Food production will need to increase in Sub-Saharan Africa, as its’ population will grow in the coming decades. Smallholder agriculture is seen as a primary source of food for local markets. However, currently, it is unknown where most of the smallholder (irrigated) agriculture takes place, and the available sources are often outdated or have gaps.
The easiest way to know where irrigation takes place is by going into the field. However, this can be costly, take time, and an area might be too big to do this economically, which is why remote sensing is interesting. We can use satellite images and field reference data to train a model that classifies irrigated agriculture in areas that we have not physically visited.
It is anything but as straightforward as this might sound. From field data collection to algorithm parameters and accuracy assessments, each step requires different choices, assumptions, and considerations, which all influence the result – the map – differently.
This Ph.D. research will identify the most appropriate choices for classifying irrigated agriculture and how they can be documented as transparently as possible. This research will also determine the consequences on the visibility of smallholder irrigated areas. This research will look at:
• How recent studies have mapped irrigated agriculture with RS to determine how transparent and useful they are, using a flow diagram developed explicitly for this;
• What the influence of mislabelling and decreasing the amount of training data is on the classified results;
• How irrigated agriculture can best be classified for large regions, with limited training data; and
• Best practices for mapping irrigation.
The goal of gathering this knowledge is to provide tools and insights into how irrigation can be classified more accurately and usefully. This knowledge will help local institutes make maps and base informed decisions on them. By knowing where irrigated agriculture occurs, the policy can be better formulated for the different regions. Consequently, better estimates can be made about water use and food production in space and time.
This project will be implemented as a Ph.D. and closely work together with the FASIMO project and staff. MSc and BSc students also have the opportunity to contribute through their thesis or internship.