Related Experiment Video
Updated: Jul 12, 2026
11:53
Spatial Multiobjective Optimization of Agricultural Conservation Practices using a SWAT Model and an Evolutionary Algorithm
Published on: December 9, 2012
12.9K
Coupling artificial neural network and sperm swarm optimization for soil temperature prediction at multiple depths
Milad Sharafi1, Mohammad Ali Ghorbani2, Rahim Barzegar3
1Department of Water Engineering, Urmia University, Urmia, Iran.
Environmental Science and Pollution Research International
|September 20, 2024
Summary
A new hybrid model, MLP-SSO, accurately predicts daily soil temperature at deeper levels. This advanced model significantly reduces prediction errors, especially at 100 cm depth, aiding irrigation and farming decisions in water-scarce regions.
Area of Science:
- Environmental Science
- Agricultural Engineering
- Data Science
Background:
- Soil temperature (ST) is crucial for water resource management and agricultural practices like planting and fertilization.
- Accurate ST prediction models are vital, particularly in water-scarce regions like Iran, and existing models show reduced errors at greater depths.
- There is a need for improved models capable of precise ST forecasting at lower depths.
Purpose of the Study:
- To develop and evaluate a novel hybrid model for accurate daily soil temperature prediction at various depths (5-100 cm).
- To enhance the prediction accuracy of the multilayer perceptron (MLP) model by integrating the Sperm Swarm Optimization Algorithm (MLP-SSO).
- To assess the model's performance in predicting soil temperature using meteorological data from synoptic stations in Iran.
Main Methods:
- Utilized multilayer perceptron (MLP) and a hybrid MLP-SSO model for daily soil temperature prediction.
- Input meteorological parameters included air temperature, relative humidity, wind speed, sunshine hours, and precipitation.
- Data were sourced from Ahvaz and Sabzevar synoptic stations in Iran, spanning from 1997 to 2022.
Main Results:
- The MLP-SSO model demonstrated superior performance compared to the standalone MLP model across all depths.
- The root mean square error (RMSE) was substantially reduced at a depth of 100 cm.
- At Ahvaz, RMSE decreased from 1.25°C (MLP) to 1.12°C (MLP-SSO); at Sabzevar, it dropped from 1.78°C to 1.49°C.
Conclusions:
- The hybrid MLP-SSO model significantly enhances the accuracy of daily soil temperature prediction, particularly at deeper soil layers.
- The findings underscore the effectiveness of the MLP-SSO model for improving soil temperature forecasting in agricultural contexts.
- This advanced model offers valuable insights for optimizing irrigation and fertilization strategies in water-limited environments.
