Guest Editorial: Statistical models for environmental issues

Abstract

DuringApril 30–May 1, 2011, the international workshop on “Digital Governance and Surveillance Hotspot GeoInformatics for Monitoring, Etiology, Early Warning, and SustainableDevelopmentwithApplications toNatural and Environmental Resources” was held atMiddle East Technical University, Ankara, Turkey. From among the invited presentations, six papers were considered for publication andwere peer reviewed. One of those papers got published earlier, in this journal in 2012, due to some technical problem. This Special Issue comprises of the accepted versions of the remaining five papers,whichprovide agood representationof the researchworkpresented at theworkshop. With the thrust on modeling, these papers reflect a panorama of recent methods based on frequentist and Bayesian paradigms, parametric and non-parametric, with linear and non-linear/directional data. Theoretical rigor, motivating real-life examples and finally, challenging data sets make this compilation an useful addition to the arena of environmental statistics. An overview of the papers is presented below. Proper choice of a model for the underlying distribution, when it has to be based on censored data often due to measuremental problems and instrumental restrictions, has been an important problem with environmental data. Surucu addresses this problem from the point of model validation in some location-scale family. A new bivariate goodness-of-fit test for censored data is proposed by combining ideas from Tiku’s test for the univariate case with concomitants of order statistics. A real-life example based on water quality evaluation, where left censoring arises naturally due to the limitation of the measuring instrument, exemplifies the use of the proposed test. A newmethod of cross-ranking is introduced byMyers and Patil to reveal structures and dependences in multivariate data. The novelty lies in the simplicity of the method