Skip to main content
ARS Home » Pacific West Area » Corvallis, Oregon » Forage Seed and Cereal Research Unit » Research » Publications at this Location » Publication #272220
Title: Sensitivity of disease management decision aids to temperature input errors associated with out-of-canopy and reduced time-resolution measurements

Author
item Pfender, William
item Gent, David - Dave
item Mahaffee, Walter - Walt

Submitted to: Plant Disease
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/7/2011
Publication Date: 5/1/2012
Citation: Pfender, W.F., Gent, D.H., Mahaffee, W.F. 2012. Sensitivity of disease management decision aids to temperature input errors associated with out-of-canopy and reduced time-resolution measurements. Plant Disease. 96:726-736.

Interpretive Summary: Decision tools to aid plant disease management decisions usually require inputs of weather data such as air temperature, but sometimes there is error associated with this weather data. We investigated errors introduced when temperature is measured only once per hour instead of every 15 minutes, and when the temperature sensor is located away from the crop instead of in the crop canopy. We studied the impact of these errors on performance of decision aids for grass stem rust and powdery mildew (grapes and hops). Decreasing time resolution from 15 min to 60 min resulted in marginally less extreme daily maximum and minimum temperatures which were important only when temperatures were near a threshold for the particular decision aid. Sensor location (in-canopy vs. standard-placement) had a larger effect than time resolution did on temperature observations. Temperature differences due to sensor location within grapes or hops were less than the location-related differences in the grass seed crop. In most cases the relatively small temperature measurement errors for grape and hops fields caused only minor differences in decision aid recommendations, unless actual temperatures were near a classification threshold. The decision aid for grass stem rust was affected by temperature input errors, which were larger than for hops and grape fields. Simple equations were devised to make corrections. These experiments and analyses use specific examples to illustrate how to determine the size of errors introduced by out-of-canopy or estimated temperature observations, and their impacts on decision aid performance. This information to makes it possible to correct for the measurement errors, and obtain more accurate predictions from disease management decision tools.

Technical Abstract: Plant disease management decision aids typically require inputs of weather elements such as air temperature. Whereas many disease models are created based on weather elements at the crop canopy, and with relatively fine time resolution, the decision aids commonly are implemented with hourly weather observations made from sensors sited at a standard placement of 1.5 m above ground vegetation or estimated for standard-placement from off-site measurements. We investigated temperature measurement errors introduced when time resolution was decreased from 15 minutes to 60 min, and in-canopy conditions were represented by observations collected by standard-placement meteorological stations, in three crops: grass seed, grapes and hops. Impact of these errors on performance of decision aids for grass stem rust and powdery mildew (grapes and hops) was assessed. Decreasing time resolution from 15 min to 60 min resulted in marginally less extreme daily maximum and minimum temperatures which were important only when temperatures were near a threshold for the particular decision aid. Sensor location (in-canopy vs. standard-placement) had a larger effect than time resolution did on temperature observations. Temperature differences due to sensor location within grapes or hops were less than the location-related differences in the grass seed crop. Effects of temperature observation errors on performance of the decision aids were affected by the magnitude of the errors as well as by the nature of the particular decision aid. The decision aids for powdery mildew are rule-based indices, and in most cases the relatively small temperature measurement errors caused only minor differences in rule outcomes and decision aid recommendations, unless actual temperatures were near a classification threshold. The decision aid for grass stem rust is a simulation model, and temperature input errors had quantitative (not threshold) effects on modeled pathogen growth and infection probability. Simple algorithms were devised to correct the temperature values or the computed infection probability so that outcomes similar to those driven by in-canopy temperature measurements were produced using standard-placement sensors. These experiments and analyses use specific decision aid examples to illustrate quantification of the uncertainty introduced by out-of-canopy or estimated temperature observations, and their impacts on decision aid performance. This information to makes it possible to correct for the measurement errors, and provides an estimate of uncertainty from this source that could be used in a subsequent analysis of relative importance among other sources of uncertainty in decision aid performance.