Elsevier

Pedosphere

Volume 33, Issue 3, June 2023, Pages 479-495
Pedosphere

Multi-step ahead soil temperature forecasting at different depths based on meteorological data: Integrating resampling algorithms and machine learning models

https://doi.org/10.1016/j.pedsph.2022.06.056Get rights and content

Abstract

Direct soil temperature (ST) measurement is time-consuming and costly; thus, the use of simple and cost-effective machine learning (ML) tools is helpful. In this study, ML approaches, including KStar, instance-based K-nearest learning (IBK), and locally weighted learning (LWL), coupled with resampling algorithms of bagging (BA) and dagging (DA) (BA-IBK, BA-KStar, BA-LWL, DA-IBK, DA-KStar, and DA-LWL) were developed and tested for multi-step ahead (3, 6, and 9 d ahead) ST forecasting. In addition, a linear regression (LR) model was used as a benchmark to evaluate the results. A dataset was established, with daily ST time-series at 5 and 50 cm soil depths in a farmland as models’ output and meteorological data as models’ input, including mean (Tmean), minimum (Tmin), and maximum (Tmax) air temperatures, evaporation (Eva), sunshine hours (SSH), and solar radiation (SR), which were collected at Isfahan Synoptic Station (Iran) for 13 years (1992–2005). Six different input combination scenarios were selected based on Pearson's correlation coefficients between inputs and outputs and fed into the models. We used 70% of the data to train the models, with the remaining 30% used for model evaluation via multiple visual and quantitative metrics. Our findings showed that Tmean was the most effective input variable for ST forecasting in most of the developed models, while in some cases the combinations of variables, including Tmean and Tmax and Tmean, Tmax, Tmin, Eva, and SSH proved to be the best input combinations. Among the evaluated models, BA-KStar showed greater compatibility, while in most cases, BA-IBK and -LWL provided more accurate results, depending on soil depth. For the 5 cm soil depth, BA-KStar had superior performance (i.e., Nash-Sutcliffe efficiency (NSE) = 0.90, 0.87, and 0.85 for 3, 6, and 9 d ahead forecasting, respectively); for the 50 cm soil depth, DA-KStar outperformed the other models (i.e., NSE = 0.88, 0.89, and 0.89 for 3, 6, and 9 d ahead forecasting, respectively). The results confirmed that all hybrid models had higher prediction capabilities than the LR model.

Section snippets

INTRODUCTION

Soil temperature (ST) plays a key role in different ecosystems, by affecting processes such as hydrological response of the soil, accumulation and degradation of organic matter, plant growth, nutrient mineralization, carbon emissions, proper time of sowing, and even micro-organism activity (Brar et al., 1992, Peng et al., 2009; Hu et al., 2016; Citakoglu, 2017). The variation of ST can alter soil characteristics and accordingly has considerable environmental outcomes with the change in carbon

Study area

Isfahan Province is located in Central Iran. It is characterized by arid and semi-arid conditions and comprises 10% of Iranian deserts. The western parts of the province encompass mountains with a mild climate. The air temperature ranges between 40.6 and 10.6 °C, the mean annual temperature is about 16.7 °C, and the mean annual rainfall is only 117 mm. Isfahan Province lies between 30° 42ʹ to 34° 30ʹ N and 49° 36ʹ to 55° E and covers an area of about 107 027 km2 (Fig. 1), of which about 5 674

Machine learning model

Instance-based K-nearest learning (IBK) is a lazy algorithm based on the K-nearest neighbor (KNN) classifier, using the Euclidean metric as a function for measuring the distance between instances (Smusz et al., 2013). Over the years, several applications of the IBK algorithm can be found in literature, for example, classification of bioactive compounds (Smusz et al., 2013), imbalance handling methods using public binary imbalanced datasets (Zhang et al., 2019), classification of foot drop gait (

Most effective input variables

It was observed that the correlation between air temperature and ST (Table II) was more important than other variables such as Eva, SSH, and SR. However, the correlation between soil and air temperatures decreased with increasing soil depth, in agreement with previous studies (Salamene et al., 2010). The influences of Eva, SSH, and SR also decreased in the deeper layers. The r values between the input variables and output revealed that Tmean was the most effective input variable for forecasting

DISCUSSION

Apart from some agronomic variables like seed availability, equipment readiness, field dryness, and tillage capacity, the timing of field crop planting is partially dependent on optimal soil temperatures for seed germination and seedling emergence (Lindstrom et al., 1976). As direct soil measurement is time-consuming and empirical models are not sufficiently accurate, the results of this study provide the bases for ST forecasting at different depths (5 and 50 cm).

The different characteristics

CONCLUSIONS

This study successfully addressed the challenge of accurately forecasting ST at two different soil depths (5 and 50 cm) using six novel hybrid models, i.e., BA-IBK, BA-KStar, BA-LWL, DA-IBK, DA-KStar, and DA-LWL. The main objectives were to develop models with reasonable prediction power and to propose them as simple and promising tools for multi-step (up to 9 d ahead) ST forecasting using Isfahan Province, Iran as a case study. The modeling process, based on r values, showed that Tmean was the

ACKNOWLEDGEMENTS

The authors would like to thank Dr. Esmaeel Dodangeh for sharing the dataset. We also would like to extend our gratitude to two anonymous reviewers for their help in enhancing the quality of the paper during the review process.

REFERENCES (82)

  • Y Gao et al.

    Thermal performance improvement of a horizontal ground-coupled heat exchanger by rainwater harvest

    Energ Buildings

    (2016)
  • K Hazama et al.

    Covariance-based locally weighted partial least squares for high-performance adaptive modeling

    Chemometr Intell Lab Syst

    (2015)
  • A Joshuva et al.

    A lazy learning approach for condition monitoring of wind turbine blade using vibration signals and histogram features

    Measurement

    (2020)
  • S Kang et al.

    Predicting spatial and temporal patterns of soil temperature based on topography, surface cover and air temperature

    Forest Ecol Manag

    (2000)
  • K Khosravi et al.

    Bedload transport rate prediction: Application of novel hybrid data mining techniques

    J Hydrol

    (2020)
  • K Khosravi et al.

    Meteorological data mining and hybrid data-intelligence models for reference evaporation simulation: A case study in Iraq

    Comput Electron Agr

    (2019)
  • K Khosravi et al.

    Quantifying hourly suspended sediment load using data mining models: Case study of a glacierized Andean catchment in Chile

    J Hydrol

    (2018)
  • J R Li et al.

    Predicting and analyzing early wake-up associated gene expressions by integrating GWAS and eQTL studies

    Biochim Biophys Acta Mol Basis Dis

    (2018)
  • B C Liu et al.

    Study of heat and moisture transfer in soil with a dry surface layer

    Int J Heat Mass Transf

    (2005)
  • H Liu et al.

    Heuristic creation of deep rule ensemble through iterative expansion of feature space

    Inf Sci

    (2020)
  • A Melesse et al.

    Suspended sediment load prediction of river systems: An artificial neural network approach

    Agr Water Manage

    (2011)
  • J Nakanishi et al.

    Composite adaptive control with locally weighted statistical learning

    Neural Netw

    (2005)
  • A Onan et al.

    A multiobjective weighted voting ensemble classifier based on differential evolution algorithm for text sentiment classification

    Expert Syst Appl

    (2016)
  • A Onan et al.

    Ensemble of keyword extraction methods and classifiers in text classification

    Expert Syst Appl

    (2016)
  • O Ozgener et al.

    A practical approach to predict soil temperature variations for geothermal (ground) heat exchangers applications

    Int J Heat Mass Transf

    (2013)
  • S Palani et al.

    An ANN application for water quality forecasting

    Mar Pollut Bull

    (2008)
  • K I Paul et al.

    Soil temperature under forests: A simple model for predicting soil temperature under a range of forest types

    Agr Forest Meteorol

    (2004)
  • S S Peng et al.

    Temperature sensitivity of soil respiration in different ecosystems in China

    Soil Biol Biochem

    (2009)
  • F Plauborg

    Simple model for 10 cm soil temperature in different soils with short grass

    Eur J Agron

    (2002)
  • H R Pourghasemi et al.

    Performance assessment of individual and ensemble data-mining techniques for gully erosion modeling

    Sci Total Environ

    (2017)
  • S Ravikumar et al.

    Fault diagnosis of self-aligning troughing rollers in belt conveyor system using K-star algorithm

    Measurement

    (2019)
  • S Samadianfard et al.

    Forecasting soil temperature at multiple-depth with a hybrid artificial neural network model coupled-hybrid firefly optimizer algorithm

    Inf Process Agr

    (2018)
  • H Sanikhani et al.

    Non-tuned data intelligent model for soil temperature estimation: A new approach

    Geoderma

    (2018)
  • H Shigemori et al.

    Optimum quality design system for steel products through locally weighted regression model

    J Process Control

    (2011)
  • S Smusz et al.

    A multidimensional analysis of machine learning methods performance in the classification of bioactive compounds

    Chemometr Intell Lab Syst

    (2013)
  • B A Tama et al.

    An empirical comparison of classification techniques for next event prediction using business process event logs

    Expert Syst Appl

    (2019)
  • C C Wei

    Comparing lazy and eager learning models for water level forecasting in river-reservoir basins of inundation regions

    Environ Modell Softw

    (2015)
  • L Xing et al.

    Daily soil temperatures predictions for various climates in United States using data-driven model

    Energy

    (2018)
  • F Yucalar et al.

    Multiple-classifiers in software quality engineering: Combining predictors to improve software fault prediction ability

    Eng Sci Technol

    (2020)
  • M Zeynoddin et al.

    A reliable linear stochastic daily soil temperature forecast model

    Soil Till Res

    (2019)
  • M Zeynoddin et al.

    Development of a linear based stochastic model for daily soil temperature prediction: One step forward to sustainable agriculture

    Comput Electron Agr

    (2020)
  • View full text