Validation of weather and leaf wetness forecasts for a lettuce downy mildew warning system

Abstract

Leaf wetness and weather forecasts were important inputs in a warning system developed previously for lettuce downy mildew in coastal California. To improve the predictive ability of this warning system, reliability of weather and leaf wetness forecasts were evaluated. Weather forecasts from two commercial companies were compared with weather data measured in or near commercial lettuce fields during 1994-1998, at several different locations in coastal California. Measured leaf wetness data were compared with forecasts by company A. and by a dew simulation model using weather forecasts of company B or measured weather data (except cloud cover) as inputs of the model. When forecast and observed weather data pooled over 24 h were compared, the forecasts appeared to be accurate, but when data were analyzed hourly, forecasts had large errors during early morning hours, which coincided with the critical period for infection by the pathogen. Correspondingly, both company A and the dew simulation model frequently gave incorrect leaf wetness forecasts during morning and evening hours although incorrect forecasts were rare from Pacific standard time 10:00 to 17:00 and 00:00 to 06:00. Weather and leaf wetness forecasts were generally better for the Salinas Valley than for the Santa Maria Valley. The errors in weather forecasts and the innate weaknesses in the dew simulation model might have resulted in the inaccuracy in leaf wetness forecasts. This, in turn, led to errors in warnings of downy mildew infection by the previous system.