Comparison of measured and modelled wetness duration as inputs to a grape disease model

Measured surface wetness duration is often used in disease risk prediction models, but is only available from a few weather stations. Wetness can be modelled from more widely available weather station networks using other meteorological variables. This study compared wetness duration measured using different methods of interpreting wetness sensor output, and from different sensor types, with wetness calculated from a classification and regression tree (CART) model. The model calculated wetness from temperature, relative humidity and wind speed. Different wetness sensors and different wetness calculation methods from the same sensor made little difference to recorded wetness duration. Total wetness duration was greater for modelled than for measured wetness at all but one of seven sites investigated. The use of modelled and measured wetness inputs into a grape botrytis prediction model indicated that modelled wetness is unsuitable for use in New Zealand without being calibrated for local conditions.