Assessing the Water Pollution of the Brahmaputra River Using Water Quality Indexes

Water quality is continuously affected by anthropogenic and environmental conditions. A significant issue of the Indian rivers is the massive water pollution, leading to the spreading of different diseases due to its daily use. Therefore, this study investigates three aspects. The first one is testing the hypothesis of the existence of a monotonic trend of the series of eight water parameters of the Brahmaputra River recorded for 17 years at ten hydrological stations. When this hypothesis was rejected, a loess trend was fitted. The second aspect is to assess the water quality using three indicators (WQI)–CCME WQI, British Colombia, and a weighted index. The third aspect is to group the years and the stations in clusters used to determine the regional (spatial) and temporal trend of the WQI series, utilizing a new algorithm. A statistical analysis does not reject the hypothesis of a monotonic trend presence for the spatially distributed data but not for the temporal ones. Hierarchical clustering based on the computed WQIs detected two clusters for the spatially distributed data and two for the temporal-distributed data. The procedure proposed for determining the WQI temporal and regional evolution provided good results in terms of mean absolute error, root mean squared error (RMSE), and mean absolute percentage error (MAPE).

Tile2Vec: Unsupervised Representation Learning for Spatially Distributed Data

Geospatial analysis lacks methods like the word vector representations and pre-trained networks that significantly boost performance across a wide range of natural language and computer vision tasks. To fill this gap, we introduce Tile2Vec, an unsupervised representation learning algorithm that extends the distributional hypothesis from natural language — words appearing in similar contexts tend to have similar meanings — to spatially distributed data. We demonstrate empirically that Tile2Vec learns semantically meaningful representations for both image and non-image datasets. Our learned representations significantly improve performance in downstream classification tasks and, similarly to word vectors, allow visual analogies to be obtained via simple arithmetic in the latent space.

Virtual Sectioning and Haptic Exploration of Volumetric Shapes in the Absence of Visual Feedback

The reduced behavior for exploration of volumetric data based on the virtual sectioning concept was compared with the free scanning at the use of the StickGrip linkage-free haptic device. Profiles of the virtual surface were simulated through the penholder displacements in relation to the pen tip of the stylus. One or two geometric shapes (cylinder, trapezoidal prism, ball, and torus) or their halves and the ripple surface were explored in the absence of visual feedback. In the free scanning, the person physically moved the stylus. In the parallel scanning, cross-sectional profiles were generated automatically starting from the location indicated by the stylus. Analysis of the performance of 18 subjects demonstrated that the new haptic visualization and exploration technique allowed to create accurate mental images, to recognize and identify virtual shapes. The mean number of errors was about 2.5% in the free scanning mode and 1.9% and 1.5% in the parallel scanning mode at the playback velocity of 28 mm/s and 42 mm/s, respectively. All participants agreed that the haptic visualization of the 3D virtual surface presented as the cross-sectional slices of the workspace was robust and easy to use. The method was developed for visualization of spatially distributed data collected by sensors.

Evaluating the effects of heteroscedasticity on the economic growth of EU regions

The hypothesis of homoscedasticity of errors is convenient for the simplification of the estimation procedures. Unfortunately, this assumption is rather restrictive in the case of the analysis of spatially distributed data. Spatial units, in fact, can be very different in size and in other economic characteristics. This circumstance suggests the presence of heteroscedasticity in this typology of data. In this paper we study the effects of heteroscedasticity in regional economic convergence. We use two different estimators of the coefficient of variance and covariance matrix recently introduced in spatial econometrics literature that take into account the heteroscedasticity highlighted by the error terms. This methodology can be considered a suitable alternative to the identification of convergence clubs that represents a very popular approach for the analysis of structural economic differences between regions. The empirical analysis concerns the estimate of conditional economic convergence on EU NUTS 2 regions for the period 1981-2004.