Article Details

Capacities of soil water reservoirs and their better regression models by combining ‘‘merged groups PCA’’ in Chongqing, China

Oleh   Liu Juan [-]
Kontributor / Dosen Pembimbing : Wei Chao-Fu, Xie Qian, Zhang Wei-Hua
Jenis Koleksi : Jurnal elektronik
Penerbit : Lain-lain
Fakultas :
Subjek :
Kata Kunci : Clay content, Multiple regression analysis, Principal component analysis, Soil physical characteristics, Soil water reservoir capacities
Sumber : ScienceDirect
Staf Input/Edit : Irwan Sofiyan  
File : 1 file
Tanggal Input : 2019-07-17 14:56:54

Generic placeholder image
2011 EJRNL PP Liu Juan 1.pdf ]


Soil water resource, together with the surface and sub-surface water resource, is essential to the regional water balance and world water cycle. A total of 90 soil samples were collected from 30 different soil profiles of dry fields throughout Chongqing, China randomly to show how soil could be a crucial part of water resources by discussing their five types of calculated soil water reservoir capacities, namely the total soil water reservoir capacity (mm) (TC), soil water storage capacity (mm) (SC), unavailable soil water reservoir capacity (mm) (UC), available soil water reservoir capacity (mm) (AC), and soil dead water storage capacity (mm) (DC) in certain layer, respectively. Overall, the total soil water reservoir capacity in 0–40 cm was about 209 mm, of which 70 mm belonged to available soil water reservoir capacity. Not all the five types of soil water reservoir capacities had significant correlations between each other. Soil structure, especially the size and quantity of soil pore was mainly determined by soil particle composition (clay, silt, and sand content). The more sand and less clay led to the more soil macropores, which provided room for soil water. Thus, clay, silt, and sand content jointly produced profound influence on soil water reservoir capacities. Nevertheless, specific water capacity and topographic factors displayed weak correlations to soil water reservoir capacities, which required further research works. Ultimately, the better regression models were achieved by multiple regression analysis coupled with ‘‘merged groups PCA’’ than by multiple regression analysis with ‘‘all variables PCA’’. UC, SC, TC and DC could be well simulated (mostly R2 > 0.70; P 0.70; P