Can farmers’ weather stations augment reference weather network for agro-meteorological mapping? Application to potato late blight in Wallonia

Details

Supervisors

Bogaert, Patrick

Faculty

Faculté des bioingénieurs

Degree label

Master [120] : bioingénieur en sciences et technologies de l'environnement, à finalité spécialisée

Abstract

enTo meet the 2009 European Directive on the sustainable use of pesticides, Wallonia has set the goal of reducing its pesticide use and associated risks by 50% by 2030. Specific efforts are needed for potato crops, which account for the highest pesticide use in the region, primarily to control late blight. Since late blight is a fungal disease highly sensitive to weather conditions, accurate field-scale weather monitoring is essential for its early detection and effective control. Late blight forecasts rely on epidemiological models, such as the Late Blight Model (LBM), which use hourly weather data and provide advice to farmers through Decision Support Systems (DSS). Currently, these models depend on weather data from the Pameseb network, which consists of reference weather stations sparsely distributed on a 30 km grid across Wallonia. To improve spatial resolution and better reflect field-scale weather conditions, this study evaluate the local accuracy of the Agromet spatial weather interpolation, which interpolates Pameseb data with a 1 km2 resolution across Wallonia based on kriging methods. The potential of low-cost agricultural weather stations (called Farmers'Weather Stations, FWS) to densify the reference weather network is further assessed in the aim to enhance the weather interpolation field-scale accuracy for DSS applications. Results show that, while Agromet accurately estimates air temperature, the interpolation lacks accuracy in relative humidity under specific microclimatic conditions, reducing the precision of late blight infection forecasts. FWS are showed to be reliable data sources, despite a consistent 3% bias in relative humidity, for which a positive correlation was observed between several FWS. Therefore, integrating FWS data into the Agromet interpolation could be useful, by assigning them lower weights than reference data in the kriging interpolation to take into account their relative uncertainty. This FWS integration could improve the local accuracy of the interpolation and the late blight forecasts precision, enabling more targeted and timely pesticide applications to achieve Wallonia's pesticide reduction goal without compromising yields. Further research is needed to assess the spatial variability of FWS bias and to identify optimal locations for FWS deployment in order to capture field-scale weather conditions that are currently missed by the Agromet interpolation.