Royal Statistics Society Environmental Statistics Afternoon - 21st May 2018

RSS Avon Local Group and RSS Environmental Statistics Section Joint Meeting

University of Bath, Room 4W 1.2 (how to get to the University of Bath)

Everyone is welcome but for practical reasons we ask you to register with the form below. For any query please contact Ilaria Prosdocimi or Evangelos Evangelou.

We encourage students and all participants to bring posters along to the event for an informal poster session after the talks: no need to submit an abstract, just bring along a poster you would like to present.


13:30 Coffee & Registration
14:00 Claire Miller (University of Glasgow) Water quality: connected networks, dimensionality reduction and missing data
14:40 Zhe Sha (University of Bristol) Bayesian Hierarchical modelling of large scale geo-spatial processes
15:20 Coffee break (posters)
16:00 Hugo Winter (EDF Energy) Modelling the risk to UK coastal infrastructure posed by extreme storms
16:40 Nicole Augustin (University of Bath) Using forest eco-system monitoring data to model tree survival for investigating climate change effects
17:00 End of talks followed by drinks reception with posters with time for further discussion


Claire Miller1
Water quality: connected networks, dimensionality reduction and missing data

1University of Glasgow

  The surface water quality for lakes and rivers is routinely monitored to protect ecosystem, animal and human health and for compliance reporting to policymakers. Current monitoring strategies enable measurements of water quality determinands to be obtained through satellite observations, automatic monitoring sensors and to be available at hundreds of spatial locations over time. There is a wealth of data available, which can provide novel information on the status of ecosystems with appropriate statistical analysis. However, the data present several challenges for statistical modelling including the data dimensionality, connected monitoring locations, missing data, calibration and validation and complex ecosystem interactions.
  Statistical developments to overcome these challenges will be presented with applications including global lake water quality, river monitoring networks and freshwater connected ecosystems.

Zhe Sha1
Bayesian Hierarchical modelling of large scale geo-spatial processes

1University of Bristol

  We introduce a framework for updating large scale geo-spatial processes using a model-data synthesis method based on Bayesian hierarchical modelling. Two major challenges come from updating large-scale Gaussian process and modelling non-stationarity. To address the first, we adopt the SPDE approach that uses a sparse Gaussian Markov random fields (GMRF) approximation to reduce the computational cost and implement the Bayesian inference by using the INLA method. For non-stationary global processes, we propose two general models that accommodate commonly-seen geo-spatial problems. Finally, we show an example of updating an estimate of global glacial isostatic adjustment (GIA) using GPS measurements.

Hugo Winter1
Modelling the risk to UK coastal infrastructure posed by extreme storms

1EDF Energy R&D UK Centre

  EDF Energy is one of the UK's largest energy companies and its largest producer of low-carbon electricity. We manage an existing UK nuclear fleet and plan for the future by developing new nuclear and renewable generation. Safety is an overriding priority for the EDF Energy Nuclear Generation fleet and as such it is important to assess the risks to these infrastructure assets; in this presentation we focus on risks associated with natural hazards.
  Storms (specifically extra-tropical cyclones) that come across the Atlantic and pass near or over the UK mainland can lead to extreme weather conditions. It is possible to observe extremes of many different environment variables (e.g. wind, rainfall, storm surge, waves, etc.) as part of a single storm event. It is vital to know not only how intense each of the separate hazards may be but also whether there are any important hazard combinations that could occur. As with any extreme value analysis problem, we also want to be able to estimate the intensity of events that are larger than any we have observed before in the historical record.
  Standard extreme value approaches would involve taking data from a site close to the infrastructure of interest and analysing this without attempting to model the physical behaviour behind the generating process (i.e. the storm). This presentation shall introduce three different projects that the EDF Energy R&D UK Centre are undertaking to try and improve the validity of extreme value results by trying to more accurately represent the storm process. These range from more advanced multivariate extreme value statistical models to numerical weather prediction tools that can be used to perturb past extreme storms (in effect creating 'black swan' events) which obey physical conservation laws.

Nicole Augustin1 , Alice Davis1 , Axel Albrecht2, Stefan Meining2, Heike Puhlmann2 and Karim Anaya-Izquierdo1
Using forest eco-system monitoring data to model tree survival for investigating climate change effects

1University of Bath, 2Forest Research Institute Baden-Württemberg (Germany)

  Forests are economically, recreationally and ecologically important, providing timber and wildlife habitat and acting as a carbon sink, among many ecosystem services. They are therefore extremely valuable to society, and it is crucial to ensure that they remain healthy. Forest health is monitored in Europe by The International Co-operative Programme on Assessment and Monitoring of Air Pollution Effects (ICP Forests) in cooperation with the European Union. More recently climate change has contributed to the decline in forest health and these data are increasingly being used to investigate the effects of climate change on forests in order to decide on forest management strategies for mitigation.
  Here we model extensive yearly data on tree mortality and crown defoliation, an indicator of tree health, from a monitoring survey carried out in Baden-Württemberg, Germany since 1983, which includes a part of the ICP transnational grid. On a changing irregular grid, defoliation, mortality and other tree and site specific variables are recorded. In some cases the grid locations are no longer observed which leads to censored data, also recruitment of trees happens throughout when new grid points are added.
  We model tree survival as a function of the predictor variables on climate, soil characteristics and deposition. We are interested in the process leading to tree mortality rather than prediction and this requires the inclusion of all potential drivers of tree mortality in the model. We use the semiparametric shared frailty model fitted using a Cox regression model which allows for random effects (frailties) taking care of dependence between neighbouring trees and non-linear smooth functions of time varying predictors and functional predictors. At each of 2385 locations 24 trees were observed between 1983 and 2016, with not all locations being observed yearly. Altogether a total of 80000 trees are observed making the analysis computationally challenging.

This meeting will adhere to the RSS Meeting Conduct Policy: all partecipants should be aware of the poilicy.