Validation of air pollution biomonitoring networks and related data modelling: a geostatistical approach† S. Gorelli, * a A. Santucci, a G. Lorenzini b and C. Nali * b Received 15th October 2008, Accepted 11th February 2009 First published as an Advance Article on the web 20th March 2009 DOI: 10.1039/b818254j This paper describes a geostatistical approach for environmental data modeling based on value descriptive analysis, variographic analysis, estimation method and validation. It shows an application related to atmospheric pollutant dispersion (ozone) detected by active biomonitoring in the province of Livorno (Italy). This work proposes a two-stage approach based on: the computation of an index on an exploratory and geostatistical data analysis to assess whether a monitoring network is well distributed in space and allows us to get reliable information about the whole study area; the derivation of a continuous representation of the variable (the Cotyledonous Damage Index) from punctual measurements. The geostatistical approach proposed is useful to define in a structured way the possible problems in a monitoring network, to control the data estimation error in the points not sampled. The case study analysed underlines that the biomonitoring network outline does not have a representative sample station distribution of the study area introducing significant errors related to the territorial generalization of the derived information. The proposed approach allows us to elaborate on the obtained data to carry out the data spatialization associating an error to them and to the evaluation model. Introduction The environmental data modelling represents a relevant step to analyze and to interpret environmental phenomena; in partic- ular, the information spatial interpolation phase has an impor- tant role for modelling the continuous distribution of the variable that characterizes environmental phenomena. This work takes into account a geostatistical approach for identifying a cluster of techniques useful to validate the moni- toring network and the data sample analysis. This analytic approach allows us to study spatial data dependence and to evaluate, not at the sample location, the data that characterize the studied phenomenon and the associate error, to understand the reliability and quality of the analytic results. This study describes the geostatistical approach for environ- mental data modelling, based on descriptive analysis of the values, on variographic analysis, on appraisal methodology and validation. Specifically, we studied an application for the diffu- sion of an atmospheric pollutant (ozone) registered by active biomonitoring in the Livorno province (Italy). 1 Through this analytic approach, it was possible to determine the biomonitoring network reliability and, successively, to elab- orate the relevant data spatial distribution. This study proposes a process of environmental data valida- tion and modelling that follows a logical flow for evaluating the monitoring network reliability, characterizing the sample data and the interpolated data consistency in the survey area (Fig. 1). Materials and methods Study area The study area was the Province of Livorno (longitude: 10  18 0 E; latitude: 43  33 0 N) of about 1212,81 km 2 . Livorno Province is an area developed along the Mediterranean sea coast in the Tuscany region. The climate is sub-oceanic to Mediterranean, with a dry period during the summer. The average annual temperature is about 15  C and calm conditions dominate although the coastal areas and the city centre are exposed to a sea–land breeze regime: in the late morning air flows inland from the open sea and the opposite holds true late in the day. 2 Fig. 1 The process of environmental data validation and modelling. a Dipartimento di Ingegneria Civile, University of Pisa, via Diotisalvi, 2, 56124 Pisa, Italy. E-mail: sgorelli@agr.unipi.it b Dipartimento di Coltivazione e Difesa delle Specie Legnose ‘‘G. Scaramuzzi’’, University of Pisa, via del Borghetto, 80 - 56124 Pisa, Italy. E-mail: cnali@agr.unipi.it † Presented at TerraData Environmetrics 2008, a recent workshop on Quality Assurance in Ecological Monitoring held on the 7 March 2008, Siena, Italy. This journal is ª The Royal Society of Chemistry 2009 J. Environ. Monit., 2009, 11, 793–797 | 793 PAPER www.rsc.org/jem | Journal of Environmental Monitoring