Development of a Technique for Improving Access to Farm Practice Information Based on Monitored Temperature Conditions

Abstract

The study designed and implemented a temperature monitoring system. The system generated timely an accurately farm practice information from monitored temperature data. This was with a view to providing critical information in advance to farmers for making proper agricultural farm management decisions. A system that monitored soil temperature at an interval of 10 minutes in a Metrological Station was set up. This comprised of a programmable interface controller (PIC) 18F452, temperature sensor, real time clock DS1307RTC and micro secure digital memory card with adapter in normal SD card application. The circuit was designed with Proteus Institute for Software Integrated Systems (ISIS) and the components were placed on the Vero board as laid out in the circuit diagram and then soldered. The firmware, written with C programming language, compiled, using Custom Computer Service Compiler (CCS C) for PIC. Subsequently, calibration was carried out and the readings from the device were validated and benchmarked with that of the standard analogue thermometer and automated data systems. Furthermore, the results were examined statistically using Auto-Regressive Integrated Moving Average (ARIMA) model with expert modeler, to predict the soil temperature and obtain farm practice information based on the recorded values. A Short Message Service (SMS) application interface was developed for Global System for Mobile Communication (GSM) module (SIM300) using visual C# programming language. The paired t-test from the analysis showed that there was no significant difference (p > 0.05) and there were positive correlations between the designed temperature system, a standard analog thermometer and an automated data logging system. From the family of the identified models, ARIMA (0,1,0) model was found to be the most adequate model that really captured the dependence in the series. The performance evaluations of the adopted model was carried out on the basis of correlation coefficient (R2) and Ljung-Box statistics with values of 0.95, 13.58 and 0.96, 19.75 respectively for soil temperature at 5 cm and 10 cm soil depths. The model was used for forecast from weeks 11- 35, 2013. The result of the analysis from the graph of the predicted soil temperature at 5 cm soil depth showed that between weeks 21-43 and 18-35 fell between 10ºC and 30ºC. Within these periods maize seeds could be planted. The developed technique would provide an improved access to farm practice information based on monitored soil temperature conditions and, thereby, bring about better decision making by related stakeholders.