Meteorological Parameters–Soil Temperature Relations in a Sub-Tropical Summer Grassland: Physically-Based and Data-Driven Modeling

Jannatul Ferdaous Progga; Mohammad Nur Hossain Khan; Mg Mostofa Amın

doi:10.5152/AUAF.2023.23126

Meteorological Parameters–Soil Temperature Relations in a Sub-Tropical Summer Grassland: Physically-Based and Data-Driven Modeling

Authors : Jannatul Ferdaous Progga, Mohammad Nur Hossain Khan, Mg Mostofa Amın

Pages : 48-56

Doi:10.5152/AUAF.2023.23126

View : 134 | Download : 437

Publication Date : 2023-05-23

Article Type : Research Paper

Abstract :The knowledge of soil temperature dynamics at different depths is paramount for the agricultural industry because soil temperature impacts the physical, chemical, and biological processes in soil. A relationship between meteorological parameters and temperature at different depths in silt loam soil was assessed by using a physically based HYDRUS-1D model and a linear regression model. Soil temperature at 5, 10, 20, 30, and 50 cm soil layers, minimum and maximum air temperature, air pressure, relative humidity, dew point, rainfall, sunshine duration, wind speed, and evaporation data collected at a weather station were used. The correlation sensitivity for the input combinations was investigated. The quantitative evaluation based on mean absolute percentage error and R2 showed that the predictions of both linear regression model and HYDRUS-1D models were satisfactory. The R2 values at 5, 10, and 20 cm depths were 0.96, 0.94, and 0.88 for linear regression model, and 0.85, 0.86, and 0.78, for HYDRUS-1D model, respectively. Similarly, the mean absolute percentage error values for linear regression model were 0.81%, 0.87%, and 1.05%, whereas 3.44%, 2.87%, and 3.73% at 5, 10, and 20 cm depths for HYDRUS-1D model, respectively. Generally, the accuracy of the models diminished with increasing the soil depth. At >30 cm soil depth, both models failed to estimate soil temperature accurately. The R2 and mean absolute percentage error values at 50 cm depth for linear regression model were 0.55% and 1.25% and 0.51% and 4.13% for HYDRUS-1D, respectively. The linear regression model performed better than the HYDRUS-1D model. Five independent variables (mean air temperature, maximum humidity, rainfall, wind speed, and evaporation) were found to significantly affect the summer-time soil temperature. Either of the methods can be used satisfactorily to predict soil temperature at 0–20 cm soil depth.
Keywords : Buharlaşma, nem, HYDRUS 1D, lineer regresyon modeli, rüzgar hızı

ORIGINAL ARTICLE URL