Comparison of different empirical methods and data-driven models for estimating reference evapotranspiration in semi-arid Central Anatolian Region of Turkey

Creative Commons License

Yurtseven I., Serengil Y.

ARABIAN JOURNAL OF GEOSCIENCES, cilt.14, sa.19, 2021 (SCI-Expanded, Scopus) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 14 Sayı: 19
  • Basım Tarihi: 2021
  • Doi Numarası: 10.1007/s12517-021-08150-8
  • Dergi Adı: ARABIAN JOURNAL OF GEOSCIENCES
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aquatic Science & Fisheries Abstracts (ASFA), Geobase, INSPEC
  • Anahtar Kelimeler: Reference evapotranspiration (ETo), Empirical methods, Artificial neural networks (ANN), Support vector machines (SVM), Random forest (RF), Multiple linear regression (MLR), LIMITED CLIMATIC DATA, ESTIMATING POTENTIAL EVAPOTRANSPIRATION, ARTIFICIAL NEURAL-NETWORK, ELEVATION GRADIENT, EVAPORATION, REGRESSION, TEMPERATURE, PERFORMANCE, CALIBRATION, FORECAST
  • Açık Arşiv Koleksiyonu: AVESİS Açık Erişim Koleksiyonu
  • İstanbul Üniversitesi-Cerrahpaşa Adresli: Evet

Özet

Evapotranspiration (ET) is a major hydrologic process to assess water budgets in terrestrial ecosystems. Since measurement of ET may involve labor intensive field technics in large areas, estimation is preferred in most cases. The FAO Penman-Monteith (PM FAO-56) is a widely recognized reference evapotranspiration (ETo) method for potential evapotranspiration calculations. The method requires a detailed and comprehensive meteorological data set; however, some empirical methods and models have attempted to calculate ET with less data. In this study, Makkink (ET_Mak), Hargreaves-Samani (ET_Har), Thornthwaite (ET_Thor), Blaney-Criddle (ET_BC), and Penman (ET_PM) were tested against the PM FAO-56. Penman method has achieved the highest accuracy among the empirical methods. In addition, the potential of artificial neural networks (ANN), support vector machines (SVM), random forest (RF), and multiple linear regression (MLR) for estimating ETo were investigated in a semi-arid Central Anatolian Region of Turkey. The results obtained with the ANN (based on multi-layer perceptron) and SVM models performed better than other tested data-driven models and empirical methods. These models could be used most effectively at elevation range of 850-1000 m. According to our results MLP, SVM, and Penman methods provided good performances in semi-arid regions in agricultural planning and water resources management studies. Furthermore, we concluded that integrating maximum temperature may result in improved accuracy in ET model simulations in semi-arid regions.