1 AIT Asian Institute of Technology

Assessment of soil salinity in a Tropical zone using ALOS PALSAR satellite imagery and ANN model

AuthorWalaiporn Phonphan
Call NumberAIT Diss. no.RS-14-07
Subject(s)Soil salinization--Remote sensing--Thailand, Northeastern

NoteA dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Remote Sensing and Geographic Information Systems
PublisherAsian Institute of Technology
Series StatementDissertation ; no. RS-14-07
AbstractSoil salinity is akey agricultural issuein many areas and,without effective investigation and management,canaggravateandexpand. Soil scientists categorize thesoil salinity level by Electrical Conductivity (EC) measurement.However, field measurements of EC values require extensive time, cost, resources, and experience. Thus, remote sensing is an attractive alternativeforcollecting and investigatingspatial data inlarger areas within short time. Due to constraints inoptical image data,microwave remote sensing data was chosen as a tool to study the tropical zone as the area of interest, which is intensely influenced by rainfall and humidity. ALOS L-band provides long wavelength with deep penetrationconsidered as reliable and useful forstudying the field under theaboveconditions.According to soil theory, the EC value isrelatedtosurface roughness and soil moisture.Severalresearchers estimated soil moisture through microwave remote sensing data, but there are fewstudies investigatingthe direct relationship between EC and Backscattering Coefficient(BC). Other researchers have studied through various models or variables that provednot quitesuccessful in derivingEC directly from BC data due to incorporating a large number of inputs and a complex multistage examination to identify salinesoil areasthroughout the learning process.Thus, the study aims to propose anestimation of EC directly from BC of microwave remote sensing data,i.e., arelationship between EC obtained from field survey and BC from microwavedata. Implementation of the study startedbyanalyzingstatistical data comprising EC data from asaline soil field survey and BC of PALSAR-ALOS databy statisticbased methods to investigate therelationship between EC and BC values. Various advanced statistical measurements such as Pearson correlation, regression modeland model fitting (polynomial, logarithm)were applied in this research. The research found thattheregression model gives an r-square value of0.743forthe relationship of EC and BC. Ther-square values were subjecttoanincrease asthepolynomialwas adjusted in the model. Artificial Neural Network (ANN) is one of the techniques to figure out the EC -BC relationshipproposed in this study.The ANN model presented anr-squarevalue of0.93,which was regarded high. BC from radar is influenced bymoisture and surface roughness, which are the main factors in identifying whether anarea has asoil salinity issue or not. Results of the research also reaffirmthe potentialof Radar BC in estimating and monitoring salineareas throughtheANN modellearning process.Therefore, the relationship between BC and EC wasused to develop asalinity modelsuitingthe real-time scenario,whichcan be successfully utilized as an effective analytical tool for EC estimation from microwave remote sensing (PALSAR ALOS) data.
Year2014
Corresponding Series Added EntryAsian Institute of Technology. Dissertation ; no. RS-14-07
TypeDissertation
SchoolSchool of Engineering and Technology (SET)
DepartmentDepartment of Information and Communications Technologies (DICT)
Academic Program/FoSRemote Sensing (RS)
Chairperson(s)Tripathi, Nitin Kumar
Examination Committee(s)Apisit Eiumnoh ;TaravudhTipdecho ;Nakamura, Shinichi
Scholarship Donor(s)SuanSunandhaRajabhat University, Thailand
DegreeThesis (Ph. D.) - Asian Institute of Technology, 2014


Usage Metrics
View Detail0
Read PDF0
Download PDF0