Modeling daily soil temperature using data-driven models and spatial distribution
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
2014, Springer-Verlag Wien. The objective of this study is to develop data-driven models, including multilayer perceptron (MLP) and adaptive neurofuzzy inference system (ANFIS), for estimating daily soil temperature at Champaign and Springfield stations in Illinois. The best input combinations (one, two, and three inputs) can be identified using MLP. The ANFIS is used to estimate daily soil temperature using the best input combinations (one, two, and three inputs). From the performance evaluation and scatter diagrams of MLP and ANFIS models, MLP 3 produces the best results for both stations at different depths (10 and 20 cm), and ANFIS 3 produces the best results for both stations at two different depths except for Champaign station at the 20 cm depth. Results of MLP are better than those of ANFIS for both stations at different depths. The MLP-based spatial distribution is used to estimate daily soil temperature using the best input combinations (one, two, and three inputs) at different depths below the ground. The MLP-based spatial distribution estimates daily soil temperature with high accuracy, but the results of MLP and ANFIS are better than those of the MLP-based spatial distribution for both stations at different depths. Data-driven models can estimate daily soil temperature successfully in this study.
