Predicting aphid (Lipaphis erysimi) growth in oilseed Brassica using near surface meteorological data from NOAA TOVS—a case study

Abstract

Pest infestation in crops is highly influenced by agrometeorological parameters. Weather based early warning of pest infestation is being practised using statistical and dynamic simulation models on point scale. Satellite based inputs and epidemiological models can extend the application to areas with irregular and non‐existing ground observations. The present study describes the use of National Oceanic and Atmospheric Administration TIROS (Television and InfraRed Operational Satellites) Operational Vertical Sounder (TOVS) near surface air temperature at 1430 h Local Apparent Time (LAT) over India (0900 Universal Time Code) for modelling onset, build‐up and peak aphid (Lipaphis erysimi) infestation on Indian Mustard (Brassica juncea) crop over Bharatpur and Kalyani, falling in semi‐arid and sub‐humid regions, respectively. The daily cumulative TOVS air temperature from 1 October 2001 showed high correlation (R 2: 0.99) with observed datasets. Exponential relationships were found to be the best empirical fit between TOVS cumulative air temperature (CATTOVS) and crop age at aphid onset (R 2: 0.7–0.99) and peak infestation (R 2: 0.91–0.95) for two stations representing semi‐arid (Bharatpur) and sub‐humid (Kalyani) agroclimatic conditions. Second order polynomial fits were found (R 2: 0.81–0.85) at both the stations between peak aphid count and CATTOVS at peak. Estimates of intermediate linear aphid build‐up to peak, computed using location‐specific linear growth rate (LGR) showed a higher standard error (SE) of 20% of mean at Kalyani (0–99), compared to 8% at Bharatpur (4–58). The common prediction models on linear start and peak were developed using TOVS noon time specific humidity (SPH) weighted thermal units and sowing dates. The standard error (SE) of estimated intermediate aphid‐build‐up became less: 4.5% of observed mean counts for pooled datasets with a common model, irrespective of sowing dates. The correlation between estimates and observations was 71%. The common model will be useful for general application in the absence of availability of local models.