Water is an essential element for existence of all life forms, and the finite nature of freshwater motivated the EU to establish the WFD to abate further degradation of watercourses and aid the member states to achieve good status of all watercourses. In England and Wales, the main reasons hindering improvements in the quality of watercourses are the diffuse sources of water pollution. Therefore in this research, we employ an IoT system embedded with turbidity sensors to detect soil-runoffs into watercourses. Considering that during the growing season these runoffs are usually accompanied by agrochemicals bounded to the soil particulates, a contamination detection algorithm was designed to detect contamination events and notify the users via email notification. Moreover, since the system is designed to determine the possibility of a contamination event and not quantitatively establish the amount of agrochemical loading, a self-designed low-cost IoT compatible water sampler with near real-time response feature was developed to capture relevant water samples. These water samples can then be examined in a laboratory environment to quantify agrochemical loading in the watercourse. System analysis performed brought to the conclusion that the system is not only cost-efficient but also power-efficient in comparison to the current on-line based multi-parametric systems. Preliminary results extracted from the controlled experiments show that the system is not only capable of identifying runoffs but is also robust to a plethora of misclassifications occurring from irrelevant events. Moreover, the system is also capable of collecting relevant water samples and issue email notification when the system decided on a successful contamination event. Since the email notification contains the source of the event origin it was concluded that the system may be utilised to identify the sources of diffuse pollution. In conclusion, when we compare the low-cost nature of the system and the benefits it offers, this research opens up the possibility of an extensive watercourse monitoring program with substantial spatial coverage. The study also exposed several limitations of the current research and possible future research directions. The future research directions include building a database of information on growing season cycle and agrochemical application schedules, energy harvesting techniques for the water sampler, unsupervised machine learning techniques to ascertain the underlying patterns in turbidity values during runoff events, extended software simulations and several extensions to the current IoT system.