machine learning and cloud computing to build and deploy Earth Observation Systems. This has enabled the development of more accurate and reliable frameworks for predicting various environmental phenomena, such as soil nutrient levels and lunar crescent visibility. Notable papers in this area include:

• A study that presents a serverless framework for composing and deploying spatio-temporal Earth Observation Systems, enabling organizations to reuse and share available functions to compose multiple systems.
• A paper that leverages machine learning to predict soil nutrient levels without reliance on laboratory tests, laying a foundation for actionable insights to improve agricultural productivity in resource-constrained areas.
• A research that integrates astronomical data with machine learning to refine the prediction of lunar crescent visibility, achieving a high predictive accuracy and highlighting the potential for further enhancements in this area.