Predicting the eye fixation locations in the gray scale images in the visual scenes with different semantic contents

In recent years, there has been considerable interest in visual attention models (saliency map of visual attention). These models can be used to predict eye fixation locations, and thus will have many applications in various fields which leads to obtain better performance in machine vision systems. Most of these models need to be improved because they are based on bottom-up computation that does not consider top-down image semantic contents and often does not match actual eye fixation locations. In this study, we recorded the eye movements (i.e., fixations) of fourteen individuals who viewed images which consist natural (e.g., landscape, animal) and man-made (e.g., building, vehicles) scenes. We extracted the fixation locations of eye movements in two image categories. After extraction of the fixation areas (a patch around each fixation location), characteristics of these areas were evaluated as compared to non-fixation areas. The extracted features in each patch included the orientation and spatial frequency. After feature extraction phase, different statistical classifiers were trained for prediction of eye fixation locations by these features. This study connects eye-tracking results to automatic prediction of saliency regions of the images. The results showed that it is possible to predict the eye fixation locations by using of the image patches around subjects’ fixation points.     

Land degradation risk dynamics assessment in red and lateritic zones of eastern plateau,India: A combine approach of K-fold CV,data mining and field validation

The phenomenon of land degradation is a serious environmental issue that affects many countries worldwide, particularly in the developing countries of sub-tropical regions like India. The assessment of land degradation is necessary for designing a mitigation plan that will reduce the adverse effects of land degradation. Recently, sensitivity models for analyzing land degradation have become a popular scientific tool for determining the spatial characteristics of this complicated environmental phenomenon. The objective of the current study is to prepare land degradation susceptibility maps for the gravely undulating red and lateritic agro-climatic zones (ACZ) of the Eastern plateau, India using hybrid techniques, i.e., integration of K-Fold cross-validation (CV) and machine learning algorithms of Reduced Error Pruning Tree (REPTree) and the ensemble of Bagging-REPTree and Boosting-REPTree. For the modelling purpose, sixteen independent land degradation conditioning factors were selected based on a multi-collinearity test, and dependent factors, i.e., gully and ravine points, were collected from published reports and field investigations. The evaluation result of the models indicates that Boosting-REPTree is the most optimal in prediction analysis, as the area under the curve (AUC) of training and validation is 0.944 and 0.928, respectively, in K-Fold 1 followed by Bagging-REPTree and REPTree. As a result, this study suggested that the ensemble of the Boosting-REPTree model can be applied as a new potential method for spatial prediction of land degradation in future research. The study also revealed that ex-situ plant species had been adopted to control soil erosion. Still, it is considered a false measure as ex-situ plant species cannot prevent erosion to an optimal level. Overall, a land degradation prevention planning map has also been suggested to measure soil erosion.     

On the possibility for generic modeling of submarine groundwater discharge

We simulate large-scale dynamics of submarine groundwater discharge (SGD) in three different coastal aquifers on the Mediterranean Sea. We subject these aquifers to a wide range of different groundwater management conditions, leading to widely different net groundwater drainage from land to sea. The resulting SGD at steady-state is quantifiable and predictable by simple linearity in the net land-determined groundwater drainage, defined as total fresh water drainage minus groundwater extraction in the coastal aquifer system. This linearity appears to be general and independent of site-specific, variable and complex details of hydrogeology, aquifer hydraulics, streamlines and salinity transition zones in different coastal systems. Also independently of site-specifics, low SGD implies high seawater content due to seawater intruding into the aquifer and mixing with fresh groundwater within a wide salinity transition zone in the aquifer. Increasing SGD implies decreasing seawater content, decreased mixing between seawater and fresh groundwater and narrowing of the salinity transition zone of brackish groundwater in the aquifer.     

Challenges of predicting the potential distribution of a slow-spreading invader: a habitat suitability map for an invasive riparian tree

Understanding the potential spread of invasive species is essential for land managers to prevent their establishment and restore impacted habitat. Habitat suitability modeling provides a tool for researchers and managers to understand the potential extent of invasive species spread. Our goal was to use habitat suitability modeling to map potential habitat of the riparian plant invader, Russian olive (Elaeagnus angustifolia). Russian olive has invaded riparian habitat across North America and is continuing to expand its range. We compiled 11 disparate datasets for Russian olive presence locations (n = 1,051 points and 139 polygons) in the western US and used Maximum entropy (Maxent) modeling to develop two habitat suitability maps for Russian olive in the western United States: one with coarse-scale water data and one with fine-scale water data. Our models were able to accurately predict current suitable Russian olive habitat (Coarse model: training AUC = 0.938, test AUC = 0.907; Fine model: training AUC = 0.923, test AUC = 0.885). Distance to water was the most important predictor for Russian olive presence in our coarse-scale water model, but it was only the fifth most important variable in the fine-scale model, suggesting that when water bodies are considered on a fine scale, Russian olive does not necessarily rely on water. Our model predicted that Russian olive has suitable habitat further west from its current distribution, expanding into the west coast and central North America. Our methodology proves useful for identifying potential future areas of invasion. Model results may be influenced by locations of cultivated individuals and sampling bias. Further study is needed to examine the potential for Russian olive to invade beyond its current range. Habitat suitability modeling provides an essential tool for enhancing our understanding of invasive species spread.     

A Habitat Suitability Index (HSI) for the Western Prairie Fringed Orchid (Platanthera praeclara) on the Sheyenne National Grassland,North Dakota,USA

The Western Prairie Fringed Orchid (Platanthera praeclara) is a threatened species found on the Sheyenne National Grassland (SNG) in southeast North Dakota, USA. The SNG is subject to management for multiple uses including biodiversity conservation, livestock grazing and recreation. Therefore, there is a need for the development of indicators of suitable orchid habitat. The orchids are continuously monitored, but understanding of the relationship between landscape properties and orchid locations is limited. In this study data that characterize topography, moisture, and groundwater were used to construct indicators of landscape suitability and an overall Habitat Suitability Index (HSI) for the orchid. A LiDAR-derived DEM and groundwater well observations were used to develop landscape indicators. The Topographic Wetness Index (TWI: a measure of moisture on the landscape), the Topographic Position Index (TPI: a measure of position on the landscape), and the distance to groundwater (DTG: a measure of the distance from the land surface to the groundwater surface) provided the best set of indicators of orchid habitat. Point-based field observations of orchid occurrence were used to develop Orchid Suitability Metrics (OSMs) that identified the range of indicator values most strongly associated with orchids. These OSMs were used to define year by year suitability zones for each indicator that were combined to create the HSI. Comparison of orchid locations with groundwater elevations showed that orchids occurred on average 0.98 ± 0.39 (2σ) m above the water table. TWI and TPI demonstrated that orchids occur near flow paths and areas of lower elevation than their surroundings. HSI values of 0.67 and above were associated with 89.8% of all orchid observations used in the analysis. The landscape indicators, OSM concept and HSI could be generally applied to monitoring and conservation management of orchid habitat and the concept may be applicable to other valued species with similar niche properties.     

Mapping Potential Amplification and Transmission Hotspots for MERS-CoV,Kenya

Dromedary camels have been implicated consistently as the source of Middle East respiratory syndrome coronavirus (MERS-CoV) human infections and attention to prevent and control it has focused on camels. To understanding the epidemiological role of camels in the transmission of MERS-CoV, we utilized an iterative empirical process in Geographic Information System (GIS) to identify and qualify potential hotspots for maintenance and circulation of MERS-CoV, and produced risk-based surveillance sites in Kenya. Data on camel population and distribution were used to develop camel density map, while camel farming system was defined using multi-factorial criteria including the agro-ecological zones (AEZs), production and marketing practices. Primary and secondary MERS-CoV seroprevalence data from specific sites were analyzed, and location-based prevalence matching with camel densities was conducted. High-risk convergence points (migration zones, trade routes, camel markets, slaughter slabs) were profiled and frequent cross-border camel movement mapped. Results showed that high camel-dense areas and interaction (markets and migration zones) were potential hotspot for transmission and spread. Cross-border contacts occurred with in-migrated herds at hotspot locations. AEZ differential did not influence risk distribution and plausible risk factors for spatial MERS-CoV hotspots were camel densities, previous cases of MERS-CoV, high seroprevalence and points of camel convergences. Although Kenyan camels are predisposed to MERS-CoV, no shedding is documented to date. These potential hotspots, determined using anthropogenic, system and trade characterizations should guide selection of sampling/surveillance sites, high-risk locations, critical areas for interventions and policy development in Kenya, as well as instigate further virological examination of camels.     

Multi objective land allocation (MOLA) for zoning Ghamishloo Wildlife Sanctuary in Iran

Protected area zoning is an approach towards decreasing conflict between the possible uses of land and providing an opportunity for policy making. GIS data processing and spatial analysis along with decision analysis techniques, were used in this study to define zones for Ghamishloo Wildlife Sanctuary according to I.U.C.N. category IV in Isfahan Province of Iran. We used multi-criteria evaluation and multi-objective land allocation for zoning the sanctuary, which covers an area of about 866 km². First, we prepared a land use map of the area using classification of the IRS 6 (AWiFS) data of May 2005. For zoning this region, nine major criteria including wildlife habitat, vegetation cover, soil, distance to historical places, water resources, road, scenic beauties in the landscape, and also to residential areas, and to the core zone were considered. We used the analytical hierarchy process to derive weights of the criteria and then applied a weighted linear combination technique to combine the factors. The degree of suitability was defined by applying Fuzzy membership function. The wildlife sanctuary was divided into four zones including conservation, recreation, rehabilitation, and cultural zones, consisting of 69%, 21%, 9.5% and 0.5% of the area, respectively. Finally, multi-objective land allocation (MOLA) function was used for allocation of the sanctuary's land area to the zones which produced reasonable results.     

Suitability index for restoration in landscapes: An alternative proposal for restoration projects

Forest fragmentation constitutes one of the main consequences of land cover change worldwide. Through this process gaps in habitat coverage are created and the ability of populations in the remaining fragments to maintain themselves is put in doubt. Hence, two options need to be considered: conserving the remaining forest fragments, and restoring habitat in some deforested patches with the aim of reestablishing the connections among the fragments. We established a mathematical index (SIR) that describes the suitability of individual habitat patches for restoration within a landscape. The index considers classes of distances among fragments and categories of habitat quality in the areas surrounding the fragments to assess habitat quality in terms of probability of dispersal and survival of propagules (especially seeds and cutting). In the present study, we created detailed maps depicting SIR values for two periods (1988 and 2011) for Sorocaba region (São Paulo State, Brazil). We derived land cover maps from satellite images for the two years of our study, and then surveyed the transition of land cover categories and landscape metrics between years. A model for the SIR was created using a map of distance classes among fragments and also a map of habitat quality established according to each land cover category. For both 1988 and 2011, pasture was the predominant land cover category. The main land cover transitions were from pasture to urban (10.6%) and from pasture to forest fragments (13.4%). Although the land cover class “wood sites” increased, the data of SIR revealed that the areas of habitat categorized as excellent and good both decreased, while habitat classes categorized as poor and very poor increased. Our model has the potential to be applied to other regions where the forest is fragmented. Hence, local policy makers will be able to use this model to determine local patches of high value for conservation and also the most ideal locations for restoration projects.     

Predicting the cover and richness of intertidal macroalgae in remote areas: a case study in the Antarctic Peninsula

Antarctica is an iconic region for scientific explorations as it is remote and a critical component of the global climate system. Recent climate change causes a dramatic retreat of ice in Antarctica with associated impacts to its coastal ecosystem. These anthropogenic impacts have a potential to increase habitat availability for Antarctic intertidal assemblages. Assessing the extent and ecological consequences of these changes requires us to develop accurate biotic baselines and quantitative predictive tools. In this study, we demonstrated that satellite‐based remote sensing, when used jointly with in situ ground‐truthing and machine learning algorithms, provides a powerful tool to predict the cover and richness of intertidal macroalgae. The salient finding was that the Sentinel‐based remote sensing described a significant proportion of variability in the cover and richness of Antarctic macroalgae. The highest performing models were for macroalgal richness and the cover of green algae as opposed to the model of brown and red algal cover. When expanding the geographical range of the ground‐truthing, even involving only a few sample points, it becomes possible to potentially map other Antarctic intertidal macroalgal habitats and monitor their dynamics. This is a significant milestone as logistical constraints are an integral part of the Antarctic expeditions. The method has also a potential in other remote coastal areas where extensive in situ mapping is not feasible.     

Summer distribution of Red-legged Partridges Alectoris rufa in relation to water availability on Mediterranean farmland

The suggestion that the summer distribution of Red-legged Partridges Alectoris rufa in the Mediterranean region is determined by the availability of surface water was examined on the agricultural farm, Alto Alentejo, southern Portugal. Partridge coveys were surveyed between 15 July and 15 August in 1993 and 1994. Using a vector-based Geographic Information System, we assessed, for each covey location and for the locations of a double number of random points, the distance to the nearest water point, distance to field boundaries, distance to water lines and land use classes. Univariate comparisons were made between the two groups of locations, and three multivariate logistic models were fitted through forward stepwise selection to the 1993, 1994 and pooled data sets to estimate the probability of sighting partridge coveys in the study area. Distance to water was significantly lower for partridge locations than for random points in both years and was the only variable selected for all logistic models. Apart from water availability, Red-legged Partridge locations were also affected by land use and distance to field boundaries.     

Feature Selection and the Class Imbalance Problem in Predicting Protein Function from Sequence

When the standard approach to predict protein function by sequence homology fails, other alternative methods can be used that require only the amino acid sequence for predicting function. One such approach uses machine learning to predict protein function directly from amino acid sequence features. However, there are two issues to consider before successful functional prediction can take place: identifying discriminatory features, and overcoming the challenge of a large imbalance in the training data. We show that by applying feature subset selection followed by undersampling of the majority class, significantly better support vector machine (SVM) classifiers are generated compared with standard machine learning approaches. As well as revealing that the features selected could have the potential to advance our understanding of the relationship between sequence and function, we also show that undersampling to produce fully balanced data significantly improves performance. The best discriminating ability is achieved using SVMs together with feature selection and full undersampling; this approach strongly outperforms other competitive learning algorithms. We conclude that this combined approach can generate powerful machine learning classifiers for predicting protein function directly from sequence.     

Topographic spatial characterisation of grey seal Halichoerus grypus breeding habitat at a sub-seal size spatial grain

Expansion within breeding colonies may critically depend upon the availability of suitable breeding habitat. Here we use topographic modelling in a GIS to characterise suitable pupping habitat and accurately predict the pattern of colonisation in an expanding grey seal breeding colony – the Isle of May (Scotland). We use high resolution images from large format aerial photographs of the colony to generate sub‐metre accurate Digital Terrain Models (DTMs). GIS modelling with these DTMs provides topographic measures of elevation, slope and ease of access to sea and freshwater pools at a 2 m grid cell size. Seal locations during the 1994 breeding season, with sex‐age class, were also digitised from the same images. We examine how the physical attributes of cells (locations) with and without pups differ and identify areas suitable for pupping but remaining unoccupied during 1994. We predict patterns of future colonisation by characterising areas differentiated by the densities of pups within 5 m grid cells and identifying areas, both occupied or unoccupied, with a potential for increased future pupping densities. Our predictions were tested by examining pup distributions observed in the 1998 breeding season. Occupied sites were significantly closer to freshwater pools and access to the sea (p<0.001) than unoccupied sites suggesting that proximity to water may restrict colony expansion before all areas of suitably flat terrain are occupied. All pup density classes occurred in sites with similar slope values and distance to pools. However, higher pupping densities occurred closer to access points (p=0.014). Pup densities observed in 1998 revealed that our 1994 predictions were accurate (p<0.0001). Only 12% of 466 grid cells had higher densities in 1998 than predicted, of which 88% differed by only 1 pup. These incorrectly classified cells occurred at the expanding edge of the colony (in a more topographically homogenous area) and at the main access points from the sea (major traffic zones). These results demonstrate the value of the accurate quantification of topographic parameters at the appropriate spatial grain (in this case below the size of the individual) for use in habitat classification and predictions of habitat utilisation.     

pathmatrix: a geographical information system tool to compute effective distances among samples

pathmatrix is a tool used to compute matrices of effective geographical distances among samples using a least‐cost path algorithm. This program is dedicated to the study of the role of the environment on the spatial genetic structure of populations. Punctual locations (e.g. individuals) or zones encompassing sample data points (e.g. demes) are used in conjunction with a species‐specific friction map representing the cost of movement through the landscape. Matrices of effective distances can then be exported to population genetic software to test, for example, for isolation by distance. pathmatrix is an extension to the geographical information system (GIS) software arcview 3.x.     

Precise vegetation restoration in arid regions based on potential natural vegetation and potential normalized difference vegetation index

Precise vegetation restoration is critical in drylands, as some inappropriate restoration attempts have even increased water scarcity and degradation in afforestation areas. Potential natural vegetation (PNV) is widely used to provide a reference for the appropriate location and vegetation type of restoration programs while the appropriate restored areas remain unknown. Therefore, we proposed a PNV–potential normalized difference vegetation index (PNDVI) coupling framework based on multiple machine learning (ML) algorithms for precise dryland vegetation restoration. Taking the lower Tarim River Basin (LTRB) with a total area of 1,182 km² as a case study, its present suitable restoration locations, area, and appropriate planting species were quantitatively estimated. The results showed that the model developed by incorporating PNDVI into PNV with easily measurable and available data such as temperature and soil properties can accurately identify dryland restoration patterns. In LTRB, the potentially suitable habitats of trees and grass are closer to the riverbank, while shrubby habitats are further away from the course, covering 1.88, 2.96, and 25.12 km², respectively. There is still enormous land potential for further expansion of the current trees and grass in the LTRB, with 2.56 and 1.54% of existing land supposed to be trees and grass, respectively. This study's novel aspect is combining PNV and PNDVI to quantify and estimate precise restoration patterns through multiple ML algorithms. The model developed here can be used to evaluate the suitable reforestation locations, area, and vegetation types in drylands and to provide a basis for precise vegetation restoration.     

Spatial Analysis of Anthropogenic Landscape Disturbance and Buruli Ulcer Disease in Benin

Background

Land use and land cover (LULC) change is one anthropogenic disturbance linked to infectious disease emergence. Current research has focused largely on wildlife and vector-borne zoonotic diseases, neglecting to investigate landscape disturbance and environmental bacterial infections. One example is Buruli ulcer (BU) disease, a necrotizing skin disease caused by the environmental pathogen Mycobacterium ulcerans (MU). Empirical and anecdotal observations have linked BU incidence to landscape disturbance, but potential relationships have not been quantified as they relate to land cover configurations.

Methodology/Principal Findings

A landscape ecological approach utilizing Bayesian hierarchical models with spatial random effects was used to test study hypotheses that land cover configurations indicative of anthropogenic disturbance were related to Buruli ulcer (BU) disease in southern Benin, and that a spatial structure existed for drivers of BU case distribution in the region. A final objective was to generate a continuous, risk map across the study region. Results suggested that villages surrounded by naturally shaped, or undisturbed rather than disturbed, wetland patches at a distance within 1200m were at a higher risk for BU, and study outcomes supported the hypothesis that a spatial structure exists for the drivers behind BU risk in the region. The risk surface corresponded to known BU endemicity in Benin and identified moderate risk areas within the boundary of Togo.

Conclusions/Significance

This study was a first attempt to link land cover configurations representative of anthropogenic disturbances to BU prevalence. Study results identified several significant variables, including the presence of natural wetland areas, warranting future investigations into these factors at additional spatial and temporal scales. A major contribution of this study included the incorporation of a spatial modeling component that predicted BU rates to new locations without strong knowledge of environmental factors contributing to disease distribution.     

Assessing the application of a geographic presence-only model for land suitability mapping

Recent advances in ecological modeling have focused on novel methods for characterizing the environment that use presence-only data and machine-learning algorithms to predict the likelihood of species occurrence. These novel methods may have great potential for land suitability applications in the developing world where detailed land cover information is often unavailable or incomplete. This paper assesses the adaptation and application of the presence-only geographic species distribution model, MaxEnt, for agricultural crop suitability mapping in a rural Thailand where lowland paddy rice and upland field crops predominant. To assess this modeling approach, three independent crop presence datasets were used including a social-demographic survey of farm households, a remote sensing classification of land use/land cover, and ground control points, used for geodetic and thematic reference that vary in their geographic distribution and sample size. Disparate environmental data were integrated to characterize environmental settings across Nang Rong District, a region of approximately 1300 sq. km in size. Results indicate that the MaxEnt model is capable of modeling crop suitability for upland and lowland crops, including rice varieties, although model results varied between datasets due to the high sensitivity of the model to the distribution of observed crop locations in geographic and environmental space. Accuracy assessments indicate that model outcomes were influenced by the sample size and the distribution of sample points in geographic and environmental space. The need for further research into accuracy assessments of presence-only models lacking true absence data is discussed. We conclude that the MaxEnt model can provide good estimates of crop suitability, but many areas need to be carefully scrutinized including geographic distribution of input data and assessment methods to ensure realistic modeling results.     

phylin: an r package for phylogeographic interpolation

phylin is a package for the r programming environment which offers different methods to spatially interpolate genetic information from phylogeographic data. These interpolations can be used to predict the spatial occurrence of different lineages within a phylogeny using a modified method of kriging, which allows the usage of a genetic distance matrix to derive a model of spatial dependence. phylin improves the available methods to generate interpolated surfaces from a phylogenetic trees by assessing the autocorrelation structure of the genetic information, interpolating the genetic data based on a statistical model, estimating the uncertainty of the predictions and identifying lineage occurrence and contact zones probability without projection of pairwise genetic distances into mid‐points between sample locations. The package also includes methods to plot interpolation surfaces and provide summary tables from the generated data and models. We provide an example of the usefulness of this tool by inferring the spatial occurrence of distinct historical evolutionary lineages of the Lataste's viper (Vipera latastei Boscá, 1878) in the Iberian Peninsula and identifying potential contact areas. The maps of phylogenetic patterns obtained with these methods provide a spatial context to test hypotheses related to processes underlying the geographic distribution of genetic diversity and to inform conservation planning.     

“交通廊道蔓延”视角下山地城市典型样带空间格局梯度分析

空间可达性的强弱在一定程度上决定着土地利用变化的方向和强度,因此交通的廊道作用对城市化进程发挥着重要作用。以重庆市轨道交通三号线沿线双侧各3000 m缓冲样带为研究区,采用面向对象分类方法将四期Landsat TM/ETM遥感影像进行土地利用制图,选择核密度估算、指数分析和梯度分析的空间分析方法,研究交通廊道的变化及其对区域空间格局变化的影响。研究表明,研究期内,样带空间的组分和格局发生了较为明显的变化。道路的空间变化和区域空间格局的变化表现出较为明显的梯度性:道路密度、土地利用的变化强度随距离交通主干道距离的远近依次减弱,道路密度等级范围内的土地利用变化强度随等级的降低出现由弱到强再转弱的特征。     

A geographical information system model for creating bioclimatic maps – examples from a high,mid-latitude city

This paper presents a method for creating large-scale bioclimatic maps with the aid of a geographical information system, GIS. Meteorological data are linked with geographical information about land use, elevation and distance to the coast, in order to generate spatial distributions of physiological equivalent temperature, PET. The model combines an air temperature map and a wind map in order to create different zones for which the thermal component is to be calculated. The advantage of the model presented is that it uses generally available information about land use, altitude and distance to the coast. Further, the model uses a GIS application, which makes it non-static. Compared to most other models, a wide range of observations are used as input. Few biometeorological studies have been performed in high-latitude areas. This paper presents bioclimatic maps for the G?teborg urban area, in Sweden, for the month of July. The results show large variations in PET during a clear, calm day at 1200 hours (Delta T 13.4 degrees C) and during average conditions in July (Delta T 6.8 degrees C), which gives an indication of the magnitude and the spatial variations within high, midlatitude, urban area in summer. The highest PET values were found in the central built-up areas and the lowest PET values in the coastal and green areas. The model generates valuable information for urban planners and decision makers when planning and constructing new areas for outdoor activities etc. This information is also useful in the fields of health and energy.