Evaluating Different Leaf Wetness Duration Models and Their Influence On a Disease-Warning System For Strawberry.

Disease-warning system is an approach to improve agricultural sustainability, since it helps to rationalize agrochemicals sprays for plant protection, reducing the production costs for farmers, the hazards to health and environment, and the residues in the final product. Leaf wetness duration (LWD) is a common input in disease-warning systems; however it is usually modeled due to its lack of observation. In this study four different models were evaluated to estimate LWD: classification and regression tree (CART), dew point depression (DPD), number of hours with relative humidity equal or greater than 90% (NHRH≥90%), and Penman-Monteith (P-M). In addition, the impact of estimated LWD on the number and timing of sprays was also evaluated. The period analyzed was from November to March during the 2011/2012 and 2012/2013 Florida strawberry seasons. A weather station located in Balm – in the Gulf Coast of Florida – was used to record air temperature, relative humidity, wind speed, net radiation, rainfall, and leaf wetness duration.

For both strawberry seasons P-M model had the best performance, considering the following statistical indexes for seasons 2011/12 and 2012/13, respectively: ME = – 0.9 h and – 0.8 h,    MAE = 1.4 h and 1.6 h, RMSE = 2.2 h and 2.4 h, R² = 0.8 and 0.8, and EF = 76% and 74%. A t-test comparing daily estimated and observed LWD showed that both P-M and DPD estimations are not statistically different at 5% of significance from observed values.

In summary, the models ranking is: P-M, DPD, CART, and NHRH≥90%.

*Note: abbreviation of statistical indexes:

- EF = efficiency

- ME = mean error

- MAE = mean absolute error

- R² = coefficient of determination

- RMSE = root-mean-squared error