Regionalization of outputs of two crop protection models using geostatistical tools and NOAA-AVHRR images

Crop protection forecasting models currently use meteorological data observed at stations to produce pest infection and development indices. The indices are then extrapolated to the regional level by assuming that the weather conditions at the stations are similar to those in neighbouring fields in the region, which is not necessarily the case. Hence, this has a significant impact on the quality of the recommendations and diagnoses based on computerized plant protection models. The regionalization of model outputs between the stations comprising the weather network, using geostatistical techniques such as cokriging in conjunction with satellite data, is a worthwhile approach for addressing this need. The objective of this study is to develop and apply a methodology for regionalization of infection indices produced by two crop protection models contained in the CIPRA (Computer Centre for Agricultural Pest Forecasting) system, using geostatistical tools and NOAA-AVHRR images. This approach will help enhance our crop pest management and forecasting capabilities while optimizing the use of pest control products in vegetable crops in Quebec. To achieve our objective, a cokriging method was applied to regionalize the model outputs using air temperature and relative humidity estimated from NOAA-AVHRR images. The results were then validated against a regionalization approach using ordinary kriging and two conventional interpolation techniques.