Temperature data is inherently error prone, with reliance on the accuracy of the sensors and continuous readings that fall within expected ranges. The Decision Aid System (DAS) is an online tool for pesticide management and recommendation that relies heavily on these temperature readings for horticulture models in order to correctly identify and provide diagnosis. It is essential that quality control is applied to these temperature readings to ensure that the real time recommendations by these models correctly reflect the current environment being read by these weather stations. In this paper, we develop and evaluate approaches to estimating the parameters of a quality control system based on historical data of a weather station. Through cross-validation, stations are evaluated on varying years of input data in order to determine the necessary amount of historical data to successfully integrate a new station into the system. Our results suggest that each parameter varies for the necessary amount of prior data to effectively apply quality control on future data, but never exceeds 5 years.
This repository holds the code use to develop methods for automated quality control and estimates parameters for weather stations, discussed in the above paper. The code was developed privately, and now made public upon completion of the project. No data can be shared due to privacy agreements, but code does not reveal any information regarding what is protected.