In order to generate variables potentially more likely to represent conditions
favorable for the development of gray leaf spot, weather data were summarized for
four periods during the growing season. Predictive models developed for other
Cercospora diseases use index values based on cumulative hours of RH above a
critical value while temperature is within a critical range (9, 56). Based on these
models, similar indices were derived for gray leaf spot. Temperature, relative
humidity, and leaf wetness were examined for the following periods: 1) 45 days
before R1 until 15 days after R1; 2) 15 days before until 15 days after R1; 3) 30 days
before R1 until R1; and 4) 45 days before R1 until 15 days before to R1 (Figure
3.1 B). These periods were chosen because they correspond to critical primary
infection periods and are relevant to the timing of fungicide application decisions. For
each period, cumulative hours of leaf wetness, temperatures between 22 and 30°C,
and RH & 90 and & 95% were generated. Since previous reports on the relationship
between temperature and RH within these ranges and gray leaf spot severity
showed that individually these variables did not tend to have a significant effect on
disease severity (4), different forms of representing these variables were explored
and their relationships with gray leaf spot severity were reassessed. Cumulative
hours of daily (600 to 1800 h) and nightly (1800 to 600 h), and mean daily and night
temperature and RH within the abovementioned ranges were generated for each of
the four periods. Cumulative daily and nightly time-duration values (TDV) (9), defined as the number of hours having both temperatures between 22 and 30°C and
RH & 90%, were also derived for each period. Bhatia and Munkvold (4) showed that
the strength of the relationship between the environment and gray leaf spot severity
improved when TDV was used as an input variable in regression models instead of
temperature and RH as individual variables. A total of 76 weather-related variables
were created and analyzed for their usefulness as input variables for model
development.