Optimization of a leaf wetness duration model

Abstract

Characterization of LWD (Leaf Wetness Duration) is essential when simulating the spread of phytopathogenic fungi. No standard for leaf wetness measurement exists and a wide range of sensors and models are currently available for estimating LWD. While the vast majority of models use hourly climate data (temperature, relative humidity, etc.) this time step is not widely available in historical records. It is also seldom available in spatially disaggregated projected (21st century) climatic data. Our study aims to develop a new method for using temperature and humidity on a daily time step to simulate leaf wetness duration. The principle behind our model consists in reconstructing hourly temperature and relative humidity based on optimized equations with the bootstrap statistical method for the studied area. Then hourly simulated data are used to estimate the daily duration of leaf wetness. The sensitivity of this process to daily and hourly temperature and relative humidity is assessed. The impact of this model on simulations of potential infections of grapevine (Vitis vinifera L.) by powdery mildew, using a key equation within a mechanistic model that simulates the pathogen's (Erysiphe necator) primary contaminations was evaluated. Results show that the error in simulating LWD and its impact on the potential intensity of the primary contamination of powdery mildew with daily climate data is low. This study has enabled us to set up an optimized modelling chain, which could be readily reproduced in different regions provided existing observational datasets enable suitable parameterization.