Rank histograms of stratified Monte-Carlo ensembles

The application of forecast ensembles to probabilistic weather prediction has spurred considerable interest in their evaluation. Such ensembles are commonly interpreted as Monte Carlo ensembles meaning that the ensemble members are perceived as random draws from a distribution. Under this interpretation, a reasonable property to ask for is statistical consistency, which demands that the ensemble members and the verification behave like draws from the same distribution. A widely used technique to assess statistical consistency of a historical dataset is the rank histogram, which uses as a criterion the number of times that the verification falls between pairs of members of the ordered ensemble. Ensemble evaluation is rendered more specific by stratification, which means that ensembles that satisfy a certain condition (e.g., a certain meteorological regime) are evaluated separately. Fundamental relationships between Monte Carlo ensembles, their rank histograms, and random sampling from the probability simplex according to the Dirichlet distribution are pointed out. Furthermore, the possible benefits and complications of ensemble stratification are discussed. The main conclusion is that a stratified Monte Carlo ensemble might appear inconsistent with the verification even though the original (unstratified) ensemble is consistent. The apparent inconsistency is merely a result of stratification. Stratified rank histograms are thus not necessarily flat. This result is demonstrated by perfect ensemble simulations and supplemented by mathematical arguments. Possible methods to avoid or remove artifacts that stratification induces in the rank histogram are suggested.

Futures basis, inventory and commodity price volatility: an empirical analysis

We employ a large dataset of physical inventory data on 21 different commodities for the period 1993–2011 to empirically analyze the behavior of commodity prices and their volatility as predicted by the theory of storage. We examine two main issues. First, we analyze the relationship between inventory and the shape of the forward curve. Low (high) inventory is associated with forward curves in backwardation (contango), as the theory of storage predicts. Second, we show that price volatility is a decreasing function of inventory for the majority of commodities in our sample. This effect is more pronounced in backwardated markets. Our findings are robust with respect to alternative inventory measures and over the recent commodity price boom.

P-Prism: a computationally efficient approach to scaling up classification rule induction

Top Down Induction of Decision Trees (TDIDT) is the most commonly used method of constructing a model from a dataset in the form of classification rules to classify previously unseen data. Alternative algorithms have been developed such as the Prism algorithm. Prism constructs modular rules which produce qualitatively better rules than rules induced by TDIDT. However, along with the increasing size of databases, many existing rule learning algorithms have proved to be computational expensive on large datasets. To tackle the problem of scalability, parallel classification rule induction algorithms have been introduced. As TDIDT is the most popular classifier, even though there are strongly competitive alternative algorithms, most parallel approaches to inducing classification rules are based on TDIDT. In this paper we describe work on a distributed classifier that induces classification rules in a parallel manner based on Prism.

Experimental and modeled UV erthemal irradiance under overcast conditions: the role of cloud optical depth

This paper evaluates the relationship between the cloud modification factor (CMF) in the ultraviolet erythe- mal range and the cloud optical depth (COD) retrieved from the Aerosol Robotic Network (AERONET) "cloud mode" algorithm under overcast cloudy conditions (confirmed with sky images) at Granada, Spain, mainly for non-precipitating, overcast and relatively homogenous water clouds. Empirical CMF showed a clear exponential dependence on experimental COD values, decreasing approximately from 0.7 for COD=10 to 0.25 for COD=50. In addition, these COD measurements were used as input in the LibRadtran radia tive transfer code allowing the simulation of CMF values for the selected overcast cases. The modeled CMF exhibited a dependence on COD similar to the empirical CMF, but modeled values present a strong underestimation with respect to the empirical factors (mean bias of 22 %). To explain this high bias, an exhaustive comparison between modeled and experimental UV erythemal irradiance (UVER) data was performed. The comparison revealed that the radiative transfer simulations were 8 % higher than the observations for clear-sky conditions. The rest of the bias (~14 %) may be attributed to the substantial underestimation of modeled UVER with respect to experimental UVER under overcast conditions, although the correlation between both dataset was high (R2 ~ 0.93). A sensitive test showed that the main reason responsible for that underestimation is the experimental AERONET COD used as input in the simulations, which has been retrieved from zenith radiances in the visible range. In this sense, effective COD in the erythemal interval were derived from an iteration procedure based on searching the best match between modeled and experimental UVER values for each selected overcast case. These effective COD values were smaller than AERONET COD data in about 80 % of the overcast cases with a mean relative difference of 22 %.

Identifying human influences on atmospheric temperature

We perform a multimodel detection and attribution study with climate model simulation output and satellite-based measurements of tropospheric and stratospheric temperature change. We use simulation output from 20 climate models participating in phase 5 of the Coupled Model Intercomparison Project. This multimodel archive provides estimates of the signal pattern in response to combined anthropogenic and natural external forcing (the finger-print) and the noise of internally generated variability. Using these estimates, we calculate signal-to-noise (S/N) ratios to quantify the strength of the fingerprint in the observations relative to fingerprint strength in natural climate noise. For changes in lower stratospheric temperature between 1979 and 2011, S/N ratios vary from 26 to 36, depending on the choice of observational dataset. In the lower troposphere, the fingerprint strength in observations is smaller, but S/N ratios are still significant at the 1% level or better, and range from three to eight. We find no evidence that these ratios are spuriously inflated by model variability errors. After removing all global mean signals, model fingerprints remain identifiable in 70% of the tests involving tropospheric temperature changes. Despite such agreement in the large-scale features of model and observed geographical patterns of atmospheric temperature change, most models do not replicate the size of the observed changes. On average, the models analyzed underestimate the observed cooling of the lower stratosphere and overestimate the warming of the troposphere. Although the precise causes of such differences are unclear, model biases in lower stratospheric temperature trends are likely to be reduced by more realistic treatment of stratospheric ozone depletion and volcanic aerosol forcing.

Beyond the computer: changing medium from digital to physical

How can organizations use digital infrastructure to realise physical outcomes? The design and construction of London Heathrow Terminal 5 is analysed to build new theoretical understanding of visualization and materialization practices in the transition from digital design to physical realisation. In the project studied, an integrated software solution is introduced as an infrastructure for delivery. The analyses articulate the work done to maintain this digital infrastructure and also to move designs beyond the closed world of the computer to a physical reality. In changing medium, engineers use heterogeneous trials to interrogate and address the limitations of an integrated digital model. The paper explains why such trials, which involve the reconciliation of digital and physical data through parallel and iterative forms of work, provide a robust practice for realizing goals that have physical outcomes. It argues that this practice is temporally different from, and at times in conflict with, building a comprehensive dataset within the digital medium. The paper concludes by discussing the implications for organizations that use digital infrastructures in seeking to accomplish goals in digital and physical media.

The short- and long-term fitness consequences of natal dispersal in a wild bird population

Dispersal is a key process in population and evolutionary ecology. Individual decisions are affected by fitness consequences of dispersal, but these are difficult to measure in wild populations. A long-term dataset on a geographically closed bird population, the Mauritius kestrel, offers a rare opportunity to explore fitness consequences. Females dispersed further when the availability of local breeding sites was limited, whereas male dispersal correlated with phenotypic traits. Female but not male fitness was lower when they dispersed longer distances compared to settling close to home. These results suggest a cost of dispersal in females. We found evidence of both short- and long-term fitness consequences of natal dispersal in females, including reduced fecundity in early life and more rapid aging in later life. Taken together, our results indicate that dispersal in early life might shape life history strategies in wild populations.

Examining the large-scale spatial coherence of European drought using regional indicators of precipitation and streamflow deficit

Droughts tend to evolve slowly and affect large areas simultaneously, which suggests that improved understanding of spatial coherence of drought would enable better mitigation of drought impacts through enhanced monitoring and forecasting strategies. This study employs an up-to-date dataset of over 500 river flow time series from 11 European countries, along with a gridded precipitation dataset, to examine the spatial coherence of drought in Europe using regional indicators of precipitation and streamflow deficit. The drought indicators were generated for 24 homogeneous regions and, for selected regions, historical drought characteristics were corroborated with previous work. The spatial coherence of drought characteristics was then examined at a European scale. Historical droughts generally have distinctive signatures in their spatio-temporal development, so there was limited scope for using the evolution of historical events to inform forecasting. Rather, relationships were explored in time series of drought indicators between regions. Correlations were generally low, but multivariate analyses revealed broad continental-scale patterns, which appear to be related to large-scale atmospheric circulation indices (in particular, the North Atlantic Oscillation and the East Atlantic West Russia pattern). A novel methodology for forecasting was developed (and demonstrated with reference to the United Kingdom), which predicts drought from drought i.e. uses spatial coherence of drought to facilitate early warning of drought in a target region, from drought which is developing elsewhere in Europe.Whilst the skill of the methodology is relatively modest at present, this approach presents a potential new avenue for forecasting, which offers significant advantages in that it allows prediction for all seasons, and also shows some potential for forecasting the termination of drought conditions.

Risk premia in covered bond markets

The authors empirically explore risk premia in mortgage covered bond markets. Using a large panel dataset of covered bond asset swap spreads, they study the impact of different legal and economic environments. Conducting an in-depth analysis of this market, the authors find significant but small differences between countries during normal market periods. However, these differences are much stronger during times of economic crisis. Moreover, they find that developments in the real estate market are of relatively little importance during stable market periods. During economic distress, however, they are of high importance for explaining risk premia in covered bond markets

What determines religious school choice? Theory and evidence from rural Bangladesh

This paper looks at the determinants of school selection in rural Bangladesh, focusing on the choice between registered Islamic and non-religious schools. Using a unique dataset on secondary school-age children from rural Bangladesh, we find that madrasah enrolment falls as household income increases. At the same time, more religious households, and those that live further away from a non-religious school are more likely to send their children to madrasahs. However, in contrast to the theory, we find that Islamic school demand does not respond to the average quality of schools in the locality.

Benchmarking energy use in licensed restaurants and pubs

Commercial kitchens often leave a large carbon footprint. A new dataset of energy performance metrics from a leading industrial partner is presented. Categorising these types of buildings is challenging. Electricity use has been analysed using data from automated meter readings (AMR) for the purpose of benchmarking and discussed in terms of factors such as size and food output. From the analysed results, consumption is found to be almost double previous sector estimates of 6480 million kWh per year. Recommendations are made to further improve the current benchmarks in order to attain robust, reliable and transparent figures, such as the introduction of normalised performance indicators to include kitchen size (m2) and kWh per thousand-pound turnover.

Stratospheric climate and variability from a general circulation model and observations

The climate and natural variability of the large-scale stratospheric circulation simulated by a newly developed general circulation model are evaluated against available global observations. The simulation consisted of a 30-year annual cycle integration performed with a comprehensive model of the troposphere and stratosphere. The observations consisted of a 15-year dataset from global operational analyses of the troposphere and stratosphere. The model evaluation concentrates on the simulation of the evolution of the extratropical stratospheric circulation in both hemispheres. The December–February climatology of the observed zonal mean winter circulation is found to be reasonably well captured by the model, although in the Northern Hemisphere upper stratosphere the simulated westerly winds are systematically stronger and a cold bias is apparent in the polar stratosphere. This Northern Hemisphere stratospheric cold bias virtually disappears during spring (March–May), consistent with a realistic simulation of the spring weakening of the mean westerly winds in the model. A considerable amount of monthly interannual variability is also found in the simulation in the Northern Hemisphere in late winter and early spring. The simulated interannual variability is predominantly caused by polar warmings of the stratosphere, in agreement with observations. The breakdown of the Northern Hemisphere stratospheric polar vortex appears therefore to occur in a realistic way in the model. However, in early winter the model severely underestimates the interannual variability, especially in the upper troposphere. The Southern Hemisphere winter (June–August) zonal mean temperature is systematically colder in the model, and the simulated winds are somewhat too strong in the upper stratosphere. Contrary to the results for the Northern Hemisphere spring, this model cold bias worsens during the Southern Hemisphere spring (September–November). Significant discrepancies between the model results and the observations are therefore found during the breakdown of the Southern Hemisphere polar vortex. For instance, the simulated Southern Hemisphere stratosphere westerly jet continuously decreases in intensity more or less in situ from June to November, while the observed stratospheric jet moves downward and poleward.

The effect of nonlinearity in CO2 heating rates on the attribution of stratospheric ozone and temperature changes

An analysis of the attribution of past and future changes in stratospheric ozone and temperature to anthropogenic forcings is presented. The analysis is an extension of the study of Shepherd and Jonsson (2008) who analyzed chemistry-climate simulations from the Canadian Middle Atmosphere Model (CMAM) and attributed both past and future changes to changes in the external forcings, i.e. the abundances of ozone-depleting substances (ODS) and well-mixed greenhouse gases. The current study is based on a new CMAM dataset and includes two important changes. First, we account for the nonlinear radiative response to changes in CO2. It is shown that over centennial time scales the radiative response in the upper stratosphere to CO2 changes is significantly nonlinear and that failure to account for this effect leads to a significant error in the attribution. To our knowledge this nonlinearity has not been considered before in attribution analysis, including multiple linear regression studies. For the regression analysis presented here the nonlinearity was taken into account by using CO2 heating rate, rather than CO2 abundance, as the explanatory variable. This approach yields considerable corrections to the results of the previous study and can be recommended to other researchers. Second, an error in the way the CO2 forcing changes are implemented in the CMAM was corrected, which significantly affects the results for the recent past. As the radiation scheme, based on Fomichev et al. (1998), is used in several other models we provide some description of the problem and how it was fixed.

Ozone database in support of CMIP5 simulations: results and corresponding radiative forcing

A continuous tropospheric and stratospheric vertically resolved ozone time series, from 1850 to 2099, has been generated to be used as forcing in global climate models that do not include interactive chemistry. A multiple linear regression analysis of SAGE I+II satellite observations and polar ozonesonde measurements is used for the stratospheric zonal mean dataset during the well-observed period from 1979 to 2009. In addition to terms describing the mean annual cycle, the regression includes terms representing equivalent effective stratospheric chlorine (EESC) and the 11-yr solar cycle variability. The EESC regression fit coefficients, together with pre-1979 EESC values, are used to extrapolate the stratospheric ozone time series backward to 1850. While a similar procedure could be used to extrapolate into the future, coupled chemistry climate model (CCM) simulations indicate that future stratospheric ozone abundances are likely to be significantly affected by climate change, and capturing such effects through a regression model approach is not feasible. Therefore, the stratospheric ozone dataset is extended into the future (merged in 2009) with multimodel mean projections from 13 CCMs that performed a simulation until 2099 under the SRES (Special Report on Emission Scenarios) A1B greenhouse gas scenario and the A1 adjusted halogen scenario in the second round of the Chemistry-Climate Model Validation (CCMVal-2) Activity. The stratospheric zonal mean ozone time series is merged with a three-dimensional tropospheric data set extracted from simulations of the past by two CCMs (CAM3.5 and GISSPUCCINI)and of the future by one CCM (CAM3.5). The future tropospheric ozone time series continues the historical CAM3.5 simulation until 2099 following the four different Representative Concentration Pathways (RCPs). Generally good agreement is found between the historical segment of the ozone database and satellite observations, although it should be noted that total column ozone is overestimated in the southern polar latitudes during spring and tropospheric column ozone is slightly underestimated. Vertical profiles of tropospheric ozone are broadly consistent with ozonesondes and in-situ measurements, with some deviations in regions of biomass burning. The tropospheric ozone radiative forcing (RF) from the 1850s to the 2000s is 0.23Wm−2, lower than previous results. The lower value is mainly due to (i) a smaller increase in biomass burning emissions; (ii) a larger influence of stratospheric ozone depletion on upper tropospheric ozone at high southern latitudes; and possibly (iii) a larger influence of clouds (which act to reduce the net forcing) compared to previous radiative forcing calculations. Over the same period, decreases in stratospheric ozone, mainly at high latitudes, produce a RF of −0.08Wm−2, which is more negative than the central Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) value of −0.05Wm−2, but which is within the stated range of −0.15 to +0.05Wm−2. The more negative value is explained by the fact that the regression model simulates significant ozone depletion prior to 1979, in line with the increase in EESC and as confirmed by CCMs, while the AR4 assumed no change in stratospheric RF prior to 1979. A negative RF of similar magnitude persists into the future, although its location shifts from high latitudes to the tropics. This shift is due to increases in polar stratospheric ozone, but decreases in tropical lower stratospheric ozone, related to a strengthening of the Brewer-Dobson circulation, particularly through the latter half of the 21st century. Differences in trends in tropospheric ozone among the four RCPs are mainly driven by different methane concentrations, resulting in a range of tropospheric ozone RFs between 0.4 and 0.1Wm−2 by 2100. The ozone dataset described here has been released for the Coupled Model Intercomparison Project (CMIP5) model simulations in netCDF Climate and Forecast (CF) Metadata Convention at the PCMDI website (http://cmip-pcmdi.llnl.gov/).

Summertime stratospheric processes at northern mid-latitudes: comparisons between MANTRA balloon measurements and the Canadian Middle Atmosphere Model

In this paper we report on a study conducted using the Middle Atmospheric Nitrogen TRend Assessment (MANTRA) balloon measurements of stratospheric constituents and temperature and the Canadian Middle Atmosphere Model (CMAM). Three different kinds of data are used to assess the inter-consistency of the combined dataset: single profiles of long-lived species from MANTRA 1998, sparse climatologies from the ozonesonde measurements during the four MANTRA campaigns and from HALOE satellite measurements, and the CMAM climatology. In doing so, we evaluate the ability of the model to reproduce the measured fields and to thereby test our ability to describe mid-latitude summertime stratospheric processes. The MANTRA campaigns were conducted at Vanscoy, Saskatchewan, Canada (52◦ N, 107◦ W)in late August and early September of 1998, 2000, 2002 and 2004. During late summer at mid-latitudes, the stratosphere is close to photochemical control, providing an ideal scenario for the study reported here. From this analysis we find that: (1) reducing the value for the vertical diffusion coefficient in CMAM to a more physically reasonable value results in the model better reproducing the measured profiles of long-lived species; (2) the existence of compact correlations among the constituents, as expected from independent measurements in the literature and from models, confirms the self-consistency of the MANTRA measurements; and (3) the 1998 measurements show structures in the chemical species profiles that can be associated with transport, adding to the growing evidence that the summertime stratosphere can be much more disturbed than anticipated. The mechanisms responsible for such disturbances need to be understood in order to assess the representativeness of the measurements and to isolate longterm trends.