Analysing the Impact of Learning Inputs - Application to Terrain Traversability Estimation

Data-driven approaches such as Gaussian Process (GP) regression have been used extensively in recent robotics literature to achieve estimation by learning from experience. To ensure satisfactory performance, in most cases, multiple learning inputs are required. Intuitively, adding new inputs can often contribute to better estimation accuracy, however, it may come at the cost of a new sensor, larger training dataset and/or more complex learning, some- times for limited benefits. Therefore, it is crucial to have a systematic procedure to determine the actual impact each input has on the estimation performance. To address this issue, in this paper we propose to analyse the impact of each input on the estimate using a variance-based sensitivity analysis method. We propose an approach built on Analysis of Variance (ANOVA) decomposition, which can characterise how the prediction changes as one or more of the input changes, and also quantify the prediction uncertainty as attributed from each of the inputs in the framework of dependent inputs. We apply the proposed approach to a terrain-traversability estimation method we proposed in prior work, which is based on multi-task GP regression, and we validate this implementation experimentally using a rover on a Mars-analogue terrain.

Estimation of cosmological parameters using adaptive importance sampling

We present a Bayesian sampling algorithm called adaptive importance sampling or population Monte Carlo (PMC), whose computational workload is easily parallelizable and thus has the potential to considerably reduce the wall-clock time required for sampling, along with providing other benefits. To assess the performance of the approach for cosmological problems, we use simulated and actual data consisting of CMB anisotropies, supernovae of type Ia, and weak cosmological lensing, and provide a comparison of results to those obtained using state-of-the-art Markov chain Monte Carlo (MCMC). For both types of data sets, we find comparable parameter estimates for PMC and MCMC, with the advantage of a significantly lower wall-clock time for PMC. In the case of WMAP5 data, for example, the wall-clock time scale reduces from days for MCMC to hours using PMC on a cluster of processors. Other benefits of the PMC approach, along with potential difficulties in using the approach, are analyzed and discussed.

A new family of multivariate heavy-tailed distributions with variable marginal amounts of tailweight: Application to robust clustering

We propose a family of multivariate heavy-tailed distributions that allow variable marginal amounts of tailweight. The originality comes from introducing multidimensional instead of univariate scale variables for the mixture of scaled Gaussian family of distributions. In contrast to most existing approaches, the derived distributions can account for a variety of shapes and have a simple tractable form with a closed-form probability density function whatever the dimension. We examine a number of properties of these distributions and illustrate them in the particular case of Pearson type VII and t tails. For these latter cases, we provide maximum likelihood estimation of the parameters and illustrate their modelling flexibility on simulated and real data clustering examples.

Bayesian mixture model estimation of aerosol particle size distributions

In this paper, we examine approaches to estimate a Bayesian mixture model at both single and multiple time points for a sample of actual and simulated aerosol particle size distribution (PSD) data. For estimation of a mixture model at a single time point, we use Reversible Jump Markov Chain Monte Carlo (RJMCMC) to estimate mixture model parameters including the number of components which is assumed to be unknown. We compare the results of this approach to a commonly used estimation method in the aerosol physics literature. As PSD data is often measured over time, often at small time intervals, we also examine the use of an informative prior for estimation of the mixture parameters which takes into account the correlated nature of the parameters. The Bayesian mixture model offers a promising approach, providing advantages both in estimation and inference.

Study of linear and nonlinear optical properties of dendrimers using density matrix renormalization group method

We have used the density matrix renormalization group (DMRG) method to study the linear and nonlinear optical responses of first generation nitrogen based dendrimers with donor acceptor groups. We have employed Pariser–Parr–Pople Hamiltonian to model the interacting pi electrons in these systems. Within the DMRG method we have used an innovative scheme to target excited states with large transition dipole to the ground state. This method reproduces exact optical gaps and polarization in systems where exact diagonalization of the Hamiltonian is possible. We have used a correction vector method which tacitly takes into account the contribution of all excited states, to obtain the ground state polarizibility, first hyperpolarizibility, and two photon absorption cross sections. We find that the lowest optical excitations as well as the lowest excited triplet states are localized. It is interesting to note that the first hyperpolarizibility saturates more rapidly with system size compared to linear polarizibility unlike that of linear polyenes.

Ecological significance of rice (Oryza sativa) planting density and nitrogen rates in managing the growth and competitive ability of itchgrass (Rottboellia cochinchinensis) in direct-seeded rice systems

Current understanding is that high planting density has the potential to suppress weeds and crop-weed interactions can be exploited by adjusting fertilizer rates. We hypothesized that (a) high planting density can be used to suppress Rottboellia cochinchinensis growth and (b) rice competitiveness against this weed can be enhanced by increasing nitrogen (N) rates. We tested these hypotheses by growing R. cochinchinensis alone and in competition with four rice planting densities (0, 100, 200, and 400 plants m-2) at four N rates (0, 50, 100, and 150 kg ha-1). At 56 days after sowing (DAS), R. cochinchinensis plant height decreased by 27-50 %, tiller number by 55-76 %, leaf number by 68-84 %, leaf area by 70-83 %, leaf biomass by 26-90 %, and inflorescence biomass by 60-84 %, with rice densities ranging from 100 to 400 plants m-2. All these parameters increased with an increase in N rate. Without the addition of N, R. cochinchinensis plants were 174 % taller than rice; whereas, with added N, they were 233 % taller. Added N favored more weed biomass production relative to rice. R. cochinchinensis grew taller than rice (at all N rates) to avoid shade, which suggests that it is a "shade-avoiding" plant. R. cochinchinensis showed this ability to reduce the effect of rice interference through increased leaf weight ratio, specific stem length, and decreased root-shoot weight ratio. This weed is more responsive to N fertilizer than rice. Therefore, farmers should give special consideration to the application timing of N fertilizer when more N-responsive weeds are present in their field. Results suggest that the growth and seed production of R. cochinchinensis can be decreased considerably by increasing rice density to 400 plants m-2. There is a need to integrate different weed control measures to achieve complete control of this noxious weed.

Variation in water extraction with maize plant density and its impact on model application

The use of maize simulation models to determine the optimum plant population for rainfed environments allows the evaluation of plant populations over multiple years and locations at a lower cost than traditional field experimentation. However the APSIM maize model that has been used to conduct some of these 'virtual' experiments assumes that the maximum rate of soil water extraction by the crop root system is constant across plant populations. This untested assumption may cause grain yield to be overestimated in lower plant populations. A field experiment was conducted to determine whether maximum rates of water extraction vary with plant population, and the maximum rate of soil water extraction was estimated for three plant populations (2.4, 3.5 and 5.5 plants m(-2)) under water limited conditions. Maximum soil water extraction rates in the field experiment decreased linearly with plant population, and no difference was detected between plant populations for the crop lower limit of soil water extraction. Re-analysis of previous maize simulation experiments demonstrated that the use of inappropriately high extraction-rate parameters at low plant populations inflated predictions of grain yield, and could cause erroneous recommendations to be made for plant population. The results demonstrate the importance of validating crop simulation models across the range of intended treatments. (C) 2013 Elsevier E.V. All rights reserved.

Stock assessment of the Queensland east coast common coral trout (Plectropomus leopardus) fishery

Common coral trout Plectropomus leopardus is an iconic fish of the Great Barrier Reef (GBR) and is the most important fish for the commercial fishery there. Most of the catch is exported live to Asia. This stock assessment was undertaken in response to falls in catch sizes and catch rates in recent years, in order to gauge the status of the stock. It is the first stock assessment ever conducted of coral trout on the GBR, and brings together a multitude of different data sources for the first time. The GBR is very large and was divided into a regional structure based on the Bioregions defined by expert committees appointed by the Great Barrier Reef Marine Park Authority (GBRMPA) as part of the 2004 rezoning of the GBR. The regional structure consists of six Regions, from the Far Northern Region in the north to the Swains and Capricorn–Bunker Regions in the south. Regions also closely follow the boundaries between Bioregions. Two of the northern Regions are split into Subregions on the basis of potential changes in fishing intensity between the Subregions; there are nine Subregions altogether, which include four Regions that are not split. Bioregions are split into Subbioregions along the Subregion boundaries. Finally, each Subbioregion is split into a “blue” population which is open to fishing and a “green” population which is closed to fishing. The fishery is unusual in that catch rates as an indicator of abundance of coral trout are heavily influenced by tropical cyclones. After a major cyclone, catch rates fall for two to three years, and rebound after that. This effect is well correlated with the times of occurrence of cyclones, and usually occurs in the same month that the cyclone strikes. However, statistical analyses correlating catch rates with cyclone wind energy did not provide significantly different catch rate trends. Alternative indicators of cyclone strength may explain more of the catch rate decline, and future work should investigate this. Another feature of catch rates is the phenomenon of social learning in coral trout populations, whereby when a population of coral trout is fished, individuals quickly learn not to take bait. Then the catch rate falls sharply even when the population size is still high. The social learning may take place by fish directly observing their fellows being hooked, or perhaps heeding a chemo-sensory cue emitted by fish that are hooked. As part of the assessment, analysis of data from replenishment closures of Boult Reef in the Capricorn–Bunker Region (closed 1983–86) and Bramble Reef in the Townsville Subregion (closed 1992–95) estimated a strong social learning effect. A major data source for the stock assessment was the large collection of underwater visual survey (UVS) data collected by divers who counted the coral trout that they sighted. This allowed estimation of the density of coral trout in the different Bioregions (expressed as a number of fish per hectare). Combined with mapping data of all the 3000 or so reefs making up the GBR, the UVS results provided direct estimates of the population size in each Subbioregion. A regional population dynamic model was developed to account for the intricacies of coral trout population dynamics and catch rates. Because the statistical analysis of catch rates did not attribute much of the decline to tropical cyclones, (and thereby implied “real” declines in biomass), and because in contrast the UVS data indicate relatively stable population sizes, model outputs were unduly influenced by the unlikely hypothesis that falling catch rates are real. The alternative hypothesis that UVS data are closer to the mark and declining catch rates are an artefact of spurious (e.g., cyclone impact) effects is much more probable. Judging by the population size estimates provided by the UVS data, there is no biological problem with the status of coral trout stocks. The estimate of the total number of Plectropomus leopardus on blue zones on the GBR in the mid-1980s (the time of the major UVS series) was 5.34 million legal-sized fish, or about 8400 t exploitable biomass, with an 2 additional 3350 t in green zones (using the current zoning which was introduced on 1 July 2004). For the offshore regions favoured by commercial fishers, the figure was about 4.90 million legal-sized fish in blue zones, or about 7700 t exploitable biomass. There is, however, an economic problem, as indicated by relatively low catch rates and anecdotal information provided by commercial fishers. The costs of fishing the GBR by hook and line (the only method compatible with the GBR’s high conservation status) are high, and commercial fishers are unable to operate profitably when catch rates are depressed (e.g., from a tropical cyclone). The economic problem is compounded by the effect of social learning in coral trout, whereby catch rates fall rapidly if fishers keep returning to the same fishing locations. In response, commercial fishers tend to spread out over the GBR, including the Far Northern and Swains Regions which are far from port and incur higher travel costs. The economic problem provides some logic to a reduction in the TACC. Such a reduction during good times, such as when the fishery is rebounding after a major tropical cyclone, could provide a net benefit to the fishery, as it would provide a margin of stock safety and make the fishery more economically robust by providing higher catch rates during subsequent periods of depressed catches. During hard times when catch rates are low (e.g., shortly after a major tropical cyclone), a change to the TACC would have little effect as even a reduced TACC would not come close to being filled. Quota adjustments based on catch rates should take account of long-term trends in order to mitigate variability and cyclone effects in data.

Optimizing Hill Seeding Density for High-Yielding Hybrid Rice in a Single Rice Cropping System in South China

Mechanical hill direct seeding of hybrid rice could be the way to solve the problems of high seeding rates and uneven plant establishment now faced in direct seeded rice; however, it is not clear what the optimum hill seeding density should be for high-yielding hybrid rice in the single-season rice production system. Experiments were conducted in 2010 and 2011 to determine the effects of hill seeding density (25 cm 615 cm, 25 cm 617 cm, 25 cm 619 cm, 25 cm 621 cm, and 25 cm 623 cm; three to five seeds per hill) on plant growth and grain yield of a hybrid variety, Nei2you6, in two fields with different fertility (soil fertility 1 and 2). In addition, in 2012 and 2013, comparisons among mechanical hill seeding, broadcasting, and transplanting were conducted with three hybrid varieties to evaluate the optimum seeding density. With increases in seeding spacing from 25 cm615 cm to 25 cm623 cm, productive tillers per hill increased by 34.2% and 50.0% in soil fertility 1 and 2. Panicles per m2 declined with increases in seeding spacing in soil fertility 1. In soil fertility 2, no difference in panicles per m2 was found at spacing ranging from 25 cm617 cm to 25 cm623 cm, while decreases in the area of the top three leaves and aboveground dry weight per shoot at flowering were observed. Grain yield was the maximum at 25 cm 617 cm spacing in both soil fertility fields. Our results suggest that a seeding density of 25 cm617 cm was suitable for high-yielding hybrid rice. These results were verified through on-farm demonstration experiments, in which mechanical hill-seeded rice at this density had equal or higher grain yield than transplanted rice

Effects of light conditions and plant density on growth and reproductive biology of Cascabela thevetia (L.) Lippold

Cascabela thevetia (L.) Lippold (Apocynaceae) is an invasive woody weed that has formed large infestations at several locations in northern Australia. Understanding the reproductive biology of C. thevetia is vital to its management. This paper reports results of a shade house experiment that determined the effects of light conditions (100% or 30% of natural light) and plant densities (one, two, four or eight plants per plot) on the growth, time to flowering and seed formation, and monthly pod production of two C. thevetia biotypes (peach and yellow). Shaded plants were significantly larger when they reached reproductive maturity than plants grown under natural light. However, plants grown under natural light flowered earlier (268 days compared with 369 days) and produced 488 more pods per pot (a 5-fold increase) over 3 years. The yellow biotype was slightly taller at reproductive maturity but significantly taller and with significantly greater aboveground biomass at the end of the study. Both biotypes flowered at a similar time under natural light and low plant densities but the yellow biotype was quicker to seed (478 versus 498 days), produced significantly more pods (364 versus 203 pods) and more shoot growth (577 g versus 550 g) than the peach biotype over 3 years. Higher densities of C. thevetia tended to significantly reduce the shoot and root growth by 981 g and 714 g per plant across all light conditions and biotypes over 3 years and increase the time taken to flower by 140 days and produce seeds by 184 days. For land managers trying to prevent establishment of C. thevetia or to control seedling regrowth once initial infestations have been treated, this study indicates that young plants have the potential to flower and produce seeds within 268 and 353 days, respectively. However, with plant growth and reproduction most likely to be slower under field conditions, annual surveillance and control activities should be sufficient to find and treat plants before they produce seeds and replenish soil seed banks. The most at-risk part of the landscape may be open areas that receive maximum sunlight, particularly within riparian habitats where plants would consistently have more favourable soil moisture conditions.

A novel method for the age estimation of Saddletail snapper (Lutjanus malabaricus) using Fourier Transform-near infrared (FT-NIR) spectroscopy

Near infrared (NIR) spectroscopy was investigated as a potential rapid method of estimating fish age from whole otoliths of Saddletail snapper (Lutjanus malabaricus). Whole otoliths from 209 Saddletail snapper were extracted and the NIR spectral characteristics were acquired over a spectral range of 800–2780 nm. Partial least-squares models (PLS) were developed from the diffuse reflectance spectra and reference-validated age estimates (based on traditional sectioned otolith increments) to predict age for independent otolith samples. Predictive models developed for a specific season and geographical location performed poorly against a different season and geographical location. However, overall PLS regression statistics for predicting a combined population incorporating both geographic location and season variables were: coefficient of determination (R2) = 0.94, root mean square error of prediction (RMSEP) = 1.54 for age estimation, indicating that Saddletail age could be predicted within 1.5 increment counts. This level of accuracy suggests the method warrants further development for Saddletail snapper and may have potential for other fish species. A rapid method of fish age estimation could have the potential to reduce greatly both costs of time and materials in the assessment and management of commercial fisheries.

The utility of near infrared spectroscopy for age estimation of deepwater sharks

Reliable age information is vital for effective fisheries management, yet age determinations are absent for many deepwater sharks as they cannot be aged using traditional methods of growth bands counts. An alternative approach to ageing using near infrared spectroscopy (NIRS) was investigated using dorsal fin spines, vertebrae and fin clips of three species of deepwater sharks. Ages were successfully estimated for the two dogfish, Squalus megalops and Squalus montalbani, and NIRS spectra were correlated with body size in the catshark, Asymbolus pallidus. Correlations between estimated-ages of the dogfish dorsal fin spines and their NIRS spectra were good, with S. megalops R2=0.82 and S. montalbani R2=0.73. NIRS spectra from S. megalops vertebrae and fin clips that have no visible growth bands were correlated with estimated-ages, with R2=0.89 and 0.76, respectively. NIRS has the capacity to non-lethally estimate ages from fin spines and fin clips, and thus could significantly reduce the numbers of sharks that need to be lethally sampled for ageing studies. The detection of ageing materials by NIRS in poorly calcified deepwater shark vertebrae could potentially enable ageing of this group of sharks that are vulnerable to exploitation.

Population density of the European provinces of Russia

Digital image

Remote sensing of boreal land cover: estimation of forest attributes and extent

Remote sensing provides methods to infer land cover information over large geographical areas at a variety of spatial and temporal resolutions. Land cover is input data for a range of environmental models and information on land cover dynamics is required for monitoring the implications of global change. Such data are also essential in support of environmental management and policymaking. Boreal forests are a key component of the global climate and a major sink of carbon. The northern latitudes are expected to experience a disproportionate and rapid warming, which can have a major impact on vegetation at forest limits. This thesis examines the use of optical remote sensing for estimating aboveground biomass, leaf area index (LAI), tree cover and tree height in the boreal forests and tundra taiga transition zone in Finland. The continuous fields of forest attributes are required, for example, to improve the mapping of forest extent. The thesis focus on studying the feasibility of satellite data at multiple spatial resolutions, assessing the potential of multispectral, -angular and -temporal information, and provides regional evaluation for global land cover data. Preprocessed ASTER, MISR and MODIS products are the principal satellite data. The reference data consist of field measurements, forest inventory data and fine resolution land cover maps. Fine resolution studies demonstrate how statistical relationships between biomass and satellite data are relatively strong in single species and low biomass mountain birch forests in comparison to higher biomass coniferous stands. The combination of forest stand data and fine resolution ASTER images provides a method for biomass estimation using medium resolution MODIS data. The multiangular data improve the accuracy of land cover mapping in the sparsely forested tundra taiga transition zone, particularly in mires. Similarly, multitemporal data improve the accuracy of coarse resolution tree cover estimates in comparison to single date data. Furthermore, the peak of the growing season is not necessarily the optimal time for land cover mapping in the northern boreal regions. The evaluated coarse resolution land cover data sets have considerable shortcomings in northernmost Finland and should be used with caution in similar regions. The quantitative reference data and upscaling methods for integrating multiresolution data are required for calibration of statistical models and evaluation of land cover data sets. The preprocessed image products have potential for wider use as they can considerably reduce the time and effort used for data processing.

Increasing the productivity of avocado orchards using high-density plantings: A review

There is strong interest in the use of high-density plantings to increase the productivity of avocado (Persea americana) orchards. Close plantings have the potential for higher yields and returns than standard or traditional plantings, especially in the early years of production. The success of this technology is dependent on the use of methods to control shoot growth and maximise light interception as the trees begin to bear fruit. We reviewed the performance of high-density orchards in different environments, and the success of efforts to control the growth of the trees through the use of dwarfing material, canopy management and growth regulators. Close plantings generally produce higher yields in the first few years of bearing compared with the yields of standard plantings. However, in most growing areas, the trees in the close plantings soon begin to crowd each other and yields decline. This usually occurs despite efforts to control shoot growth by pruning the trees or by applying growth regulators. Efforts to breed dwarfing rootstocks that can control the growth of mature trees have been largely unsuccessful. In the absence of dwarfing material, effective canopy management appears to be the largest barrier to success of high-density orchards. Further research on the use of different pruning strategies and growth regulators to control the growth of the trees and maximise light interception is required. There are potential problems with some of the growth regulators persisting in the harvested fruit and soil under certain circumstances.