How Much Is Enough? Data Requirements for Practical Irrigation Decision-Making in Vietnamese Coffee Production

In making irrigation decisions, farmers typically rely on local observation and experience, such as observing crops and neighbors' actions. Research has mainly focused on understanding crop water requirements to improve farming practices, but it is important to acknowledge that farmers have unique perspectives, access to diverse local "signals", and experience. The challenge is to strike a balance between complex technical assessments of field conditions (the science) and harnessing farmers' skills to manage their irrigation in ways that maximize yield and quality. This study established a basis for specifying minimum data requirements for pragmatic, but not necessarily perfect, irrigation decision-making for small-scale Vietnamese coffee farmers. This study focuses on three areas in Dak Lak province in the Central Highlands of Vietnam. To explore the role of monitoring in irrigation management, two contrasting monitoring systems were set up to collect soil, weather, and irrigation data. We also compared a variety of water balance models with different data requirements, with a focus on processes that used "passive data collection", i.e., farmers do not manually collect data, rather data can be accessed readily from external sources. In Vietnam, traditional hosepipe irrigation is applied where it is impractical to know the volume of applied water. The proposed Low Data Model (LDM) is suited to more informed irrigation scheduling decisions, which have potential to improve the likelihood of coffee growers adopting measurement-based decision-making. While researchers may seek a detailed daily sub-millimeter understanding of soil water dynamics, farmers require practical decision support if there is to be any adoption of improved methods. This study offers a simple and practical approach for irrigation scheduling rather than a model using numerically perfect data that is unachievable in the field. The work demonstrates that on-site rainfall data is essential. However, other data can be collected passively to reduce the burden of data collection on users. This approach may enhance the likelihood of model-based irrigation scheduling being adopted by coffee farmers in Vietnam.