HESS Opinions ``Ensembles, uncertainty and flood prediction''

Ensemble predictions are being used more frequently to model the propagation of uncertainty through complex, coupled meteorological, hydrological and coastal models, with the goal of better characterising flood risk. In this paper, we consider the issues that we judge to be important when designing and evaluating ensemble predictions, and make recommendations for the guidance of future research.

A data collection case study supporting requirements oriented prediction and management in software developments

Consider the statement "this project should cost X and has risk of Y". Such statements are used daily in industry as the basis for making decisions. The work reported here is part of a study aimed at providing a rational and pragmatic basis for such statements. Of particular interest are predictions made in the requirements and early phases of projects. A preliminary model has been constructed using Bayesian Belief Networks and in support of this, a programme to collect and study data during the execution of various software development projects commenced in May 2002. The data collection programme is undertaken under the constraints of a commercial industrial regime of multiple concurrent small to medium scale software development projects. Guided by pragmatism, the work is predicated on the use of data that can be collected readily by project managers; including expert judgements, effort, elapsed times and metrics collected within each project.

A GIS-based empirical model for vegetation prediction in Lefka Ori, Crete

The aim of the study was to establish and verify a predictive vegetation model for plant community distribution in the alti-Mediterranean zone of the Lefka Ori massif, western Crete. Based on previous work three variables were identified as significant determinants of plant community distribution, namely altitude, slope angle and geomorphic landform. The response of four community types against these variables was tested using classification trees analysis in order to model community type occurrence. V-fold cross-validation plots were used to determine the length of the best fitting tree. The final 9node tree selected, classified correctly 92.5% of the samples. The results were used to provide decision rules for the construction of a spatial model for each community type. The model was implemented within a Geographical Information System (GIS) to predict the distribution of each community type in the study site. The evaluation of the model in the field using an error matrix gave an overall accuracy of 71%. The user's accuracy was higher for the Crepis-Cirsium (100%) and Telephium-Herniaria community type (66.7%) and relatively lower for the Peucedanum-Alyssum and Dianthus-Lomelosia community types (63.2% and 62.5%, respectively). Misclassification and field validation points to the need for improved geomorphological mapping and suggests the presence of transitional communities between existing community types.

Prediction of global and local model quality in CASP8 using the ModFOLD server

The development of effective methods for predicting the quality of three-dimensional (3D) models is fundamentally important for the success of tertiary structure (TS) prediction strategies. Since CASP7, the Quality Assessment (QA) category has existed to gauge the ability of various model quality assessment programs (MQAPs) at predicting the relative quality of individual 3D models. For the CASP8 experiment, automated predictions were submitted in the QA category using two methods from the ModFOLD server-ModFOLD version 1.1 and ModFOLDclust. ModFOLD version 1.1 is a single-model machine learning based method, which was used for automated predictions of global model quality (QMODE1). ModFOLDclust is a simple clustering based method, which was used for automated predictions of both global and local quality (QMODE2). In addition, manual predictions of model quality were made using ModFOLD version 2.0-an experimental method that combines the scores from ModFOLDclust and ModFOLD v1.1. Predictions from the ModFOLDclust method were the most successful of the three in terms of the global model quality, whilst the ModFOLD v1.1 method was comparable in performance to other single-model based methods. In addition, the ModFOLDclust method performed well at predicting the per-residue, or local, model quality scores. Predictions of the per-residue errors in our own 3D models, selected using the ModFOLD v2.0 method, were also the most accurate compared with those from other methods. All of the MQAPs described are publicly accessible via the ModFOLD server at: http://www.reading.ac.uk/bioinf/ModFOLD/. The methods are also freely available to download from: http://www.reading.ac.uk/bioinf/downloads/.

Results of an aqueous source term model for a radiological risk assessment of the Drigg LLW Site, U.K.

A radionuclide source term model has been developed which simulates the biogeochemical evolution of the Drigg low level waste (LLW) disposal site. The DRINK (DRIgg Near field Kinetic) model provides data regarding radionuclide concentrations in groundwater over a period of 100,000 years, which are used as input to assessment calculations for a groundwater pathway. The DRINK model also provides input to human intrusion and gaseous assessment calculations through simulation of the solid radionuclide inventory. These calculations are being used to support the Drigg post closure safety case. The DRINK model considers the coupled interaction of the effects of fluid flow, microbiology, corrosion, chemical reaction, sorption and radioactive decay. It represents the first direct use of a mechanistic reaction-transport model in risk assessment calculations.

Ensemble prediction for nowcasting with a convection-permitting model - II: forecast error statistics

A 24-member ensemble of 1-h high-resolution forecasts over the Southern United Kingdom is used to study short-range forecast error statistics. The initial conditions are found from perturbations from an ensemble transform Kalman filter. Forecasts from this system are assumed to lie within the bounds of forecast error of an operational forecast system. Although noisy, this system is capable of producing physically reasonable statistics which are analysed and compared to statistics implied from a variational assimilation system. The variances for temperature errors for instance show structures that reflect convective activity. Some variables, notably potential temperature and specific humidity perturbations, have autocorrelation functions that deviate from 3-D isotropy at the convective-scale (horizontal scales less than 10 km). Other variables, notably the velocity potential for horizontal divergence perturbations, maintain 3-D isotropy at all scales. Geostrophic and hydrostatic balances are studied by examining correlations between terms in the divergence and vertical momentum equations respectively. Both balances are found to decay as the horizontal scale decreases. It is estimated that geostrophic balance becomes less important at scales smaller than 75 km, and hydrostatic balance becomes less important at scales smaller than 35 km, although more work is required to validate these findings. The implications of these results for high-resolution data assimilation are discussed.

Automated tertiary structure prediction with accurate local model quality assessment using the IntFOLD-TS method

The IntFOLD-TS method was developed according to the guiding principle that the model quality assessment would be the most critical stage for our template based modelling pipeline. Thus, the IntFOLD-TS method firstly generates numerous alternative models, using in-house versions of several different sequence-structure alignment methods, which are then ranked in terms of global quality using our top performing quality assessment method – ModFOLDclust2. In addition to the predicted global quality scores, the predictions of local errors are also provided in the resulting coordinate files, using scores that represent the predicted deviation of each residue in the model from the equivalent residue in the native structure. The IntFOLD-TS method was found to generate high quality 3D models for many of the CASP9 targets, whilst also providing highly accurate predictions of their per-residue errors. This important information may help to make the 3D models that are produced by the IntFOLD-TS method more useful for guiding future experimental work

Comparing the UK Met Office climate prediction system DePreSys with idealized predictability in the HadCM3 model

The initial condition effect on climate prediction skill over a 2-year hindcast time-scale has been assessed from ensemble HadCM3 climate model runs using anomaly initialization over the period 1990–2001, and making comparisons with runs without initialization (equivalent to climatological conditions), and to anomaly persistence. It is shown that the assimilation improves the prediction skill in the first year globally, and in a number of limited areas out into the second year. Skill in hindcasting surface air temperature anomalies is most marked over ocean areas, and is coincident with areas of high sea surface temperature and ocean heat content skill. Skill improvement over land areas is much more limited but is still detectable in some cases. We found little difference in the skill of hindcasts using three different sets of ocean initial conditions, and we obtained the best results by combining these to form a grand ensemble hindcast set. Results are also compared with the idealized predictability studies of Collins (Clim. Dynam. 2002; 19: 671–692), which used the same model. The maximum lead time for which initialization gives enhanced skill over runs without initialization varies in different regions but is very similar to lead times found in the idealized studies, therefore strongly supporting the process representation in the model as well as its use for operational predictions. The limited 12-year period of the study, however, means that the regional details of model skill should probably be further assessed under a wider range of observational conditions.

Multiple-project discount rates for cost-benefit analysis in construction projects: A formal risk model for microgeneration renewable energy technologies

The traditional economic approach for appraising the costs and benefits of construction project Net Present Values involves the calculation of net returns for each investment option under different discount rates. An alternative approach consists of multiple-project discount rates based on risk modelling. The example of a portfolio of microgeneration renewable energy technology (MRET) is presented to demonstrate that risks and future available budget for re-investment can be taken into account when setting discount rates for construction project specifications in presence of uncertainty. A formal demonstration is carried out through a reversed intertemporal approach of applied general equilibrium. It is demonstrated that risk and the estimated available budget for future re-investment can be included in the simultaneous assessment of the costs and benefits of multiple projects.