Janette Walde

Classifiers vs. input variables—The drivers in image classification for land cover mapping

The study investigates the performance of image classifiers for landscape-scale land cover mapping and the relevance of ancillary data for the classification success in order to assess and to quantify the importance of these components in image classification. Specifically tested are the performance of maximum likelihood classification (MLC), artificial neural networks (ANN) and discriminant analysis (DA) based on Landsat7 ETM+ spectral data in combination with topographic measures and NDVI. ANN produced high accuracies of more than 75% also with limited input information, while MLC and DA produced comparable results only by incorporating ancillary data into the classification process. The superiority of ANN classification was less pronounced on the level of the single land cover classes. The use of ancillary data generally increased classification accuracy and showed a similar potential for increasing classification accuracy than the selection of the classifier. Therefore, a stronger focus on the development of appropriate and optimised sets of input variables is suggested. Also the definition and selection of land cover classes has shown to be crucial and not to be simply adaptable from existing land cover class schemes. A stronger research focus towards discriminating land cover classes by their typical spectral, topographic or seasonal properties is therefore suggested to advance image classification.

Lactic acid and methane: Improved exploitation of biowaste potential

This feasibility study investigated a two-step biorefining approach to increase the value gained by recycling of organic municipal solid waste. Firstly, lactic acid was produced via batch fermentation at 37°C using the indigenous microbiome. Experiments revealed an optimal fermentation period of 24h resulting in high yields of lactic acid (up to 37gkg(-1)). The lactic acid proportion of total volatile fatty acid content reached up to 83%. Lactobacilli were selectively enriched to up to 75% of the bacterial community. Additionally conversion of organic matter to lactic acid was increased from 22% to 30% through counteracting end product inhibition by continuous lactic acid extraction. Secondly, fermentation residues were used as co-substrate in biomethane production yielding up to 618±41Nmlbiomethaneg(-1) volatile solids. Digestate, the only end product of this process can be used as organic fertilizer.

Stability and predictive value of anti-JCV antibody index in multiple sclerosis: A 6-year longitudinal study.

Background Risk of natalizumab-related progressive multifocal leukoencephalopathy is associated with the presence of anti-JC-virus (JCV) antibodies. Objective To investigate the longitudinal evolution of anti-JCV antibody index and to determine the predictive value of baseline anti-JCV antibody index for long-term stability of anti-JCV antibody status. Methods MS patients from the MS centre of Medical University of Innsbruck, who had serum sampling for a time period of 4–6 years at intervals of 6±3 months, were included in this retrospective, longitudinal study. Anti-JCV antibody serological status and index were determined by 2-step second-generation anti-JCV antibody assay. Results 154 patients were included in this study. Median follow-up time was 63.7 months, with median 11 samples available per patient. At baseline, 111 (72.1%) patients were anti-JCV antibody positive. Baseline anti-JCV antibody index significantly correlated with age (R = 0.22, p = 0.005); there was no difference with respect to sex, disease duration or previously used disease-modifying treatment. During follow-up anti-JCV antibody status changed from negative to positive or vice versa in 17% of patients. In seronegative patients at baseline, baseline anti-JCV antibody index was significantly lower in those remaining seronegative at follow-up compared to those converting to seropositivity (median 0.16 vs. 0.24, p = 0.002). In seropositive patients at baseline, index was higher in those remaining seropositive compared to those reverting to seronegativity (2.6 vs. 0.45, p<10−7). Baseline anti-JCV antibody index >0.90 predicted stable positive serostatus (sensitivity 88.7%, specificity 96.5%) and <0.20 stable negative serostatus (sensitivity 61.3%, specificity 97.6%). Conclusions Anti-JCV antibody index remained relatively stable over 6-year follow-up with annual serostatus change of ~3%. Baseline anti-JCV antibody index predicted stable negative and stable positive JCV serostatus.

Exploring socio-cultural values of ecosystem service categories in the Central Alps: the influence of socio-demographic factors and landscape type

Abstract The socio-cultural assessment of ecosystem services has been proposed as a promising tool for eliciting people’s preferences towards ecosystem services. Despite an increasing integration of the socio-cultural perspective in ecosystem service research, little knowledge exists about linkages between landscape and the socio-cultural values people assign to ecosystem services. This paper combines a socio-cultural valuation approach with the use of landscape pictures to analyse and compare people’s perceived importance of the provisioning, regulating, and cultural ecosystem service categories across three landscape types (i.e. larch meadows, spruce forests, and hay meadow). A survey with 470 tourists visiting the region of South Tyrol (Italy) was conducted to link people’s perceived importance to their socio-demographic background and to the landscape types explored. The results show that regulating ecosystem services are preferred over provisioning and cultural services, whereby environmental awareness is found to be more influential than formal education levels regarding the perceived importance of regulating services. The results further demonstrate that cultural background is an important driver in determining people’s perceived importance of cultural services. The underlying landscape types, however, exert an even stronger influence on people’s socio-cultural valuation of ecosystem service categories. This finding suggests that the focus of most ecosystem services assessments on the study area as a whole risks mistakenly attributing differences in people’s socio-cultural values to socio-demographic characteristics only. A better knowledge of the spatial integration of socio-cultural values, however, could help with anticipating the consequences of changes in the landscape and provide better guidance for future landscape planning.

Performance contest between MLE and GMM for huge spatial autoregressive models

Abstract When using maximum likelihood estimation for spatial models, a well known problem is the computation of the logarithm of the determinant of the Jacobian, especially for problems with a huge number of observation units. In the recent literature there are various promising approaches to account for these numerical difficulties, relying on alternative decompositions or approximations. Recently, a general method of moments approach for estimating these models was developed. We compare all these different approaches with respect to their root mean-squared errors of the estimates and investigate the size and power of hypotheses tests with respect to the spatial correlation and the regression parameters.

Lag or Error? — Detecting the Nature of Spatial Correlation

Theory often suggests spatial correlations without being explicit about the exact form. Hence, econometric tests are used for model choice. So far, mainly Lagrange Multiplier tests based on ordinary least squares residuals are employed to decide whether and in which form spatial correlation is present in Cliff-Ord type spatial models. In this paper, the model selection is based both on likelihood ratio and Wald tests using estimates for the general model and information criteria. The results of the conducted large Monte Carlo study suggest that Wald tests on the spatial parameters after estimation of the general model are the most reliable approach to reveal the nature of spatial correlation.

Social Capital Formation and Intra Familial Correlation: A Social Panel Perspective

Social capital is widely regarded to constitute an important indicator for the economic performance of a society. This paper analyzes the impact of various socio-demographic characteristics on social capital. Proxy variables for social capital are obtained from a comprehensive principal components analysis exercise using survey data from the British Household Panel Survey (BHPS). The BHPS provides information on social and economic change at the individual and household levels in Britain and the UK with an annual survey of ca. 10,000 individuals from ca. 5,000 households. Based on the 13th wave of this database, we investigate the impact of exogenous qualities, individually acquired characteristics, and of the social environment using a spatial auto-regression framework. The results show that the formation of social capital can be modeled to a very high degree of statistical accuracy. The structural effect from the households contributes substantially to the social capital level of each household member. Thus, the social capital formation can be based equally on individual measures (such as education) and social contagion processes.

Statistical aspects of multilayer perceptrons under data limitations

Based on three case studies, the impact of sample size and sample randomness on the predictive accuracy of multilayer perceptrons (MLP) is investigated. The MLP prove to be useful for classification problems. Although they are dependent on the sample size and the non-linearity of the underlying problem, they achieve predictions superior to the classical methods. A so-called saturation curve describes the dependency of the network performance on the sample size. This function enables the user to evaluate the achieved network performance and the usefulness of additional data. For reliable and generalizable results, the calculation of prediction intervals for the network is essential. It is demonstrated that the network leads to narrower confidence intervals of the performance measures in comparison to classical methods even for small sample sizes. The experiments show the validity of the law, for even relatively small sample sizes, that the standard error of the hit ratio decreases by one over the square root of the sample size. Therefore, the suggestion is to estimate the standard error for a given sample size by randomly drawing smaller sample sizes, and then rescaling the standard error accordingly.

Simplified and still meaningful: assessing butterfly habitat quality in grasslands with data collected by pupils

Public participation in scientific research, now commonly referred to as citizen science, is increasingly promoted as a possibility to overcome the large-scale data limitations related to biodiversity and conservation research. Furthermore, public data-collection projects can stimulate public engagement and provide transformative learning situations. However, biodiversity monitoring depends on sound data collection and warranted data quality. Therefore, we investigated if and how trained and supervised pupils are able to systematically collect data about the occurrence of diurnal butterflies, and how this data could contribute to a permanent butterfly monitoring system. We developed a specific assessment scheme suitable for laypeople and applied it at 35 sampling sites in Tyrol, Austria. Data quality and its explanatory power to predict butterfly habitat quality was investigated comparing data collected by pupils with independent assessments of professional butterfly experts. Despite substantial identification uncertainties for some species or species groups, the data collected by pupils was successfully used to predict the general habitat quality for butterflies using a linear regression model (r² = 0.73, p <0.001). Applying the proposed method in a citizen science context with laypeople could support both the long term monitoring of butterfly habitat quality, as well as the efficient selection of sites for professional in-depth assessments.

End-product inhibition and acidification limit biowaste fermentation efficiency.

Converting waste to resource may mitigate environmental pollution and global resource limitation. The platform chemical lactic acid can be produced from biowaste and its liquid fraction after solid-liquid separation. A fermentation step for lactic acid production prior to the conversion of biowaste to methane and organic fertilizer would increase the biowastes value. Despite the huge potential and promising results of the treatment procedure, the reasons for efficiency loss observed previously need to be addressed in order to pave the way for an up-scaling of the fermentation process. Therefore, biowaste was fermented applying pH control, acid extraction and glucose addition in order to counteract reasons such as acidification, end-product inhibition and carbon limitation, respectively. The fermentation was competitive compared to other renewable lactic acid production substrates and reached a maximum productivity of >5 g Clactic acidg(-1)Ch(-1) and a concentration exceeding 30 g L(-1). A combination of acidification and end-product inhibition was identified as major obstacle. Lactobacillus crispatus and its closest relatives were identified as key lactic acid producers within the process using Miseq Illumina sequencing.