Paper Information

Title: 

ESTIMATING OF CATION EXCHANGE CAPACITY IN TWO DOMINANT SOIL FAMILIES OF CHARMAHAL-VA-BAKHTIARI PROVINCE USING ARTIFICIAL NEURAL NETWORK

Type: PAPER
Author(s): MOHAJER R.,SALEHI MOHAMMAD HASAN,BEIGI HARCHEGANI HABIB ELAH
 
 
 
Name of Seminar: HAMAYESHE MANTAGHEI KARBORD FANAVARIHAYE NOVIN DAR KESHAVARZI
Type of Seminar:  CONGRESS
Sponsor:  Islamic Azad University, Shiraz
Date:  2008Volume 1
 
 
Abstract: 

SOIL QUALITY MAPS SERVE AS BASE MAPS FOR THE VARIOUS USES IN AGRICULTURE, NATURAL RESOURCES AND ENVIRONMENT. HAVE SPECIAL IMPOTENT. ONE OF THE SOIL CHARACTERISTICS THAT CAN IMPROVE QUALITY OF SOIL MAPS IS CATION EXCHANGE CAPACITY (CEC). AS MEASURING CEC IS TIME CONSUMING AND COSTLY, IT IS USEFUL TO A FIND AN INDIRECT, SIMPLE AND CHEAP METHOD TO ESTIMATE CEC. PEDOTRANSFER FUNCTIONS ARE USED FOR ESTIMATION OF SOIL PROPERTIES. THESE FUNCTIONS ARE DEVELOPED BY DIFFERENT METHODS INCLUDING REGRESSION AND ARTIFICIAL NEURAL NETWORKS. WHEN SOILS ARE GROUPED BY TAXONOMIC ORDER, HORIZON OR LAYER, ACCURACY OF PREDICTIVE MODELS, IN GENERAL, HAS BEEN SHOWN TO IMPROVE. THE PURPOSES OF THIS RESEARCH ARE (1) UPGRADING THE SOIL MAPS BY DETERMINING THE CEC IN TWO DOMINANT SOIL FAMILIES IN CHAHARMAHAL-VA-BAKHTIARI PROVINCE; (2) DEVELOPING OF PTFS FOR CEC USING METHOD OF NEURAL NETWORKS, (3) ASSESSING THE EFFECT OF SOIL PARTITIONING INTO FAMILIES AND LAYERS ON THE QUALITY OF MODELS. THE STUDIED AREA CONSISTED OF THREE DELINATIONS OF TWO CONSOCIATION MAP UNITS OF SHAHRAK AND CHARMAHAL SERIES. SOIL SAMPLES WERE COLLECTED FROM EACH SOIL FAMILY FROM TWO DEPTHS OF 0-20 AND 30-50 CM. THE MEASURED SOIL PHYSICO-CHEMICAL PROPERTIES ARE: SOIL TEXURE, ORGANIC MATTER, EQUIVALENT CACO3, WATER CONTENT AT -1500 KPA (PWP), PH AND CEC. NEURAL NETWORK MODELS WERE DEVELOPED BY JMP 5.0 SOFTWARE R2 AND RMSE WERE USED TO EVALUATE AND SELECT BEST MODELS FOR ALL SAMPLES, FOR TOP- AND SUBSOIL LAYERS AND BOTH SOIL FAMILIES. THE RESULTS SHOWED THAT CEC PEDOTRANSFER FUNCTIONS MAY BE CONSTRUCTED WITH THE USE OF NEUARAL NETWORK METHOD.
IN GENERAL, PARTITIONING OF SOILS INTO LAYERS AND FAMILIES INCREASED THE ACCURACY OF MODELS. AMONGST THE MEASURED PROPERTIES, OM%, CLAY% AND PWP% WERE BEST PARAMETERS FOR ESTIMATING CEC. AS EVIDENT FROM R2 AND RMSE, AT ALL LEVELS OF PARTITIONING, NEURAL NETWORK DERIVED MODELS ESTIMATE CEC RELATIVELY ACCURATELY AND WITH HIGH PRECISION.

 
Keyword(s): CEC, PEDOTRANSFER FUNCTIONS, NEURAL NETWORK, SOIL FAMILY AND PARTITIONING
 
Yearly Visit 22   tarjomyar
 
Latest on Blog
Enter SID Blog