A carregar...
Publicação
Pedotransfer functions for estimating soil water retention properties of northern China agricultural soils: Development and needs
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.77 MB | Adobe PDF |
Orientador(es)
Resumo(s)
Many agro-environmental studies focusing on the efficient management of
soils and water resources make use of soil water simulation models. Reliable
soil hydraulic properties are critical for ensuring the accuracy of model simulations.
However, soil hydraulic parameters in northern China are generally
derived using external pedotransfer functions (PTFs) that do not take into
account the specificities of local edaphoclimatic conditions due to the lack of a
better alternative. Therefore, the main objective of this paper was to develop
PTFs for estimating the soil water retention curve (SWRC) in northern China
agricultural soils (named PTF-ANC). A total of 440 soil horizons were collected
from the existing literature. A flexible soil-textural conversion program was
first developed to harmonize soil particle-size data into the United States
Department of Agriculture (USDA) classification system. The SWRC parameters
of the van Genuchten model were also generated by curve fitting. Then,
the PTF-ANC were developed using artificial neural networks, with soil texture
and bulk density being used as input data and with a basic three-layer
back-propagation neural network. The PTF-ANC showed an acceptable accuracy
when predicting the SWRC, indicating a strong application potential for
northern China soils. Comparison of estimates with two widely used external
PTFs also showed that these were not suitable for characterizing the SWRC of
northern China agricultural soils. This is due to the fact that the main soil
textures (silt and silty loam textures) found in northern China soils were
misrepresented in those external soil databases. Overall, this paper presented
the absolute necessity of developing specific PTFs for northern China
agricultural soils
Descrição
Palavras-chave
neural network non-linear fitting pedotransfer functions soil hydraulic parameters soil water model textural data conversion
Contexto Educativo
Citação
Irrig. and Drain. 2021;1–16
Editora
Wiley