FAMOUS, faster: using parallel computing techniques to accelerate the FAMOUS/HadCM3 climate model with a focus on the radiative transfer algorithm

We have optimised the atmospheric radiation algorithm of the FAMOUS climate model on several hardware platforms. The optimisation involved translating the Fortran code to C and restructuring the algorithm around the computation of a single air column. Instead of the existing MPI-based domain decomposition, we used a task queue and a thread pool to schedule the computation of individual columns on the available processors. Finally, four air columns are packed together in a single data structure and computed simultaneously using Single Instruction Multiple Data operations. The modified algorithm runs more than 50 times faster on the CELL’s Synergistic Processing Elements than on its main PowerPC processing element. On Intel-compatible processors, the new radiation code runs 4 times faster. On the tested graphics processor, using OpenCL, we find a speed-up of more than 2.5 times as compared to the original code on the main CPU. Because the radiation code takes more than 60% of the total CPU time, FAMOUS executes more than twice as fast. Our version of the algorithm returns bit-wise identical results, which demonstrates the robustness of our approach. We estimate that this project required around two and a half man-years of work.

An assessment of observed vertical flux divergence in long-term eddy-covariance measurements over two Midwestern forest ecosystems

Vertical divergence of CO2 fluxes is observed over two Midwestern AmeriFlux forest sites. The differences in ensemble averaged hourly CO2 fluxes measured at two heights above canopy are relatively small (0.2–0.5 μmol m−2 s−1), but they are the major contributors to differences (76–256 g C m−2 or 41.8–50.6%) in estimated annual net ecosystem exchange (NEE) in 2001. A friction velocity criterion is used in these estimates but mean flow advection is not accounted for. This study examines the effects of coordinate rotation, averaging time period, sampling frequency and co-spectral correction on CO2 fluxes measured at a single height, and on vertical flux differences measured between two heights. Both the offset in measured vertical velocity and the downflow/upflow caused by supporting tower structures in upwind directions lead to systematic over- or under-estimates of fluxes measured at a single height. An offset of 1 cm s−1 and an upflow/downflow of 1° lead to 1% and 5.6% differences in momentum fluxes and nighttime sensible heat and CO2 fluxes, respectively, but only 0.5% and 2.8% differences in daytime sensible heat and CO2 fluxes. The sign and magnitude of both offset and upflow/downflow angle vary between sonic anemometers at two measurement heights. This introduces a systematic and large bias in vertical flux differences if these effects are not corrected in the coordinate rotation. A 1 h averaging time period is shown to be appropriate for the two sites. In the daytime, the absolute magnitudes of co-spectra decrease with height in the natural frequencies of 0.02–0.1 Hz but increase in the lower frequencies (<0.01 Hz). Thus, air motions in these two frequency ranges counteract each other in determining vertical flux differences, whose magnitude and sign vary with averaging time period. At night, co-spectral densities of CO2 are more positive at the higher levels of both sites in the frequency range of 0.03–0.4 Hz and this vertical increase is also shown at most frequencies lower than 0.03 Hz. Differences in co-spectral corrections at the two heights lead to a positive shift in vertical CO2 flux differences throughout the day at both sites. At night, the vertical CO2 flux differences between two measurement heights are 20–30% and 40–60% of co-spectral corrected CO2 fluxes measured at the lower levels of the two sites, respectively. Vertical differences of CO2 flux are relatively small in the daytime. Vertical differences in estimated mean vertical advection of CO2 between the two measurement heights generally do not improve the closure of the 1D (vertical) CO2 budget in the air layer between the two measurement heights. This may imply the significance of horizontal advection. However, a reliable assessment of mean advection contributions in annual NEE estimate at these two AmeriFlux sites is currently an unsolved problem.

Multi-scale sensible heat fluxes in the urban environment from large aperture scintillometry and eddy covariance

Sensible heat fluxes (QH) are determined using scintillometry and eddy covariance over a suburban area. Two large aperture scintillometers provide spatially integrated fluxes across path lengths of 2.8 km and 5.5 km over Swindon, UK. The shorter scintillometer path spans newly built residential areas and has an approximate source area of 2-4 km2, whilst the long path extends from the rural outskirts to the town centre and has a source area of around 5-10 km2. These large-scale heat fluxes are compared with local-scale eddy covariance measurements. Clear seasonal trends are revealed by the long duration of this dataset and variability in monthly QH is related to the meteorological conditions. At shorter time scales the response of QH to solar radiation often gives rise to close agreement between the measurements, but during times of rapidly changing cloud cover spatial differences in the net radiation (Q*) coincide with greater differences between heat fluxes. For clear days QH lags Q*, thus the ratio of QH to Q* increases throughout the day. In summer the observed energy partitioning is related to the vegetation fraction through use of a footprint model. The results demonstrate the value of scintillometry for integrating surface heterogeneity and offer improved understanding of the influence of anthropogenic materials on surface-atmosphere interactions.

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.

A peak reduction scheduling algorithm for storage devices on the low voltage network

Reinforcing the Low Voltage (LV) distribution network will become essential to ensure it remains within its operating constraints as demand on the network increases. The deployment of energy storage in the distribution network provides an alternative to conventional reinforcement. This paper presents a control methodology for energy storage to reduce peak demand in a distribution network based on day-ahead demand forecasts and historical demand data. The control methodology pre-processes the forecast data prior to a planning phase to build in resilience to the inevitable errors between the forecasted and actual demand. The algorithm uses no real time adjustment so has an economical advantage over traditional storage control algorithms. Results show that peak demand on a single phase of a feeder can be reduced even when there are differences between the forecasted and the actual demand. In particular, results are presented that demonstrate when the algorithm is applied to a large number of single phase demand aggregations that it is possible to identify which of these aggregations are the most suitable candidates for the control methodology.

Aircraft-based observations of air–sea turbulent fluxes around the British Isles

Observations of turbulent fluxes of momentum, heat and moisture from low-level aircraft data are presented. Fluxes are calculated using the eddy covariance technique from flight legs typically ∼40 m above the sea surface. Over 400 runs of 2 min (∼12 km) from 26 flights are evaluated. Flight legs are mainly from around the British Isles although a small number are from around Iceland and Norway. Sea-surface temperature (SST) observations from two on-board sensors (the ARIES interferometer and a Heimann radiometer) and a satellite-based analysis (OSTIA) are used to determine an improved SST estimate. Most of the observations are from moderate to strong wind speed conditions, the latter being a regime short of validation data for the bulk flux algorithms that are necessary for numerical weather prediction and climate models. Observations from both statically stable and unstable atmospheric boundary-layer conditions are presented. There is a particular focus on several flights made as part of the DIAMET (Diabatic influence on mesoscale structures in extratropical storms) project. Observed neutral exchange coefficients are in the same range as previous studies, although higher for the momentum coefficient, and are broadly consistent with the COARE 3.0 bulk flux algorithm, as well as the surface exchange schemes used in the ECMWF and Met Office models. Examining the results as a function of aircraft heading shows higher fluxes and exchange coefficients in the across-wind direction, compared to along-wind (although this comparison is limited by the relatively small number of along-wind legs). A multi-resolution spectral decomposition technique demonstrates a lengthening of spatial scales in along-wind variances in along-wind legs, implying the boundary-layer eddies are elongated in the along-wind direction. The along-wind runs may not be able to adequately capture the full range of turbulent exchange that is occurring because elongation places the largest eddies outside of the run length.

Daytime CO2 urban surface fluxes from airborne measurements, eddy-covariance observations and emissions inventory in Greater London

Airborne measurements within the urban mixing layer (360 m) over Greater London are used to quantify CO2 emissions at the meso-scale. Daytime CO2 fluxes, calculated by the Integrative Mass Boundary Layer (IMBL) method, ranged from 46 to 104 μmol CO2 m−2 s−1 for four days in October 2011. The day-to-day variability of IMBL fluxes is at the same order of magnitude as for surface eddy-covariance fluxes observed in central London. Compared to fluxes derived from emissions inventory, the IMBL method gives both lower (by −37%) and higher (by 19%) estimates. The sources of uncertainty of applying the IMBL method in urban areas are discussed and guidance for future studies is given.

Simple M-factor algorithm for improved estimation of the basic maximum usable frequency of radio waves reflected from the ionospheric F-region

The equations of Milsom are evaluated, giving the ground range and group delay of radio waves propagated via the horizontally stratified model ionosphere proposed by Bradley and Dudeney. Expressions for the ground range which allow for the effects of the underlying E- and F1-regions are used to evaluate the basic maximum usable frequency or M-factors for single F-layer hops. An algorithm for the rapid calculation of the M-factor at a given range is developed, and shown to be accurate to within 5%. The results reveal that the M(3000)F2-factor scaled from vertical-incidence ionograms using the standard URSI procedure can be up to 7.5% in error. A simple addition to the algorithm effects a correction to ionogram values to make these accurate to 0.5%.

Comparing hybrid data assimilation methods on the Lorenz 1963 model with increasing nonlinearity

We systematically compare the performance of ETKF-4DVAR, 4DVAR-BEN and 4DENVAR with respect to two traditional methods (4DVAR and ETKF) and an ensemble transform Kalman smoother (ETKS) on the Lorenz 1963 model. We specifically investigated this performance with increasing nonlinearity and using a quasi-static variational assimilation algorithm as a comparison. Using the analysis root mean square error (RMSE) as a metric, these methods have been compared considering (1) assimilation window length and observation interval size and (2) ensemble size to investigate the influence of hybrid background error covariance matrices and nonlinearity on the performance of the methods. For short assimilation windows with close to linear dynamics, it has been shown that all hybrid methods show an improvement in RMSE compared to the traditional methods. For long assimilation window lengths in which nonlinear dynamics are substantial, the variational framework can have diffculties fnding the global minimum of the cost function, so we explore a quasi-static variational assimilation (QSVA) framework. Of the hybrid methods, it is seen that under certain parameters, hybrid methods which do not use a climatological background error covariance do not need QSVA to perform accurately. Generally, results show that the ETKS and hybrid methods that do not use a climatological background error covariance matrix with QSVA outperform all other methods due to the full flow dependency of the background error covariance matrix which also allows for the most nonlinearity.

Bit-index sort: a fast non-comparison integer sorting algorithm for permutations

This paper describes a fast integer sorting algorithm, herein referred as Bit-index sort, which is a non-comparison sorting algorithm for partial per-mutations, with linear complexity order in execution time. Bit-index sort uses a bit-array to classify input sequences of distinct integers, and exploits built-in bit functions in C compilers supported by machine hardware to retrieve the ordered output sequence. Results show that Bit-index sort outperforms in execution time to quicksort and counting sort algorithms. A parallel approach for Bit-index sort using two simultaneous threads is included, which obtains speedups up to 1.6.

Search engine liability for autocomplete suggestions: personality, privacy and the power of the algorithm

This article is concerned with the liability of search engines for algorithmically produced search suggestions, such as through Google’s ‘autocomplete’ function. Liability in this context may arise when automatically generated associations have an offensive or defamatory meaning, or may even induce infringement of intellectual property rights. The increasing number of cases that have been brought before courts all over the world puts forward questions on the conflict of fundamental freedoms of speech and access to information on the one hand, and personality rights of individuals— under a broader right of informational self-determination—on the other. In the light of the recent judgment of the Court of Justice of the European Union (EU) in Google Spain v AEPD, this article concludes that many requests for removal of suggestions including private individuals’ information will be successful on the basis of EU data protection law, even absent prejudice to the person concerned.

Differences between the CME fronts tracked by an expert, an automated algorithm, and the Solar Stormwatch project

Observations from the Heliospheric Imager (HI) instruments aboard the twin STEREO spacecraft have enabled the compilation of several catalogues of coronal mass ejections (CMEs), each characterizing the propagation of CMEs through the inner heliosphere. Three such catalogues are the Rutherford Appleton Laboratory (RAL)-HI event list, the Solar Stormwatch CME catalogue, and, presented here, the J-tracker catalogue. Each catalogue uses a different method to characterize the location of CME fronts in the HI images: manual identification by an expert, the statistical reduction of the manual identifications of many citizen scientists, and an automated algorithm. We provide a quantitative comparison of the differences between these catalogues and techniques, using 51 CMEs common to each catalogue. The time-elongation profiles of these CME fronts are compared, as are the estimates of the CME kinematics derived from application of three widely used single-spacecraft-fitting techniques. The J-tracker and RAL-HI profiles are most similar, while the Solar Stormwatch profiles display a small systematic offset. Evidence is presented that these differences arise because the RAL-HI and J-tracker profiles follow the sunward edge of CME density enhancements, while Solar Stormwatch profiles track closer to the antisunward (leading) edge. We demonstrate that the method used to produce the time-elongation profile typically introduces more variability into the kinematic estimates than differences between the various single-spacecraft-fitting techniques. This has implications for the repeatability and robustness of these types of analyses, arguably especially so in the context of space weather forecasting, where it could make the results strongly dependent on the methods used by the forecaster.

A robust sclera segmentation algorithm

Sclera segmentation is shown to be of significant importance for eye and iris biometrics. However, sclera segmentation has not been extensively researched as a separate topic, but mainly summarized as a component of a broader task. This paper proposes a novel sclera segmentation algorithm for colour images which operates at pixel-level. Exploring various colour spaces, the proposed approach is robust to image noise and different gaze directions. The algorithm’s robustness is enhanced by a two-stage classifier. At the first stage, a set of simple classifiers is employed, while at the second stage, a neural network classifier operates on the probabilities’ space generated by the classifiers at stage 1. The proposed method was ranked the 1st in Sclera Segmentation Benchmarking Competition 2015, part of BTAS 2015, with a precision of 95.05% corresponding to a recall of 94.56%.

A Genetic Algorithm for the selection of structural MRI features for classification of Mild Cognitive Impairment and Alzheimer's Disease

This work investigates the problem of feature selection in neuroimaging features from structural MRI brain images for the classification of subjects as healthy controls, suffering from Mild Cognitive Impairment or Alzheimer’s Disease. A Genetic Algorithm wrapper method for feature selection is adopted in conjunction with a Support Vector Machine classifier. In very large feature sets, feature selection is found to be redundant as the accuracy is often worsened when compared to an Support Vector Machine with no feature selection. However, when just the hippocampal subfields are used, feature selection shows a significant improvement of the classification accuracy. Three-class Support Vector Machines and two-class Support Vector Machines combined with weighted voting are also compared with the former and found more useful. The highest accuracy achieved at classifying the test data was 65.5% using a genetic algorithm for feature selection with a three-class Support Vector Machine classifier.