Evaluating forecasts of the evolution of the cloudy boundary layer using diurnal composites of radar and lidar observations

Observations of boundary-layer cloud have been made using radar and lidar at Chilbolton, Hampshire, UK. These have been compared with output from 7 different global and regional models. Fifty-five cloudy days have been composited to reveal the mean diurnal variation of cloud top and base heights, cloud thickness and liquid water path of the clouds. To enable like-for-like comparison between model and observations, the observations have been averaged on to the grid of each model. The composites show a distinct diurnal cycle in observed cloud; the cloud height exhibits a sinusoidal variation throughout the day with a maximum at around 1600 and a minimum at around 0700 UTC. This diurnal cycle is captured by six of the seven models analysed, although the models generally under-predict both cloud top and cloud base heights throughout the day. The two worst performing models in terms of cloud boundaries also have biases of around a factor of two in liquid water path; these were the only two models that did not include an explicit formulation for cloud-top entrainment.

Verification of cloud-fraction forecasts

Cloud radar and lidar can be used to evaluate the skill of numerical weather prediction models in forecasting the timing and placement of clouds, but care must be taken in choosing the appropriate metric of skill to use due to the non- Gaussian nature of cloud-fraction distributions. We compare the properties of a number of different verification measures and conclude that of existing measures the Log of Odds Ratio is the most suitable for cloud fraction. We also propose a new measure, the Symmetric Extreme Dependency Score, which has very attractive properties, being equitable (for large samples), difficult to hedge and independent of the frequency of occurrence of the quantity being verified. We then use data from five European ground-based sites and seven forecast models, processed using the ‘Cloudnet’ analysis system, to investigate the dependence of forecast skill on cloud fraction threshold (for binary skill scores), height, horizontal scale and (for the Met Office and German Weather Service models) forecast lead time. The models are found to be least skillful at predicting the timing and placement of boundary-layer clouds and most skilful at predicting mid-level clouds, although in the latter case they tend to underestimate mean cloud fraction when cloud is present. It is found that skill decreases approximately inverse-exponentially with forecast lead time, enabling a forecast ‘half-life’ to be estimated. When considering the skill of instantaneous model snapshots, we find typical values ranging between 2.5 and 4.5 days. Copyright c 2009 Royal Meteorological Society

A new method for evaluating regional air quality forecasts

A Kriging interpolation method is combined with an object-based evaluation measure to assess the ability of the UK Met Office's dispersion and weather prediction models to predict the evolution of a plume of tracer as it was transported across Europe. The object-based evaluation method, SAL, considers aspects of the Structure, Amplitude and Location of the pollutant field. The SAL method is able to quantify errors in the predicted size and shape of the pollutant plume, through the structure component, the over- or under-prediction of the pollutant concentrations, through the amplitude component, and the position of the pollutant plume, through the location component. The quantitative results of the SAL evaluation are similar for both models and close to a subjective visual inspection of the predictions. A negative structure component for both models, throughout the entire 60 hour plume dispersion simulation, indicates that the modelled plumes are too small and/or too peaked compared to the observed plume at all times. The amplitude component for both models is strongly positive at the start of the simulation, indicating that surface concentrations are over-predicted by both models for the first 24 hours, but modelled concentrations are within a factor of 2 of the observations at later times. Finally, for both models, the location component is small for the first 48 hours after the start of the tracer release, indicating that the modelled plumes are situated close to the observed plume early on in the simulation, but this plume location error grows at later times. The SAL methodology has also been used to identify differences in the transport of pollution in the dispersion and weather prediction models. The convection scheme in the weather prediction model is found to transport more pollution vertically out of the boundary layer into the free troposphere than the dispersion model convection scheme resulting in lower pollutant concentrations near the surface and hence a better forecast for this case study.

Spatiotemporal behavior of the TIGGE medium-range ensemble forecasts

Using the recently-developed mean–variance of logarithms (MVL) diagram, together with the TIGGE archive of medium-range ensemble forecasts from nine different centres, an analysis is presented of the spatiotemporal dynamics of their perturbations, showing how the differences between models and perturbation techniques can explain the shape of their characteristic MVL curves. In particular, a divide is seen between ensembles based on singular vectors or empirical orthogonal functions, and those based on bred vector, Ensemble Transform with Rescaling or Ensemble Kalman Filter techniques. Consideration is also given to the use of the MVL diagram to compare the growth of perturbations within the ensemble with the growth of the forecast error, showing that there is a much closer correspondence for some models than others. Finally, the use of the MVL technique to assist in selecting models for inclusion in a multi-model ensemble is discussed, and an experiment suggested to test its potential in this context.

Judgments, forecasts and decisions: an analysis of fund managers over time

Decision theory is the study of models of judgement involved in, and leading to, deliberate and (usually) rational choice. In real estate investment there are normative models for the allocation of assets. These asset allocation models suggest an optimum allocation between the respective asset classes based on the investors’ judgements of performance and risk. Real estate is selected, as other assets, on the basis of some criteria, e.g. commonly its marginal contribution to the production of a mean variance efficient multi asset portfolio, subject to the investor’s objectives and capital rationing constraints. However, decisions are made relative to current expectations and current business constraints. Whilst a decision maker may believe in the required optimum exposure levels as dictated by an asset allocation model, the final decision may/will be influenced by factors outside the parameters of the mathematical model. This paper discusses investors' perceptions and attitudes toward real estate and highlights the important difference between theoretical exposure levels and pragmatic business considerations. It develops a model to identify “soft” parameters in decision making which will influence the optimal allocation for that asset class. This “soft” information may relate to behavioural issues such as the tendency to mirror competitors; a desire to meet weight of money objectives; a desire to retain the status quo and many other non-financial considerations. The paper aims to establish the place of property in multi asset portfolios in the UK and examine the asset allocation process in practice, with a view to understanding the decision making process and to look at investors’ perceptions based on an historic analysis of market expectation; a comparison with historic data and an analysis of actual performance.

The production and consumption of commercial real estate market forecasts

Whilst the vast majority of the research on property market forecasting has concentrated on statistical methods of forecasting future rents, this report investigates the process of property market forecast production with particular reference to the level and effect of judgemental intervention in this process. Expectations of future investment performance at the levels of individual asset, sector, region, country and asset class are crucial to stock selection and tactical and strategic asset allocation decisions. Given their centrality to investment performance, we focus on the process by which forecasts of rents and yields are generated and expectations formed. A review of the wider literature on forecasting suggests that there are strong grounds to expect that forecast outcomes are not the result of purely mechanical calculations.

Early warning with calibrated and sharper probabilistic forecasts

Given a nonlinear model, a probabilistic forecast may be obtained by Monte Carlo simulations. At a given forecast horizon, Monte Carlo simulations yield sets of discrete forecasts, which can be converted to density forecasts. The resulting density forecasts will inevitably be downgraded by model mis-specification. In order to enhance the quality of the density forecasts, one can mix them with the unconditional density. This paper examines the value of combining conditional density forecasts with the unconditional density. The findings have positive implications for issuing early warnings in different disciplines including economics and meteorology, but UK inflation forecasts are considered as an example.

Ensemble forecasts of a flood-producing storm: comparison of the influence of model-state perturbations and parameter modifications

High-resolution ensemble simulations (Δx = 1 km) are performed with the Met Office Unified Model for the Boscastle (Cornwall, UK) flash-flooding event of 16 August 2004. Forecast uncertainties arising from imperfections in the forecast model are analysed by comparing the simulation results produced by two types of perturbation strategy. Motivated by the meteorology of the event, one type of perturbation alters relevant physics choices or parameter settings in the model's parametrization schemes. The other type of perturbation is designed to account for representativity error in the boundary-layer parametrization. It makes direct changes to the model state and provides a lower bound against which to judge the spread produced by other uncertainties. The Boscastle has genuine skill at scales of approximately 60 km and an ensemble spread which can be estimated to within ∼ 10% with only eight members. Differences between the model-state perturbation and physics modification strategies are discussed, the former being more important for triggering and the latter for subsequent cell development, including the average internal structure of convective cells. Despite such differences, the spread in rainfall evaluated at skilful scales is shown to be only weakly sensitive to the perturbation strategy. This suggests that relatively simple strategies for treating model uncertainty may be sufficient for practical, convective-scale ensemble forecasting.

On reliability analysis of multi-categorical forecasts

Reliability analysis of probabilistic forecasts, in particular through the rank histogram or Talagrand diagram, is revisited. Two shortcomings are pointed out: Firstly, a uniform rank histogram is but a necessary condition for reliability. Secondly, if the forecast is assumed to be reliable, an indication is needed how far a histogram is expected to deviate from uniformity merely due to randomness. Concerning the first shortcoming, it is suggested that forecasts be grouped or stratified along suitable criteria, and that reliability is analyzed individually for each forecast stratum. A reliable forecast should have uniform histograms for all individual forecast strata, not only for all forecasts as a whole. As to the second shortcoming, instead of the observed frequencies, the probability of the observed frequency is plotted, providing and indication of the likelihood of the result under the hypothesis that the forecast is reliable. Furthermore, a Goodness-Of-Fit statistic is discussed which is essentially the reliability term of the Ignorance score. The discussed tools are applied to medium range forecasts for 2 m-temperature anomalies at several locations and lead times. The forecasts are stratified along the expected ranked probability score. Those forecasts which feature a high expected score turn out to be particularly unreliable.

Predicting outliers in ensemble forecasts

An ensemble forecast is a collection of runs of a numerical dynamical model, initialized with perturbed initial conditions. In modern weather prediction for example, ensembles are used to retrieve probabilistic information about future weather conditions. In this contribution, we are concerned with ensemble forecasts of a scalar quantity (say, the temperature at a specific location). We consider the event that the verification is smaller than the smallest, or larger than the largest ensemble member. We call these events outliers. If a K-member ensemble accurately reflected the variability of the verification, outliers should occur with a base rate of 2/(K + 1). In operational forecast ensembles though, this frequency is often found to be higher. We study the predictability of outliers and find that, exploiting information available from the ensemble, forecast probabilities for outlier events can be calculated which are more skilful than the unconditional base rate. We prove this analytically for statistically consistent forecast ensembles. Further, the analytical results are compared to the predictability of outliers in an operational forecast ensemble by means of model output statistics. We find the analytical and empirical results to agree both qualitatively and quantitatively.

Estimating reliability and resolution of probability forecasts through decomposition of the empirical score

References (20)Cited By (1)Export CitationAboutAbstract Proper scoring rules provide a useful means to evaluate probabilistic forecasts. Independent from scoring rules, it has been argued that reliability and resolution are desirable forecast attributes. The mathematical expectation value of the score allows for a decomposition into reliability and resolution related terms, demonstrating a relationship between scoring rules and reliability/resolution. A similar decomposition holds for the empirical (i.e. sample average) score over an archive of forecast–observation pairs. This empirical decomposition though provides a too optimistic estimate of the potential score (i.e. the optimum score which could be obtained through recalibration), showing that a forecast assessment based solely on the empirical resolution and reliability terms will be misleading. The differences between the theoretical and empirical decomposition are investigated, and specific recommendations are given how to obtain better estimators of reliability and resolution in the case of the Brier and Ignorance scoring rule.