Agriculture faces unprecedented challenges in a rapidly evolving world, defined by expanding global populations and climatic uncertainties. This necessitates the maximisation of yields, while remaining as environmentally friendly as possible. In addressing these complex obstacles, the implementation of AI within the agricultural domain has emerged as a pivotal factor. One approach is to utilise AIs capacity to handle large volumes of data and generate plans that achieve specific objectives based on multiple factors. HTN planning, a well-established AI planning technique, proves effective in generating efficient plans for real-world situations. This study commences with systematical analysing the domain of agriculture, by looking at irrigation, fertilizing and pest management. We explore how uncertainty, in the form of weather events, affects the irrigation planning. To implement this uncertainty, we use risk-aware HTN planning, which enables decision making based on a probability distribution of the cost of a action and a given risk attitude. We implement our model in JSHOP2 and evaluate it in terms of correctness, scalability and precision. The result is a model, that plans according to a given risk attitude in an efficient and sustainable way, by only using as much water as necessary to maximize the yield of a plant. Furthermore, it establishes a good foundation to expand upon it, with integrating multiple sources of uncertainty in the future.