Performance of Leaf Wetness Sensors for Applicability In Decision-support Systems for Management of Citrus, Blueberry, and Strawberry Diseases

Abstract

Disease Alert Systems (DAS) in the Agroclimate decision-support system provide site-specific information to aid citrus,
blueberry, and strawberry growers to decide when a fungicide application would be required. All of these DAS use
disease models based on temperature and leaf wetness (LW) duration data to predict when weather conditions are
favorable for disease development and control measures are needed. Daily environmental data are obtained from the
weather stations of the Florida Agricultural Weather Network (FAWN). Previous research has shown that the electrical resistance-based Campbell 237-L leaf wetness (LW) sensors provide reliable data. However, they require painting
and in-situ calibration, which is not easily done by growers. Conversely, Decagon LW dielectric sensors come ready to
use by the manufacturer, with pre-established thresholds for wet and dry conditions. However, their performance in
the field is uncertain. We compared the LW estimations provided by Campbell 237-L and Decagon dielectric sensors
installed in the same station in Plant City, Florida. We performed comparisons of every sensor combination using
15-minute observations and maximum daily LW duration. The sensors of the same manufacturer had high (> 0.90)
Pearson’s correlation coefficient (Pc), low (< 1.0) mean absolute error (MAE), and high k agreement indices (> 0.9),
which indicate a strong correlation. However, when comparing Campbell and Decagon sensors, the precision was
lower as indicated by Pc of approximately 0.8, MAE around 2.0 hours, and k-indices around 0.8. Nevertheless, the
estimations MAE were within the acceptable range for DAS applicability. Decagon dielectric sensors could be used in
the FAWN weather stations to provide reliable LW estimations.