A Tool for Classification and Regression Using Random Forest Methodology: Applications to Landslide Susceptibility Mapping and Soil Thickness Modeling

Classification and regression problems are a central issue in geosciences. In this paper, we present Classification and Regression Treebagger (ClaReT), a tool for classification and regression based on the random forest (RF) technique. ClaReT is developed in Matlab and has a simple graphic user interface (GUI) that simplifies the model implementation process, allows the standardization of the method, and makes the classification and regression process reproducible. This tool performs automatically the feature selection based on a quantitative criterion and allows testing a large number of explanatory variables. First, it ranks and displays the parameter importance; then, it selects the optimal configuration of explanatory variables; finally, it performs the classification or regression for an entire dataset. It can also provide an evaluation of the results in terms of misclassification error or root mean squared error. We tested the applicability of ClaReT in two case studies. In the first one, we used ClaReT in classification mode to identify the better subset of landslide conditioning variables (LCVs) and to obtain a landslide susceptibility map (LSM) of the Arno river basin (Italy). In the second case study, we used ClaReT in regression mode to produce a soil thickness map of the Terzona catchment, a small sub-basin of the Arno river basin. In both cases, we performed a validation of the results and a comparison with other state-of-the-art techniques. We found that ClaReT produced better results, with a more straightforward and easy application and could be used as a valuable tool to assess the importance of the variables involved in the modeling.     

Microbial mediated arsenic biotransformation in wetlands

Arsenic (As) is a pervasive environmental toxin and carcinogenic metalloid. It ranks at the top of the US priority List of Hazardous Substances and causes worldwide human health problems. Wetlands, including natural and artificial ecosystems (i.e. paddy soils) are highly susceptible to As enrichment; acting not only as repositories for water but a host of other elemental/chemical moieties. While macroscale processes (physical and geological) supply As to wetlands, it is the micro-scale biogeochemistry that regulates the fluxes of As and other trace elements from the semi-terrestrial to neighboring plant/aquatic/atmospheric compartments. Among these fine-scale events, microbial mediated As biotransformations contribute most to the element’s changing forms, acting as the ‘switch’ in defining a wetland as either a source or sink of As. Much of our understanding of these important microbial catalyzed reactions follows relatively recent scientific discoveries. Here we document some of these key advances, with focuses on the implications that wetlands and their microbial mediated transformation pathways have on the global As cycle, the chemistries of microbial mediated As oxidation, reduction and methylation, and future research priorities areas.

Open image in new window

     

Biogenic polycyclic aromatic hydrocarbons in sediments of the San Joaquin River in California (USA), and current paradigms on their formation

Biogenic perylene and higher plant pentacyclic triterpenoid-derived alkylated and partially aromatized tetra- and pentacyclic derivatives of chrysene (3,4,7-trimethyl- and 3,3,7-trimethyl-1,2,3,4-tetrahydrochrysene, THC) and picene (1,2,9-trimethyl- and 2,2,9-trimethyl-1,2,3,4-tetrahydropicene, THP) were two- to four-fold more abundant than pyrogenic PAH in two sediment cores from the San Joaquin River in Northern California (USA). In a core from Venice Cut (VC), located in the river, PAH concentrations varied little downcore and the whole-core PAH concentration (biogenics?+?pyrogenics) was 250.6?±?73.7 ng g^?1 dw; biogenic PAH constituted 67?±?4 % of total PAH. THC were 26?±?9 % of total biogenic PAH, THP were 36?±?7 %, and perylene was 38?±?7 %. PAH distributions in a core from Franks Tract (FT), a former wetland that was converted to an agricultural tract in the late 1800s and flooded in 1938, were more variable. Surface sediments were dominated by pyrogenic PAH so that biogenic PAH were only ~30 % of total PAH. Deeper in the core, biogenic PAH constituted 60–93 % of total PAH; THC, THP and perylene were 31?±?28 %, 24?±?32 %, and 45?±?36 % of biogenic PAH. At 100–103 cm depth, THP constituted 80 % of biogenic PAH and at 120–123 cm perylene was 95 % of biogenic PAH. Current concepts related to precursors and transformation processes responsible for the diagenetic generation of perylene and triterpenoid-derived PAH are discussed. Distributions of biogenic PAH in VC and FT sediments suggest that they may not form diagenetically within these sediments but rather might be delivered pre-formed from the river’s watershed.     

Suitability of passive sampling for the monitoring of pharmaceuticals in Finnish surface waters

The occurrence of five pharmaceuticals, consisting of four anti-inflammatory and one antiepileptic drug, was studied by passive sampling and grab sampling in northern Lake Päijänne and River Vantaa. The passive sampling was performed by using Chemcatcher® sampler with a SDB-RPS Empore disk as a receiving phase. In Lake Päijänne, the sampling was conducted during summer 2013 at four locations near the discharge point of a wastewater treatment plant and in the years 2013 and 2015 at four locations along River Vantaa. The samples were analyzed by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) in the multiple reaction monitoring mode. The concentrations of carbamazepine, diclofenac, ibuprofen, ketoprofen, and naproxen in Lake Päijänne determined by passive sampling ranged between 1.4–2.9 ng L^?1, 15–35 ng L^?1, 13–31 ng L^?1, 16–27 ng L^?1, and 3.3–32 ng L^?1, respectively. Similarly, the results in River Vantaa ranged between 1.2–40 ng L^?1, 15–65 ng L^?1, 13–33 ng L^?1, 16–31 ng L^?1, and 3.3–6.4 ng L^?1. The results suggest that the Chemcatcher passive samplers are suitable for detecting pharmaceuticals in lake and river waters.     

Coupling fishery dynamics,human health and social learning in a model of fish-borne pollution exposure

Pollution-induced illnesses are caused by toxicants that result from human activity and are often entirely preventable. However, where industrial priorities have undermined responsible governance, exposed populations must reduce their exposure by resorting to voluntary protective measures and demanding emissions abatement. This paper presents a coupled human–environment system model that represents the effects of water pollution on the health and livelihood of a fishing community. The model is motivated by an incident from 1949 to 1968 in Minamata, Japan, where methylmercury effluent from a local factory poisoned fish populations and humans who ate them. We model the critical role of risk perception in driving both social learning and the protective feedbacks against pollution exposure. These feedbacks are undermined in the presence of social misperceptions such as stigmatization of the injured. Through numerical simulation and scenario analysis, we compare our model results with historical datasets from Minamata, and find that the conditions for an ongoing pollution epidemic are highly unlikely without social misperception. We also find trade-offs between human health outcomes, the viability of the polluting industry and the survival of the fishery. We conclude that an understanding of human–environment interactions and misperception effects is highly relevant to the resolution of contemporary pollution problems, and merits further study.     

Connecting the dots: mapping habitat connectivity for tigers in central India

Large connected landscapes are paramount to maintain top predator populations. Across their range, tiger (Panthera tigris) populations occur in small fragmented patches of habitat, often isolated by large distances in human-dominated landscapes. We assessed connectivity between 16 protected areas (PAs) in central India, a global priority landscape for tiger conservation, using data on land use and land cover, human population density, and transportation infrastructure. We identified and prioritized movement routes using a combination of least-cost corridor modeling and circuit theory. Our analyses suggest that there are several opportunities to maintain connectivity in this landscape. We mapped a total of thirty-five linkages in the region and calculated metrics to estimate their quality and importance. The highest quality linkages as measured by the ratio of cost-weighted distance to Euclidean distance are Kanha–Phen/Bandhavgarh–SanjayGhasidas/Melghat–Satpura, and cost-weighted distance to least-cost path length are Nawegaon–Tadoba/Achanakmar–SanjayGhasidas/Kanha–Phen. We used current flow centrality to evaluate the contribution of each PA and linkage toward facilitating animal movement. Values are highest for Kanha and Pench tiger reserves, and the linkages between Kanha–Phen, Kanha–Pench, and Pench–Satpura, suggesting that these PAs and linkages play a critical role in maintaining connectivity in central India. In addition, smaller areas such as Bor, Nawegaon, and Phen have high centrality scores relative to their areas and thus may act as important stepping stones. We mapped pinch points, which are sections of the linkages where tiger movement is restricted due to unfavorable habitat, transportation networks, human habitation, or a combination of factors. Currently, very limited data exist on tiger movement outside of PAs to validate model results. Regional-scale connectivity mapping efforts can assist managers and policy makers to develop strategic plans for balancing wildlife conservation and other land uses in the landscape.     

The city as a refuge for insect pollinators

Research on urban insect pollinators is changing views on the biological value and ecological importance of cities. The abundance and diversity of native bee species in urban landscapes that are absent in nearby rural lands evidence the biological value and ecological importance of cities and have implications for biodiversity conservation. Lagging behind this revised image of the city are urban conservation programs that historically have invested in education and outreach rather than programs designed to achieve high‐priority species conservation results. We synthesized research on urban bee species diversity and abundance to determine how urban conservation could be repositioned to better align with new views on the ecological importance of urban landscapes. Due to insect pollinators’ relatively small functional requirements—habitat range, life cycle, and nesting behavior—relative to larger mammals, we argue that pollinators put high‐priority and high‐impact urban conservation within reach. In a rapidly urbanizing world, transforming how environmental managers view the city can improve citizen engagement and contribute to the development of more sustainable urbanization.     

A new approach for the modelling of chestnut wood photo-degradation monitored by different spectroscopic techniques

The aim of this work is to study the colour and chemical modifications of the surfaces in chestnut wood samples as a consequence of irradiating in a controlled environment. The changes were investigated by a new analytical approach by combining traditional techniques such as reflectance spectrophotometry in the visible range and Fourier transform infrared spectroscopy with new hyperspectral imaging, in order to obtain forecast models to describe the phenomenon. The statistical elaboration of the experimental data allowed to validate the measurements and to obtain models enabling to relate the investigated parameters; the elaboration of the hyperspectral images by chemometric methods allowed for studying the changes in the reflectance spectra. A result of great importance is the possibility to correlate the oxidation of wood chemical components with the colour change in a totally non-invasive modality. This result is particularly relevant in the field of cultural heritage and in general in the control processes of wooden materials.