Günlük Toprak Sıcaklığının Zaman Serisi Tahmini için Derin Öğrenme Yaklaşımı

Öz

Toprak sıcaklığı (ST), güneş enerjisi, tarım, hidroloji, jeoloji, tarımbilim ve çevre çalışmaları gibi birçok alanda önemli uygulamalara sahip kritik bir klimatolojik parametredir. ST'nin zaman serisi tahmini için doğru modellerin geliştirilmesi, özellikle tarımsal ve çevresel süreçlerin optimize edilmesi açısından büyük önem taşımaktadır. Bu çalışmada, saatlik toprak sıcaklığı tahmini için derin öğrenme yaklaşımlarından biri olan uzun-kısa vadeli bellek (LSTM) sinir ağı kullanılmıştır. Türkiye'nin farklı iklim bölgelerini temsil eden dört ölçüm istasyonundan (Adana, Ağrı, İzmir ve Ankara) 2016–2019 yılları arasındaki saatlik ST verileri analiz edilmiştir. 20 cm, 50 cm ve 100 cm derinliklerdeki ST değerleri, geçmiş ST verileri kullanılarak tahmin edilmiştir. LSTM modelinin performansını değerlendirmek için kök ortalama kare hata (RMSE), ortalama mutlak hata (MAE) ve korelasyon katsayısı (R) gibi istatistiksel ölçütler kullanılmıştır. Sonuçlar, tahmin doğruluğunun toprak derinliği arttıkça iyileştiğini göstermektedir. En düşük MAE (0,0385°C) değeri Ağrı'da 100 cm derinlikte elde edilirken, en düşük RMSE (0,0500°C) değeri ise aynı derinlikte Ankara'da gözlemlenmiş ve her iki durumda da en yüksek R değeri (0,9999) elde edilmiştir. Bulgular, LSTM yönteminin ST'nin zamansal desenlerini yakalamada son derece etkili olduğunu ve farklı iklim bölgelerinde başarıyla kullanılabileceğini doğrulamaktadır

Anahtar Kelimeler

• Derin öğrenme
• Uzun-kısa vadeli bellek (LSTM)
• Toprak sıcaklığı
• Zaman serisi tahmini

Deep Learning Approach for Time-Series Prediction of Daily Soil Temperature

Öz

Soil temperature (ST) is a crucial climatological parameter with significant applications in solar energy, agriculture, hydrology, geology, agronomy, and environmental studies. Developing accurate models for time-series prediction of ST is particularly important for optimizing agricultural and environmental processes. This study applies a long short-term memory (LSTM) neural network, a deep learning approach, for predicting hourly soil temperature. Hourly ST data from four measurement stations (Adana, Ağrı, İzmir, and Ankara), representing different climatic regions of Turkey, were analyzed over the period 2016–2019. ST values at depths of 20 cm, 50 cm, and 100 cm were estimated using historical ST data. To assess the performance of the LSTM model, statistical metrics such as root mean square error (RMSE), mean absolute error (MAE), and correlation coefficient (R) were used. Results indicate that prediction accuracy improves with increasing soil depth. The lowest MAE (0.0385°C) was recorded at 100 cm depth in Ağrı, while the lowest RMSE (0.0500°C) was observed in Ankara at the same depth, both with the highest R-value of 0.9999. The findings confirm that LSTM is highly effective in capturing the temporal patterns of ST and can be successfully employed in various climatic regions.

Anahtar Kelimeler

• Deep learning
• Long short-term memory (LSTM)
• Soil temperature
• Time series prediction

Kaynakça

1. Ballouch M., Akay F., Erdem S., Tartuk M., Nurdağ TF., Yurdagül HH. Forecasting call center arrivals using machine learning. Osmaniye Korkut Ata Üniversitesi Fen Bilimleri Enstitüsü Dergisi 2021; 4(1): 96-101.
2. Bilgili M. Prediction of soil temperature using regression and artificial neural network models. Meteorol Atmos Phys 2010; 110(1): 59-70.
3. Farhangmehr V., Imanian H., Mohammadian A., Cobo JH., Shirkhani H., Payeur P. A spatiotemporal CNN-LSTM deep learning model for predicting soil temperature in diverse large-scale regional climates. Sci Total Environ 2025; 968: 178901.
4. Geng Q., Wang L., Li Q. Soil temperature prediction based on explainable artificial intelligence and LSTM. Front Environ Sci 2024; 12: 1-16.
5. Guleryuz D. Estimation of soil temperatures with machine learning algorithms—Giresun and Bayburt stations in Turkey. Theor Appl Climatol 2022; 147(1-2): 109-125.
6. Gurlek C. The estimation of monthly mean soil temperature at different depths in Sivas province, Turkey by Artificial Neural Networks. Commun Soil Sci Plant Anal. 2023; 54(3): 408–430.
7. Hosseinzadeh Talaee P. Daily soil temperature modeling using neuro-fuzzy approach. Theor Appl Climatol 2014; 118(3): 481-489.
8. Kim S., Singh VP. Modeling daily soil temperature using data-driven models and spatial distribution. Theor Appl Climatol 2014; 118(3): 465-479.

9. Li Q., Hao H., Zhao Y., Geng Q., Liu G., Zhang Y., et al. GANs-LSTM model for soil temperature estimation from meteorological: a new approach. IEEE Access 2020; 8: 59427-59443.
10. Li X., Peng L., Yao X., Cui S., Hu Y., You C., et al. Long short-term memory neural network for air pollutant concentration predictions: method development and evaluation. Environ Pollut 2017; 231: 997-1004.
11. Mehdizadeh S., Ahmadi F., Kozekalani Sales A. Modelling daily soil temperature at different depths via the classical and hybrid models. Meteorol Appl 2020; 27(4): 1-15.
12. Mehdizadeh S., Fathian F., Safari MJS., Khosravi A. Developing novel hybrid models for estimation of daily soil temperature at various depths. Soil Tillage Res 2020; 197: 104513.
13. Mehdizadeh S., Mohammadi B., Bao Pham Q., Nguyen Khoi D., Thi Thuy Linh N. Implementing novel hybrid models to improve indirect measurement of the daily soil temperature: Elman neural network coupled with gravitational search algorithm and ant colony optimization. Meas J Int Meas Confed 2020; 165: 108127.
14. Moazenzadeh R., Mohammadi B. Assessment of bio-inspired metaheuristic optimisation algorithms for estimating soil temperature. Geoderma 2019; 353: 152-171.
15. Qing X., Niu Y. Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM. Energy 2018; 148: 461-468.
16. Samadianfard S., Asadi E., Jarhan S., Kazemi H., Kheshtgar S., Kisi O., et al. Wavelet neural networks and gene expression programming models to predict short-term soil temperature at different depths. Soil Tillage Res 2018; 175: 37-50.
17. Temur A. Comparison of ARIMA, LSTM and hybrid models in establishing sales budgets: A case of production facility. Sakarya University; 2019.
18. Yildirim A., Bilgili M., Ozbek A. One-hour-ahead solar radiation forecasting by MLP, LSTM, and ANFIS approaches. Meteorol Atmos Phys 2023; 135(1): 1-17.
19. Zeynoddin M., Bonakdari H., Ebtehaj I., Esmaeilbeiki F., Gharabaghi B., Zare Haghi D. A reliable linear stochastic daily soil temperature forecast model. Soil Tillage Res 2019; 189: 73-87.
20. Zounemat-Kermani M. Hydrometeorological parameters in prediction of soil temperature by means of artificial neural network: case study in Wyoming. J Hydrol Eng 2013; 18(6): 707-718.