Data Quality and Information Quality – the Case of the Negative Prognosis Plant Protection Model

Abstract

The Negative Prognosis model (NPM) is a meteorological plant protection model that has been used in the farming practice for a long time now for controlling potato late blight, caused by Phytophthora infestans. NPM takes hourly air temperature, relative humidity (RH) and precipitation sums as input, generating recommendations on the first protective treatment date. The objective of the present work was to determine the impact of weather data quality on the NPM recommendations quality. Sensitivity analysis was conducted through inspection of source code and simulation. Data for simulation were prepared so as to measure a possible delay of the recommended date of the first treatment. Simulation experiments were performed to assess the recommendations quality for the direct-measurement data and the prepared data. It was established that the recommendations depend on the measurement precision of the RH near the value of 87% and air temperature near the values of 10, 12, 14, 16, 18 and 24 °C. The decisive factor is RH measurement precision. Small variations of RH of 2–3% may cause a serious miscalculation of the recommended date of the first treatment of 8-11 days.