Optimization for the design of environmental monitoring networks in routine and emergency settings S.J. Melles 1,2 , G.B.M. Heuvelink 1 , C.J.W. Twenhöfel 3 , A. van Dijk 3 , P. Hiemstra 4 , O. Baume 1 , U. Stöhlker 5 1 Environmental Sciences Group, Wageningen University, The Netherlands 2 current address: Environment Canada, stephanie.melles@ec.gc.ca 3 National Institute for Public Health and the Environment, The Netherlands 4 Faculty of Geosciences, Utrecht University, The Netherlands 5 Federal Office for Radiation Protection, Bundesamt für Strahlenschutz, Germany Abstract. The design of radiation monitoring networks were optimized by combin- ing a geostatistical assessment of routine prediction error with simulation modelling to assess network signalling function in emergency settings. A physical atmospheric dispersion model was used to simulate radioactive releases throughout the study area under different accident scenarios and varying weather conditions (e.g. small nu- clear power plant accidents and mock human-caused radioactive emissions). Net- work signalling function was defined as the ability to detect radioactivity above a critical threshold within 3 hours of a nuclear release. Spatial simulated annealing was used to obtain optimal monitoring designs by moving stations around and ac- cepting those designs that reduced a weighted sum of two criteria (prediction error of mean annual background radiation and network signalling function). Results were promising and the method should prove useful for assessing the efficacy of hazard monitoring networks designed to detect the unlikely event of a nuclear emergency. 1 I NTRODUCTION Radiation monitoring networks are designed to detect gamma dose rates emitted by both natural and artificial radionuclides. The importance of these networks is without question given the potential for accidents like the radioactive release at Three Mile Island in Penn- sylvania (1979) and the Chernobyl nuclear power plant (NPP) accident in the Ukraine (1986). Currently there are 436 NPPs operating worldwide [8] and that number is set to increase. The probability of an attack with a dirty bomb is difficult to estimate, but may be even higher than the probability of an NPP type accident [8]. In addition, public fears related to the risks of radiation tend to be amplified [8]. The National Institute for Public Health and the Environment (RIVM) operates the Dutch National Radioactivity Monitoring network, and in Germany, the Federal Office for Radiation Protection (BfS) is the agency responsible for the German network. Moni- toring stations are more or less uniformly spread across the two countries, with increased densities near nuclear power plants and along country borders [6]. However, there is a need to coordinate the sampling design of radiation monitoring networks amongst these and other countries because hazard releases have trans boundary properties [7]. In order to optimize a sampling design, one must first select an appropriate criterion with which to evaluate the suitability of a given design. Also referred to as the objective function, the criterion must encompass the sometimes conflicting objectives of a moni- toring network. In cases where environmental variables are being mapped, it is gener- ally appropriate to use model-based geostatistical approaches that rely on a pre-specified