Learning-aided predictor integration for system performance prediction

The integration of multiple predictors promises higher prediction accuracy than the accuracy that can be obtained with a single predictor. The challenge is how to select the best predictor at any given moment. Traditionally, multiple predictors are run in parallel and the one that generates the best result is selected for prediction. In this paper, we propose a novel approach for predictor integration based on the learning of historical predictions. Compared with the traditional approach, it does not require running all the predictors simultaneously. Instead, it uses classification algorithms such as k-Nearest Neighbor (k-NN) and Bayesian classification and dimension reduction technique such as Principal Component Analysis (PCA) to forecast the best predictor for the workload under study based on the learning of historical predictions. Then only the forecasted best predictor is run for prediction. Our experimental results show that it achieved 20.18% higher best predictor forecasting accuracy than the cumulative MSE based predictor selection approach used in the popular Network Weather Service system. In addition, it outperformed the observed most accurate single predictor in the pool for 44.23% of the performance traces.

A long-term improvement in Danish stream fauna: Analyses of temporal dynamics and community alignment of a biotic index

Danish streams have for some decades shown a significant improvement in ecological quality. This is based on a time-series of a subset of 247 sites from the nationwide monitoring program with reoccurring annual sampling. The ecological quality is determined using the benthic macroinvertebrate fauna as bio-indicators via the Danish Stream Fauna Index (DSFI), and expressed in categorically ranked scores, with 1 as the lowest, and 7 as the highest quality. We analysed a data set on DSFI scores and the associated taxon lists from 2004 to 2013, totaling to 2411 individual samples or communities. Our main objectives were to 1) examine the temporal multidirectional dynamics underlying the overall net improving trend in the DSFI scores, and 2) to elucidate how the DSFI responds to differences in faunal community composition. Our analysis showed that most sites exhibited unchanged DSFI over the 10 year period (53%), although inter-annual shifts were still observable within these sites. The DSFI interval 1–2 showed the highest proportion of positive shifts, whereas interval 3–4 the highest proportion of negative shifts. Improvement measures should therefore be directed more specifically towards intermediate quality streams. Sites with moderate to high scores (DSFI 4–7) were most stable. Turnover (i.e., replacement of taxa) was the dominating component of overall beta diversity, while the richness change component (i.e., loss/gain of taxa) was negligible. The specific DSFI scores encompassed wide ranges of community composition, and showed a weak, but significant correlation between differences in scores and the community composition. This is to a certain degree advantageous since it gives the categorical scores robustness and plasticity, and thus makes the DSFI capable of handling natural variation in communities. The ideal biotic index should allow for natural variation, but do so while maintaining the capability of separating ecologically different communities. Our findings consequently give rise to concern to whether the DSFI is too sensitive to stochastic variation in samples with insufficient precision to assign sites to correct ecological quality classes.     

Wolverines and declining snowpack: response to comments

The critiques by DeVink et al. (2010) and McKelvey et al. (2010) are flawed for several reasons. We show here that, contrary to what DeVink et al. claim, the influence of annual pelt price on wolverine harvest returns is essentially negligible. DeVink et al. also suggest that our results show the influence of snowpack on trapper success, rather than on actual wolverine population dynamics. This is unlikely, since most of the snowpack terms in our models are at 1- or 2-year time lags, whereas the impact of snow conditions on trapper success can only manifest in the current year. Both DeVink et al. and McKelvey et al. claim that wolverine populations across Canada are actually increasing, but provide no quantitative data to support this claim. Both sets of authors present alternative explanations for the declines in harvest returns, but none of those explanations are mutually exclusive with our own, and none can explain the significance of time-lagged snowpack on annual harvest returns. McKelvey et al.’s claim that our results represent a spurious correlation, as well as other points that they raise, suggests either a superficial understanding or deliberate misrepresentation of our methods and can simply reflect their underlying philosophical biases.     

Modelling rainfall and canopy controls on net-precipitation beneath selectively-logged tropical forest

Understanding spatio-temporal patterns in rainfall received beneath tropical forest is required for eco- hydrological modelling of soil-water status, river behaviour, soil erosion, nutrient loss and wet-canopy evaporation. As selective-logging of tropical forest leaves a very complex mosaic of canopy types, it is likely to add to the spatio-temporal complexity of this sub-canopy or net precipitation. As a precursor to addressing this problem, the analysis presented here will examine the two dominant biophysical controls on sub-canopy precipitation. These controls are: (a) the spatial and temporal patterns in above-canopy or gross rainfall, and (b) the rate of wet-canopy evaporation associated with each type of canopy structure created by selective-forestry. For this study, over 400 raingauges were installed within a 10 km² area of lowland dipterocarp forest affected by selective-forestry some 9-years prior to this work. Gauges were located beneath various canopy types and within large openings. The spatial distribution of gross rainfall (monitored within the openings) was modelled using variography, while the effects of different canopy types on sub-canopy preciptation was analysed by comparing 6-month totals. The temporal distribution of gross rainfall over an 11-year record collected at the same site (Danum Valley Field Centre) was modelled with Data-Based-Mechanistic (DBM) approaches. These DBM approaches were also applied to the rainfall time-series of the two adjacent meteorological stations; all three gauges being contained within a 5000 km² region of Eastern Sabah in Malaysian Borneo.Strong diurnal modulation was apparent within gross rainfall for the inland rainforest site, with a distribution consistent with a dominance of local convective rain cells. A similarly strong cycle coincident with the periodicity of the El Niño-Southern Oscillation (ENSO) was present within all of the region's rainfall records, though marked differences in annual and intra-annual seasonality were apparent. The preliminary variogram modelling indicated that a deterministic drift was present within the local-scale gross rainfall data, probably related to local topographic effects. Notwithstanding the need to remove this drift, the work indicated that spatial models of gross rainfall could be identified and used to interpret similar models of net-precipitation. During the ENSO drought-period monitored, the lowland dipterocarp forest allowed 91% of the gross rainfall to reach the ground as throughfall. These rates were, however, reduced to between 80%–86% beneath representative plots of moderately impacted to creeper-covered, highly damaged patches of forest.     

Temporal patterns of benthic community development in an Arctic fjord (Kongsfjorden, Svalbard): results of a 24-year manipulation study

We investigated temporal patterns of recolonisation and disturbance in a benthic hard bottom community in high-arctic Kongsfjorden from 1980 to 2003 through annual photographic surveys. A manipulative sampling design was applied, where half of the study area (treatment areas) was cleared at the beginning of the study. Twenty-three different taxa and groups of benthic epifauna were found in the photographs. The benthic community structures of treatments and controls converged within the first decade, but significant differences prevailed until ≤13 years after the start of the study. We could distinguish between three different time intervals with increased inter-annual changes. While the observed differences during the first two intervals could be attributed to recolonisation and succession, the changes in interval 3 were mostly due to increased external forcing and characterised by low inter-group and high inter-annual differences. During this interval, brown algae (mainly Desmarestia) and the sea urchin Strongylocentrotus droebachiensis emerged in high densities, while sea anemone populations declined. Different recolonisation patterns for individual species were related to life span, rate of maturity, predators and larval settlement. We could not find a climax stage in the succession for the benthic community at Kvadehuken, presumably due to the constant level of disturbances at the site.     

山楂叶螨种群空间格局及其应用的研究Ⅰ.成螨在苹果树上的三维空间格局

考察了山楂叶螨成螨在树内的三维空间格局及其聚集与扩散的趋势,并用聚块性指标研究了其聚集强度的时序变化。结果表明,在垂直和水平方向,山楂叶螨成螨在树内的空间格局均为聚集型,呈单个个体或小的个体群分布,以下层树冠的聚集强度为最高。根据种群的空间格局,给出了上、中、下3层树冠的数据代换公式。     

Measuring phenological variability from satellite imagery

Abstract. Vegetation phenological phenomena are closely related to seasonal dynamics of the lower atmosphere and are therefore important elements in global models and vegetation monitoring. Normalized difference vegetation index (NDVI) data derived from the National Oceanic and Atmospheric Administration's Advanced Very High Resolution Radiometer (AVHRR) satellite sensor offer a means of efficiently and objectively evaluating phenological characteristics over large areas. Twelve metrics linked to key phenological events were computed based on time-series NDVI data collected from 1989 to 1992 over the conterminous United States. These measures include the onset of greenness, time of peak NDVI, maximum NDVI, rate of greenup, rate of senescence, and integrated NDVI. Measures of central tendency and variability of the measures were computed and analyzed for various land cover types. Results from the analysis showed strong coincidence between the satellite-derived metrics and predicted phenological characteristics. In particular, the metrics identified interannual variability of spring wheat in North Dakota, characterized the phenology of four types of grasslands, and established the phenological consistency of deciduous and coniferous forests. These results have implications for large-area land cover mapping and monitoring. The utility of remotely sensed data as input to vegetation mapping is demonstrated by showing the distinct phenology of several land cover types. More stable information contained in ancillary data should be incorporated into the mapping process, particularly in areas with high phenological variability. In a regional or global monitoring system, an increase in variability in a region may serve as a signal to perform more detailed land cover analysis with higher resolution imagery.     

生物信息学基因表达差异分析

基因的差异化表达由多种因素共同导致,并且与许多疾病的发生和发展有密切联系,对差异化表达的基因进行生物信息学以及生物统计学的分析对于研究细胞调节机制和疾病机理有着重要意义。目前,对差异化表达的基因有以下几种主流的研究方法：DNA微阵列（DNA microarray）,抑制性消减杂交（SSH）,基因表达连续性分析（SAGE）,代表性差异分析（RDA）,以及mRNA差异显示PCR（mRNA DDRT-PCR）。目前许多基因差异化表达数据是建立在时段（time series）基础上,因此对基于时间变化的基因差异化表达分析变得尤为重要。本文将对差异化表达基因的几种主流方法进行详细阐述,并介绍一种基于傅里叶函数的时段基因差异化表达分析。     

Assessing temporal scales and patterns in time series: Comparing methods based on redundancy analysis

Time-series modelling techniques are powerful tools for studying temporal scaling structures and dynamics present in ecological and other complex systems and are gaining popularity for assessing resilience quantitatively. Among other methods, canonical ordinations based on redundancy analysis are increasingly used for determining temporal scaling patterns that are inherent in ecological data. However, modelling outcomes and thus inference about ecological dynamics and resilience may vary depending on the approaches used. In this study, we compare the statistical performance, logical consistency and information content of two approaches: (i) asymmetric eigenvector maps (AEM) that account for linear trends and (ii) symmetric distance-based Moran's eigenvector maps (MEM), which requires detrending of raw data to remove linear trends prior to analysis. Our comparison is done using long-term water quality data (25 years) from three Swedish lakes. This data set therefore provides the opportunity for assessing how the modelling approach used affects performance and inference in time series modelling. We found that AEM models had consistently more explanatory power than MEM, and in two out of three lakes AEM extracted one more temporal scale than MEM. The scale-specific patterns detected by AEM and MEM were uncorrelated. Also individual water quality variables explaining these patterns differed between methods, suggesting that inferences about systems dynamics are dependent on modelling approach. These findings suggest that AEM might be more suitable for assessing dynamics in time series analysis compared to MEM when temporal trends are relevant. The AEM approach is logically consistent with temporal autocorrelation where earlier conditions can influence later conditions but not vice versa. The symmetric MEM approach, which ignores the asymmetric nature of time, might be suitable for addressing specific questions about the importance of correlations in fluctuation patterns where there are no confounding elements of linear trends or a need to assess causality.     

Statistical modeling of valley fever data in Kern County,California

Coccidioidomycosis (valley fever) is a fungal infection found in the southwestern US, northern Mexico, and some places in Central and South America. The fungus that causes it (Coccidioides immitis) is normally soil-dwelling but, if disturbed, becomes air-borne and infects the host when its spores are inhaled. It is thus natural to surmise that weather conditions that foster the growth and dispersal of the fungus must have an effect on the number of cases in the endemic areas. We present here an attempt at the modeling of valley fever incidence in Kern County, California, by the implementation of a generalized auto regressive moving average (GARMA) model. We show that the number of valley fever cases can be predicted mainly by considering only the previous history of incidence rates in the county. The inclusion of weather-related time sequences improves the model only to a relatively minor extent. This suggests that fluctuations of incidence rates (about a seasonally varying background value) are related to biological and/or anthropogenic reasons, and not so much to weather anomalies.