Probabilistic evaluation of time series models : a comparison of several approaches

Several methods are examined which allow to produce forecasts for time series in the form of probability assignments. The necessary concepts are presented, addressing questions such as how to assess the performance of a probabilistic forecast. A particular class of models, cluster weighted models (CWMs), is given particular attention. CWMs, originally proposed for deterministic forecasts, can be employed for probabilistic forecasting with little modification. Two examples are presented. The first involves estimating the state of (numerically simulated) dynamical systems from noise corrupted measurements, a problem also known as filtering. There is an optimal solution to this problem, called the optimal filter, to which the considered time series models are compared. (The optimal filter requires the dynamical equations to be known.) In the second example, we aim at forecasting the chaotic oscillations of an experimental bronze spring system. Both examples demonstrate that the considered time series models, and especially the CWMs, provide useful probabilistic information about the underlying dynamical relations. In particular, they provide more than just an approximation to the conditional mean.

The impact of installer business models on the uptake of residential microgeneration in the UK: Evidence from a nation-wide survey

This paper arises from a doctoral thesis comparing the impact of alternative installer business models on the rate at which microgeneration is taken up in homes and installation standards across the UK. The paper presents the results of the first large-scale academic survey of businesses certified to install residential microgeneration. The aim is to systematically capture those characteristics which define the business model of each surveyed company, and relate these to the number, location and type of technologies that they install, and the quality of these installations. The methodology comprised a pilot web survey of 235 certified installer businesses, which was carried out in June last year and achieved a response rate of 30%. Following optimisation of the design, the main web survey was emailed to over 2000 businesses between October and December 2011, with 317 valid responses received. The survey is being complemented during summer 2012 by semi-structured interviews with a representative sample of installers who completed the main survey. The survey results are currently being analysed. The early results indicate an emerging and volatile market where solar PV, solar hot water and air source heat pumps are the dominant technologies. Three quarters of respondents are founders of their installer business, while only 22 businesses are owned by another company. Over half of the 317 businesses have five employees or less, while 166 businesses are no more than four years old. In addition, half of the businesses stated that 100% of their employees work on microgeneration-related activities. 85% of the surveyed companies have only one business location in the UK. A third of the businesses are based either in the South West or South East regions of England. This paper outlines the interim results of the survey combined with the outcomes from additional interviews with installers to date. The research identifies some of the business models underpinning microgeneration installers and some of the ways in which installer business models impact on the rate and standards of microgeneration uptake. A tentative conclusion is that installer business models are profoundly dependent on the levels and timing of support from the UK Feed-in Tariffs and Renewable Heat Incentive.

Anti-proliferative effects of physiological concentrations of enterolactone in models of prostate tumourigenesis

SCOPE: There is evidence that a mammalian lignan, enterolactone (ENL), decreases the proliferation rate of prostate cancer cells, although previous studies have used concentrations difficult to achieve through dietary modification. We have therefore investigated the anti-proliferative effects of ENL in an in vitro model of prostate tumourigenesis at concentrations reported to occur in a range of male populations. METHODS AND RESULTS: The effects of 0.1 and 1 μM ENL on three markers of viability and proliferation (metabolic activity, growth kinetics, and cell cycle progression) were assessed in the RWPE-1, WPE1-NA22, WPE1-NB14, WPE1-NB11, WPE1-NB26, LNCaP, and PC-3 cell lines over 72 h. Based on these data, we quantified the expression levels of 12 genes involved in the control of DNA replication initiation using TaqMan real-time PCR in the WPE1-NA22, WPE1-NB14, WPE1-NB11, and WPE1-NB26 cell lines. ENL significantly inhibited the abnormal proliferation of the WPE1-NB14 and WPE1-NB11 cell lines and appears to be a consequence of decreased expression of abnormal chromatin licensing and DNA replication factor 1. CONCLUSION: In contrast to previous studies, concentrations of ENL that are reported after dietary intervention restrict the proliferation of early-stage tumourigenic prostate cell lines by inhibiting the abnormal formation of complexes that initiate DNA replication.

Fit for policy? Some evidence on the application of development viability models in the UK planning system

This paper investigates the application and use of development viability models in the formation of planning policies in the UK. Particular attention is paid to three key areas; the assumed development scheme in development viability models, the use of forecasts and the debate concerning Threshold Land Value. The empirical section reports on the results of an interview survey involving the main producers of development viability models and appraisals. It is concluded that, although development viability models have intrinsic limitations associated with model composition and input uncertainties, the most significant limitations are related to the ways that they have been adapted for use in the planning system. In addition, it is suggested that the contested nature of Threshold Land Value is an example of calculative practices providing a façade of technocratic rationality in the planning system.

The role of rainfall intensity and soil in determining rates of flow through cryoturbated chalk

The effects of different water application rates (3, 10, 15 and 30 mm/h) and of topsoil removal on the rate of downward water movement through the cryoturbated chalk zone in southern England were investigated in situ. During and after each application of water, changes in water content and matric potential of the profile were monitored and percolate was collected in troughs. The measured water breakthrough time showed that water moved to 1.2 m depth quickly (in 8.2 h) even with application rate as low as 3 mm/h and that the time was only 3 h when water was applied at a rate of 15 mm/ h. These breakthrough times were about 150 and 422 fold shorter, respectively, than those expected if the water had been conducted by the matrix alone. Percolate was collected in troughs within 3.5 h at 1.2 m depth when water was applied at 30 mm/h and the quantity collected indicated that a significant amount of the surface applied water moved downward through inter-aggregate pores. The small increase in volumetric water content (about 3%) in excess of matrix water content resulted in a large increase in pore water velocities, from 0.20 to 5.3 m/d. The presence of soil layer had effect on the time taken for water to travel through the cryoturbated chalk layer and in the soil layer, water took about 1-2 h to pass thorough, depending on the intensity.

Northern winter stratospheric temperature and ozone responses to ENSO inferred from an ensemble of Chemistry Climate Models

The connection between the El Ni˜no Southern Oscillation (ENSO) and the Northern polar stratosphere has been established from observations and atmospheric modeling. Here a systematic inter-comparison of the sensitivity of the modeled stratosphere to ENSO in Chemistry Climate Models (CCMs) is reported. This work uses results from a number of the CCMs included in the 2006 ozone assessment. In the lower stratosphere, the mean of all model simulations reports a warming of the polar vortex during strong ENSO events in February–March, consistent with but smaller than the estimate from satellite observations and ERA40 reanalysis. The anomalous warming is associated with an anomalous dynamical increase of column ozone north of 70� N that is accompanied by coherent column ozone decrease in the Tropics, in agreement with that deduced from the NIWA column ozone database, implying an increased residual circulation in the mean of all model simulations during ENSO. The spread in the model responses is partly due to the large internal stratospheric variability and it is shown that it crucially depends on the representation of the tropospheric ENSO teleconnection in the models.

A strategy for process-oriented validation of coupled chemistry–climate models

Evaluating CCMs with the presented framework will increase our confidence in predictions of stratospheric ozone change.

Uncertainties and assessments of chemistry-climate models of the stratosphere

In recent years a number of chemistry-climate models have been developed with an emphasis on the stratosphere. Such models cover a wide range of time scales of integration and vary considerably in complexity. The results of specific diagnostics are here analysed to examine the differences amongst individual models and observations, to assess the consistency of model predictions, with a particular focus on polar ozone. For example, many models indicate a significant cold bias in high latitudes, the “cold pole problem”, particularly in the southern hemisphere during winter and spring. This is related to wave propagation from the troposphere which can be improved by improving model horizontal resolution and with the use of non-orographic gravity wave drag. As a result of the widely differing modelled polar temperatures, different amounts of polar stratospheric clouds are simulated which in turn result in varying ozone values in the models. The results are also compared to determine the possible future behaviour of ozone, with an emphasis on the polar regions and mid-latitudes. All models predict eventual ozone recovery, but give a range of results concerning its timing and extent. Differences in the simulation of gravity waves and planetary waves as well as model resolution are likely major sources of uncertainty for this issue. In the Antarctic, the ozone hole has probably reached almost its deepest although the vertical and horizontal extent of depletion may increase slightly further over the next few years. According to the model results, Antarctic ozone recovery could begin any year within the range 2001 to 2008. The limited number of models which have been integrated sufficiently far indicate that full recovery of ozone to 1980 levels may not occur in the Antarctic until about the year 2050. For the Arctic, most models indicate that small ozone losses may continue for a few more years and that recovery could begin any year within the range 2004 to 2019. The start of ozone recovery in the Arctic is therefore expected to appear later than in the Antarctic.

The ModFOLD4 server for the quality assessment of 3D protein models

Once you have generated a 3D model of a protein, how do you know whether it bears any resemblance to the actual structure? To determine the usefulness of 3D models of proteins, they must be assessed in terms of their quality by methods that predict their similarity to the native structure. The ModFOLD4 server is the latest version of our leading independent server for the estimation of both the global and local (per-residue) quality of 3D protein models. The server produces both machine readable and graphical output, providing users with intuitive visual reports on the quality of predicted protein tertiary structures. The ModFOLD4 server is freely available to all at: http://www.reading.ac.uk/bioinf/ModFOLD/.

Historical and idealized climate model experiments: an intercomparison of Earth system models of intermediate complexity

Both historical and idealized climate model experiments are performed with a variety of Earth system models of intermediate complexity (EMICs) as part of a community contribution to the Intergovernmental Panel on Climate Change Fifth Assessment Report. Historical simulations start at 850 CE and continue through to 2005. The standard simulations include changes in forcing from solar luminosity, Earth's orbital configuration, CO2, additional greenhouse gases, land use, and sulphate and volcanic aerosols. In spite of very different modelled pre-industrial global surface air temperatures, overall 20th century trends in surface air temperature and carbon uptake are reasonably well simulated when compared to observed trends. Land carbon fluxes show much more variation between models than ocean carbon fluxes, and recent land fluxes appear to be slightly underestimated. It is possible that recent modelled climate trends or climate–carbon feedbacks are overestimated resulting in too much land carbon loss or that carbon uptake due to CO2 and/or nitrogen fertilization is underestimated. Several one thousand year long, idealized, 2 × and 4 × CO2 experiments are used to quantify standard model characteristics, including transient and equilibrium climate sensitivities, and climate–carbon feedbacks. The values from EMICs generally fall within the range given by general circulation models. Seven additional historical simulations, each including a single specified forcing, are used to assess the contributions of different climate forcings to the overall climate and carbon cycle response. The response of surface air temperature is the linear sum of the individual forcings, while the carbon cycle response shows a non-linear interaction between land-use change and CO2 forcings for some models. Finally, the preindustrial portions of the last millennium simulations are used to assess historical model carbon-climate feedbacks. Given the specified forcing, there is a tendency for the EMICs to underestimate the drop in surface air temperature and CO2 between the Medieval Climate Anomaly and the Little Ice Age estimated from palaeoclimate reconstructions. This in turn could be a result of unforced variability within the climate system, uncertainty in the reconstructions of temperature and CO2, errors in the reconstructions of forcing used to drive the models, or the incomplete representation of certain processes within the models. Given the forcing datasets used in this study, the models calculate significant land-use emissions over the pre-industrial period. This implies that land-use emissions might need to be taken into account, when making estimates of climate–carbon feedbacks from palaeoclimate reconstructions.

Toward vicarious calibration of microwave remote-sensing satellites in arid environments

The Soil Moisture and Ocean Salinity (SMOS) satellite marks the commencement of dedicated global surface soil moisture missions, and the first mission to make passive microwave observations at L-band. On-orbit calibration is an essential part of the instrument calibration strategy, but on-board beam-filling targets are not practical for such large apertures. Therefore, areas to serve as vicarious calibration targets need to be identified. Such sites can only be identified through field experiments including both in situ and airborne measurements. For this purpose, two field experiments were performed in central Australia. Three areas are studied as follows: 1) Lake Eyre, a typically dry salt lake; 2) Wirrangula Hill, with sparse vegetation and a dense cover of surface rock; and 3) Simpson Desert, characterized by dry sand dunes. Of those sites, only Wirrangula Hill and the Simpson Desert are found to be potentially suitable targets, as they have a spatial variation in brightness temperatures of <4 K under normal conditions. However, some limitations are observed for the Simpson Desert, where a bias of 15 K in vertical and 20 K in horizontal polarization exists between model predictions and observations, suggesting a lack of understanding of the underlying physics in this environment. Subsequent comparison with model predictions indicates a SMOS bias of 5 K in vertical and 11 K in horizontal polarization, and an unbiased root mean square difference of 10 K in both polarizations for Wirrangula Hill. Most importantly, the SMOS observations show that the brightness temperature evolution is dominated by regular seasonal patterns and that precipitation events have only little impact.

Theory and generalization of Monte Carlo models of the BOLD signal source

The dependency of the blood oxygenation level dependent (BOLD) signal on underlying hemodynamics is not well understood. Building a forward biophysical model of this relationship is important for the quantitative estimation of the hemodynamic changes and neural activity underlying functional magnetic resonance imaging (fMRI) signals. We have developed a general model of the BOLD signal which can model both intra- and extravascular signals for an arbitrary tissue model across a wide range of imaging parameters. The model of the BOLD signal was instantiated as a look-up-table (LuT), and was verified against concurrent fMRI and optical imaging measurements of activation induced hemodynamics. Magn Reson Med, 2008. © 2008 Wiley-Liss, Inc.

A cross-calibration of GMS-5 thermal channels against ATSR-2

Motivated by the importance to weather and climate of the Indo-Pacific seas, we present a new calibration of the Visible Infrared Spin-Scan Radiometer (VISSR) on the geostationary meteorological satellite, GMS-5. VISSR imagery has significant potential for exploring the dynamics of the ocean and air–sea interactions in this poorly characterized region, by virtue of the VISSR's surface temperature retrieval capability and hourly sampling. However, the calibration of the thermal imagery supplied by the Japanese Meteorological Agency (JMA) is inconsistent with the spectral characteristics of the channels, and published details of the JMA calibration procedure are scant. We use the well-characterized Along-Track Scanning Radiometer 2 (ATSR-2) as a reference, and determine calibration corrections for GMS-5 VISSR. We obtain more credible VISSR brightness temperatures and demonstrate sea surface temperature (SST) retrieval that validates well against in situ measurements (bias ∼0.3 and scatter ∼0.4 K). Comparison with a widely used sea surface temperature analysis shows that the GMS-5 VISSR SST fields capture important spatial structure, absent in the analysis.

On the lack of stratospheric dynamical variability in low-top versions of the CMIP5 models

We describe the main differences in simulations of stratospheric climate and variability by models within the fifth Coupled Model Intercomparison Project (CMIP5) that have a model top above the stratopause and relatively fine stratospheric vertical resolution (high-top), and those that have a model top below the stratopause (low-top). Although the simulation of mean stratospheric climate by the two model ensembles is similar, the low-top model ensemble has very weak stratospheric variability on daily and interannual time scales. The frequency of major sudden stratospheric warming events is strongly underestimated by the low-top models with less than half the frequency of events observed in the reanalysis data and high-top models. The lack of stratospheric variability in the low-top models affects their stratosphere-troposphere coupling, resulting in short-lived anomalies in the Northern Annular Mode, which do not produce long-lasting tropospheric impacts, as seen in observations. The lack of stratospheric variability, however, does not appear to have any impact on the ability of the low-top models to reproduce past stratospheric temperature trends. We find little improvement in the simulation of decadal variability for the high-top models compared to the low-top, which is likely related to the fact that neither ensemble produces a realistic dynamical response to volcanic eruptions.

On the difficulties of present theoretical models to predict the oxidation state of atomic Au adsorbed on regular sites of CeO2(111)

The electronic structure and oxidation state of atomic Au adsorbed on a perfect CeO2(111) surface have been investigated in detail by means of periodic density functional theory-based calculations, using the LDA+U and GGA+U potentials for a broad range of U values, complemented with calculations employing the HSE06 hybrid functional. In addition, the effects of the lattice parameter a0 and of the starting point for the geometry optimization have also been analyzed. From the present results we suggest that the oxidation state of single Au atoms on CeO2(111) predicted by LDA+U, GGA+U, and HSE06 density functional calculations is not conclusive and that the final picture strongly depends on the method chosen and on the construction of the surface model. In some cases we have been able to locate two well-defined states which are close in energy but with very different electronic structure and local geometries, one with Au fully oxidized and one with neutral Au. The energy difference between the two states is typically within the limits of the accuracy of the present exchange-correlation potentials, and therefore, a clear lowest-energy state cannot be identified. These results suggest the possibility of a dynamic distribution of Au0 and Au+ atomic species at the regular sites of the CeO2(111) surface.