HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text Free Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Agyare, W. A. Articles by Vlek, P. L. G. Search for Related Content PubMed Articles by Agyare, W. A. Articles by Vlek, P. L. G. GeoRef GeoRef Citation Agricola Articles by Agyare, W. A. Articles by Vlek, P. L. G. Related Collections Hydraulic Conductivity Pedotransfer Functions Soil Physics

Artificial Neural Network Estimation of Saturated Hydraulic Conductivity

^a Savanna Agricultural Research Inst. (SARI), CSIR, P.O. Box 52, Tamale, Ghana
^b Dep. of Geography, Seoul National Univ., Shilim-Dong, Kwanak-Gu, Seoul, Korea
^c Center for Development Research (ZEF), Univ. of Bonn, Walter-Flex-Str. 3, 53113 Bonn, Germany

View larger version (33K): [in this window] [in a new window]   FIG. 1. Sensitivity of different soil input parameters for different data forms, (A) raw, (B) normalized, (C) z-scored, and (D) minimum–maximum (0–1) data, for estimating saturated hydraulic conductivity in the artificial neural network. (SABST, subangular blocky structure; GST, granular structure; WKSG, weak structure; MSG, moderately strong structure; SSG, strong structure; FSS, fine structural size; MedSS, medium structural size; CSS, coarse and medium structural size; OC, organic carbon; CEC, cation exchange capacity; BD, bulk density.

View larger version (16K): [in this window] [in a new window]   FIG. 2. Variation in saturated hydraulic conductivity estimation with increasing number of input parameters for (A) coefficient of determination (R2) and (B) normalized mean square error (NMSE) in the artificial neural network.

View larger version (20K): [in this window] [in a new window]   FIG. 3. Trend of training data size effect on (A) coefficient of determination (R2) and (B) normalized mean square error (NMSE) for training and testing data in estimating saturated hydraulic conductivity using combined data from Ejura and Tamale sites in the artificial neural network.

View larger version (26K): [in this window] [in a new window]   FIG. 4. Measured compared to estimated saturated hydraulic conductivity (Ks) for (A) training and (B) testing data for 60 randomly selected Ejura topsoil datasets in the artificial neural network with Ks transformed between 0 and 1.

View larger version (17K): [in this window] [in a new window]   FIG. 5. Comparison of coefficient of determination (R2) for estimated saturated hydraulic conductivity for different testing data using training data from the same and different sites and indicating their SE bars in the artificial neural network (a, b, c shows Bonferroni mean separation test, with same letter indicating not significantly different at p < 0.05).

View larger version (16K): [in this window] [in a new window]   FIG. 6. Comparison of normalized mean square error (NMSE) for estimated saturated hydraulic conductivity for different testing data using training data from the same and different sites and indicating their SE bars in the artificial neural network.

View larger version (28K): [in this window] [in a new window]   FIG. 7. Comparison of measured (Ksm) and estimated (Kse) saturated hydraulic conductivity for Tamale (A and B) and Ejura sites (C and D)– using training data from different site– in the artificial neural network with Ks transformed between 0 and 1.