Fast identification algorithms for Gaussian process model

A class identification algorithms is introduced for Gaussian process(GP)models.The fundamental approach is to propose a new kernel function which leads to a covariance matrix with low rank,a property that is consequently exploited for computational efficiency for both model parameter estimation and model predictions.The objective of either maximizing the marginal likelihood or the Kullback–Leibler (K–L) divergence between the estimated output probability density function(pdf)and the true pdf has been used as respective cost functions.For each cost function,an efficient coordinate descent algorithm is proposed to estimate the kernel parameters using a one dimensional derivative free search, and noise variance using a fast gradient descent algorithm. Numerical examples are included to demonstrate the effectiveness of the new identification approaches.

An aggregated wind power generation model based on MERRA reanalysis data: MATLAB model and example data for the April 2014 wind farm distribution of Great Britain.

The MATLAB model is contained within the compressed folders (versions are available as .zip and .tgz). This model uses MERRA reanalysis data (>34 years available) to estimate the hourly aggregated wind power generation for a predefined (fixed) distribution of wind farms. A ready made example is included for the wind farm distribution of Great Britain, April 2014 ("CF.dat"). This consists of an hourly time series of GB-total capacity factor spanning the period 1980-2013 inclusive. Given the global nature of reanalysis data, the model can be applied to any specified distribution of wind farms in any region of the world. Users are, however, strongly advised to bear in mind the limitations of reanalysis data when using this model/data. This is discussed in our paper: Cannon, Brayshaw, Methven, Coker, Lenaghan. "Using reanalysis data to quantify extreme wind power generation statistics: a 33 year case study in Great Britain". Submitted to Renewable Energy in March, 2014. Additional information about the model is contained in the model code itself, in the accompanying ReadMe file, and on our website: http://www.met.reading.ac.uk/~energymet/data/Cannon2014/

Forecast uncertainty—ex Ante and ex Post: U.S. inflation and output growth

Survey respondents who make point predictions and histogram forecasts of macro-variables reveal both how uncertain they believe the future to be, ex ante, as well as their ex post performance. Macroeconomic forecasters tend to be overconfident at horizons of a year or more, but overestimate (i.e., are underconfident regarding) the uncertainty surrounding their predictions at short horizons. Ex ante uncertainty remains at a high level compared to the ex post measure as the forecast horizon shortens. There is little evidence of a link between individuals’ ex post forecast accuracy and their ex ante subjective assessments.

Development of a wind gust model to estimate gust speeds and their return periods

Spatially dense observations of gust speeds are necessary for various applications, but their availability is limited in space and time. This work presents an approach to help to overcome this problem. The main objective is the generation of synthetic wind gust velocities. With this aim, theoretical wind and gust distributions are estimated from 10 yr of hourly observations collected at 123 synoptic weather stations provided by the German Weather Service. As pre-processing, an exposure correction is applied on measurements of the mean wind velocity to reduce the influence of local urban and topographic effects. The wind gust model is built as a transfer function between distribution parameters of wind and gust velocities. The aim of this procedure is to estimate the parameters of gusts at stations where only wind speed data is available. These parameters can be used to generate synthetic gusts, which can improve the accuracy of return periods at test sites with a lack of observations. The second objective is to determine return periods much longer than the nominal length of the original time series by considering extreme value statistics. Estimates for both local maximum return periods and average return periods for single historical events are provided. The comparison of maximum and average return periods shows that even storms with short average return periods may lead to local wind gusts with return periods of several decades. Despite uncertainties caused by the short length of the observational records, the method leads to consistent results, enabling a wide range of possible applications.

Using reanalysis data to quantify extreme wind power generation statistics : a 33 year case study in Great Britain

With a rapidly increasing fraction of electricity generation being sourced from wind, extreme wind power generation events such as prolonged periods of low (or high) generation and ramps in generation, are a growing concern for the efficient and secure operation of national power systems. As extreme events occur infrequently, long and reliable meteorological records are required to accurately estimate their characteristics. Recent publications have begun to investigate the use of global meteorological “reanalysis” data sets for power system applications, many of which focus on long-term average statistics such as monthly-mean generation. Here we demonstrate that reanalysis data can also be used to estimate the frequency of relatively short-lived extreme events (including ramping on sub-daily time scales). Verification against 328 surface observation stations across the United Kingdom suggests that near-surface wind variability over spatiotemporal scales greater than around 300 km and 6 h can be faithfully reproduced using reanalysis, with no need for costly dynamical downscaling. A case study is presented in which a state-of-the-art, 33 year reanalysis data set (MERRA, from NASA-GMAO), is used to construct an hourly time series of nationally-aggregated wind power generation in Great Britain (GB), assuming a fixed, modern distribution of wind farms. The resultant generation estimates are highly correlated with recorded data from National Grid in the recent period, both for instantaneous hourly values and for variability over time intervals greater than around 6 h. This 33 year time series is then used to quantify the frequency with which different extreme GB-wide wind power generation events occur, as well as their seasonal and inter-annual variability. Several novel insights into the nature of extreme wind power generation events are described, including (i) that the number of prolonged low or high generation events is well approximated by a Poission-like random process, and (ii) whilst in general there is large seasonal variability, the magnitude of the most extreme ramps is similar in both summer and winter. An up-to-date version of the GB case study data as well as the underlying model are freely available for download from our website: http://www.met.reading.ac.uk/~energymet/data/Cannon2014/.

Real-time factor model forecasting and the effects of instability

Factor forecasting models are shown to deliver real-time gains over autoregressive models for US real activity variables during the recent period, but are less successful for nominal variables. The gains are largely due to the Financial Crisis period, and are primarily at the shortest (one quarter ahead) horizon. Excluding the pre-Great Moderation years from the factor forecasting model estimation period (but not from the data used to extract factors) results in a marked fillip in factor model forecast accuracy, but does the same for the AR model forecasts. The relative performance of the factor models compared to the AR models is largely unaffected by whether the exercise is in real time or is pseudo out-of-sample.

Assessing macro uncertainty in real-time when data are subject to revision

Model-based estimates of future uncertainty are generally based on the in-sample fit of the model, as when Box-Jenkins prediction intervals are calculated. However, this approach will generate biased uncertainty estimates in real time when there are data revisions. A simple remedy is suggested, and used to generate more accurate prediction intervals for 25 macroeconomic variables, in line with the theory. A simulation study based on an empirically-estimated model of data revisions for US output growth is used to investigate small-sample properties.

Output sustainability to exogenous and endogenous shocks: evidence from emerging economies

This paper studies the impact of exogenous and endogenous shocks (exogenous shock is used interchangeably with external shock; endogenous shock is used interchangeably with domestic shock) on output fluctuations in post-communist countries during the 2000s. The first part presents the analytical framework and formulates a research hypothesis. The second part presents vector autoregressive estimation and analysis model proposed by Pesaran (2004) and Pesaran and Smith (2006) that relates bank real lending, the cyclical component of output and spreads and accounts for cross-sectional dependence (CD) across the countries. Impulse response functions show that exogenous positive shock lead to a drop in output sustainability for 9 over 12 Central Eastern European countries and Russia, when the endogenous shock is mild and ambiguous. Moreover, the effect of exogenous shock is more significant during the crises. Variance decompositions show that exogenous shock in the aftermath of crisis had a substantial impact on economic activity of emerging economies.

Long-run restrictions and survey forecasts of output, consumption and investment

We consider the extent to which long-horizon survey forecasts of consumption, investment and output growth are consistent with theory-based steady-state values, and whether imposing these restrictions on long-horizon forecasts will enhance their accuracy. The restrictions we impose are consistent with a two-sector model in which the variables grow at different rates in steady state. The restrictions are imposed by exponential-tilting of simple auxiliary forecast densities. We show that imposing the consumption-output restriction yields modest improvements in the long-horizon output growth forecasts, and larger improvements in the forecasts of the cointegrating combination of consumption and output: the transformation of the data on which accuracy is assessed plays an important role.

Future changes of wind energy potentials over Europe in a large CMIP5 multi-model ensemble

A statistical-dynamical downscaling method is used to estimate future changes of wind energy output (Eout) of a benchmark wind turbine across Europe at the regional scale. With this aim, 22 global climate models (GCMs) of the Coupled Model Intercomparison Project Phase 5 (CMIP5) ensemble are considered. The downscaling method uses circulation weather types and regional climate modelling with the COSMO-CLM model. Future projections are computed for two time periods (2021–2060 and 2061–2100) following two scenarios (RCP4.5 and RCP8.5). The CMIP5 ensemble mean response reveals a more likely than not increase of mean annual Eout over Northern and Central Europe and a likely decrease over Southern Europe. There is some uncertainty with respect to the magnitude and the sign of the changes. Higher robustness in future changes is observed for specific seasons. Except from the Mediterranean area, an ensemble mean increase of Eout is simulated for winter and a decreasing for the summer season, resulting in a strong increase of the intra-annual variability for most of Europe. The latter is, in particular, probable during the second half of the 21st century under the RCP8.5 scenario. In general, signals are stronger for 2061–2100 compared to 2021–2060 and for RCP8.5 compared to RCP4.5. Regarding changes of the inter-annual variability of Eout for Central Europe, the future projections strongly vary between individual models and also between future periods and scenarios within single models. This study showed for an ensemble of 22 CMIP5 models that changes in the wind energy potentials over Europe may take place in future decades. However, due to the uncertainties detected in this research, further investigations with multi-model ensembles are needed to provide a better quantification and understanding of the future changes.

Projected changes in tropical cyclones over Vietnam and the South China Sea using a 25 km regional climate model perturbed physics ensemble

The regional climate modelling system PRECIS, was run at 25 km horizontal resolution for 150 years (1949-2099) using global driving data from a five member perturbed physics ensemble (based on the coupled global climate model HadCM3). Output from these simulations was used to investigate projected changes in tropical cyclones (TCs) over Vietnam and the South China Sea due to global warming (under SRES scenario A1B). Thirty year climatological mean periods were used to look at projected changes in future (2069-2098) TCs compared to a 1961-1990 baseline. Present day results were compared qualitatively with IBTrACS observations and found to be reasonably realistic. Future projections show a 20-44 % decrease in TC frequency, although the spatial patterns of change differ between the ensemble members, and an increase of 27-53 % in the amount of TC associated precipitation. No statistically significant changes in TC intensity were found, however, the occurrence of more intense TCs (defined as those with a maximum 10 m wind speed > 35 m/s) was found to increase by 3-9 %. Projected increases in TC associated precipitation are likely caused by increased evaporation and availability of atmospheric water vapour, due to increased sea surface and atmospheric temperature. The mechanisms behind the projected changes in TC frequency are difficult to link explicitly; changes are most likely due to the combination of increased static stability, increased vertical wind shear and decreased upward motion, which suggest a decrease in the tropical overturning circulation.

Interactions between wind-blown snow redistribution and melt ponds in a coupled ocean–sea ice model

Introducing a parameterization of the interactions between wind-driven snow depth changes and melt pond evolution allows us to improve large scale models. In this paper we have implemented an explicit melt pond scheme and, for the first time, a wind dependant snow redistribution model and new snow thermophysics into a coupled ocean–sea ice model. The comparison of long-term mean statistics of melt pond fractions against observations demonstrates realistic melt pond cover on average over Arctic sea ice, but a clear underestimation of the pond coverage on the multi-year ice (MYI) of the western Arctic Ocean. The latter shortcoming originates from the concealing effect of persistent snow on forming ponds, impeding their growth. Analyzing a second simulation with intensified snow drift enables the identification of two distinct modes of sensitivity in the melt pond formation process. First, the larger proportion of wind-transported snow that is lost in leads directly curtails the late spring snow volume on sea ice and facilitates the early development of melt ponds on MYI. In contrast, a combination of higher air temperatures and thinner snow prior to the onset of melting sometimes make the snow cover switch to a regime where it melts entirely and rapidly. In the latter situation, seemingly more frequent on first-year ice (FYI), a smaller snow volume directly relates to a reduced melt pond cover. Notwithstanding, changes in snow and water accumulation on seasonal sea ice is naturally limited, which lessens the impacts of wind-blown snow redistribution on FYI, as compared to those on MYI. At the basin scale, the overall increased melt pond cover results in decreased ice volume via the ice-albedo feedback in summer, which is experienced almost exclusively by MYI.

Bayesian model comparison with un-normalised likelihoods

Models for which the likelihood function can be evaluated only up to a parameter-dependent unknown normalizing constant, such as Markov random field models, are used widely in computer science, statistical physics, spatial statistics, and network analysis. However, Bayesian analysis of these models using standard Monte Carlo methods is not possible due to the intractability of their likelihood functions. Several methods that permit exact, or close to exact, simulation from the posterior distribution have recently been developed. However, estimating the evidence and Bayes’ factors for these models remains challenging in general. This paper describes new random weight importance sampling and sequential Monte Carlo methods for estimating BFs that use simulation to circumvent the evaluation of the intractable likelihood, and compares them to existing methods. In some cases we observe an advantage in the use of biased weight estimates. An initial investigation into the theoretical and empirical properties of this class of methods is presented. Some support for the use of biased estimates is presented, but we advocate caution in the use of such estimates.

Bayesian analysis of uncertainty in the GlobCover 2009 land cover product at climate model grid scale

Land cover data derived from satellites are commonly used to prescribe inputs to models of the land surface. Since such data inevitably contains errors, quantifying how uncertainties in the data affect a model’s output is important. To do so, a spatial distribution of possible land cover values is required to propagate through the model’s simulation. However, at large scales, such as those required for climate models, such spatial modelling can be difficult. Also, computer models often require land cover proportions at sites larger than the original map scale as inputs, and it is the uncertainty in these proportions that this article discusses. This paper describes a Monte Carlo sampling scheme that generates realisations of land cover proportions from the posterior distribution as implied by a Bayesian analysis that combines spatial information in the land cover map and its associated confusion matrix. The technique is computationally simple and has been applied previously to the Land Cover Map 2000 for the region of England and Wales. This article demonstrates the ability of the technique to scale up to large (global) satellite derived land cover maps and reports its application to the GlobCover 2009 data product. The results show that, in general, the GlobCover data possesses only small biases, with the largest belonging to non–vegetated surfaces. In vegetated surfaces, the most prominent area of uncertainty is Southern Africa, which represents a complex heterogeneous landscape. It is also clear from this study that greater resources need to be devoted to the construction of comprehensive confusion matrices.

Coupling an aerosol box model with one-dimensional flow: a tool for understanding observations of new particle formation events

Field observations of new particle formation and the subsequent particle growth are typically only possible at a fixed measurement location, and hence do not follow the temporal evolution of an air parcel in a Lagrangian sense. Standard analysis for determining formation and growth rates requires that the time-dependent formation rate and growth rate of the particles are spatially invariant; air parcel advection means that the observed temporal evolution of the particle size distribution at a fixed measurement location may not represent the true evolution if there are spatial variations in the formation and growth rates. Here we present a zero-dimensional aerosol box model coupled with one-dimensional atmospheric flow to describe the impact of advection on the evolution of simulated new particle formation events. Wind speed, particle formation rates and growth rates are input parameters that can vary as a function of time and location, using wind speed to connect location to time. The output simulates measurements at a fixed location; formation and growth rates of the particle mode can then be calculated from the simulated observations at a stationary point for different scenarios and be compared with the ‘true’ input parameters. Hence, we can investigate how spatial variations in the formation and growth rates of new particles would appear in observations of particle number size distributions at a fixed measurement site. We show that the particle size distribution and growth rate at a fixed location is dependent on the formation and growth parameters upwind, even if local conditions do not vary. We also show that different input parameters used may result in very similar simulated measurements. Erroneous interpretation of observations in terms of particle formation and growth rates, and the time span and areal extent of new particle formation, is possible if the spatial effects are not accounted for.