Spatial and machine learning methods of satellite imagery analysis for Sustainable Development Goals

The United Nations (UN) and World Bank have set Sustainable Development Goals (SDGs), with the aim for countries to reach targets related to important aspects of quality of life by 2030. An essential element of sustainable development is achieving social and economic aims to improve human quality of life, while conserving and managing natural resources. Earth observation data, such as satellite imagery data, are increasingly being used for monitoring the SDGs, and statistical machine learning methods are commonly used to analyse these types of data. However, current methods often exclude the spatial information inherent in earth observation data, which can provide useful insights. In this paper we review how spatial information is currently measured for remote sensing data, describe spatial machine learning methods in the literature and opportunities for further development of spatial methods. We also describe a minimum set of requirements to measure SDGs from satellite imagery data