Root System Water Consumption Pattern Identification on Time Series Data.

Area of Science:

• Agricultural Engineering
• Data Science
• Environmental Science

Background:

• Soil and meteorological sensors are crucial in agriculture for optimizing resource use and minimizing environmental impact.
• Low-power networks facilitate data capture from these sensors, enabling precision agriculture.

Purpose of the Study:

• To enhance soil moisture prediction and irrigation systems using advanced time series analysis.
• To identify irrigation and consumption patterns through outlier detection and pattern recognition in sensor data.

Main Methods:

• Application of time series analysis for outlier detection and pattern recognition on soil moisture sensor data.
• Comparison of three novel algorithms against the existing detection technique.
• Evaluation of algorithm performance using precision metrics on testing sets.

Main Results:

• The Series Strings Comparison (SSC) algorithm achieved an average precision of 0.872 on testing sets.
• The proposed algorithms significantly reduced the number of false positives compared to the current system.
• The SSC algorithm demonstrated a substantial improvement over the existing system's precision of 0.348.

Conclusions:

• The developed time series analysis methods, particularly the SSC algorithm, offer a significant improvement for soil moisture prediction and irrigation systems.
• Accurate detection of outliers and patterns in sensor data leads to more efficient irrigation and reduced environmental impact.
• The findings support the integration of advanced data analysis techniques in smart farming for enhanced agricultural practices.