Machine learning offers the potential for effective and efficient classification of remotely sensed imagery. The strengths of machine learning include the capacity to handle data of high dimensionality and to map classes with very complex characteristics. Nevertheless, implementing a machine-learning classification is not straightforward, and the literature provides conflicting advice regarding many key issues. This article therefore provides an overview of machine learning from an applied perspective. We focus on the relatively mature methods of support vector machines, single decision trees (DTs), Random Forests, boosted DTs, artificial neural networks, and k-nearest neighbours (k-NN). Issues considered include the choice of algorithm, training data requirements, user-defined parameter selection and optimization, feature space impacts and reduction, and computational costs. We illustrate these issues through applying machine-learning classification to two publically available remotely sensed data sets.
High spatial resolution (1–5 m) remotely sensed datasets are increasingly being used to map land covers over large geographic areas using supervised machine learning algorithms. Although many studies have compared machine learning classification methods, sample selection methods for acquiring training and validation data for machine learning, and cross-validation techniques for tuning classifier parameters are rarely investigated, particularly on large, high spatial resolution datasets. This work, therefore, examines four sample selection methods—simple random, proportional stratified random, disproportional stratified random, and deliberative sampling—as well as three cross-validation tuning approaches—k-fold, leave-one-out, and Monte Carlo methods. In addition, the effect on the accuracy of localizing sample selections to a small geographic subset of the entire area, an approach that is sometimes used to reduce costs associated with training data collection, is investigated. These methods are investigated in the context of support vector machines (SVM) classification and geographic object-based image analysis (GEOBIA), using high spatial resolution National Agricultural Imagery Program (NAIP) orthoimagery and LIDAR-derived rasters, covering a 2,609 km2 regional-scale area in northeastern West Virginia, USA. Stratified-statistical-based sampling methods were found to generate the highest classification accuracy. Using a small number of training samples collected from only a subset of the study area provided a similar level of overall accuracy to a sample of equivalent size collected in a dispersed manner across the entire regional-scale dataset. There were minimal differences in accuracy for the different cross-validation tuning methods. The processing time for Monte Carlo and leave-one-out cross-validation were high, especially with large training sets. For this reason, k-fold cross-validation appears to be a good choice. Classifications trained with samples collected deliberately (i.e., not randomly) were less accurate than classifiers trained from statistical-based samples. This may be due to the high positive spatial autocorrelation in the deliberative training set. Thus, if possible, samples for training should be selected randomly; deliberative samples should be avoided.
This letter evaluates the effectiveness of a new kernel-based extreme learning machine (ELM) algorithm for a land cover classification using both multi-and hyperspectral remote-sensing data. The results are compared with the most widely used algorithms -support vector machines (SVMs). The results are compared in terms of the ease of use (in terms of the number of user-defined parameters), classification accuracy and computation cost. A radial basis kernel function was used with both the SVM and the kernel-based extreme-learning machine algorithms to ensure compatibility in the comparison of the two algorithms. The results suggest that the new algorithm is similar to, or more accurate than, SVM in terms of classification accuracy, has notable lower computational cost and does not require the implementation of a multiclass strategy.
In the Appalachian region of the eastern United States, mountaintop removal mining (MTM) is a dominant driver of land-cover change, impacting 6.8% of the largely forested 4.86 million ha coal fields region. Recent catastrophic flooding and documented biological impairment downstream of MTM has drawn sharp criticism to this practice. Despite its extent, scale, and use since the 1970s, the impact of MTM on hydrology is poorly understood. Therefore, the goal of this study was a multiscale evaluation to establish the nature of hydrologic impacts associated with MTM. To quantify the extent of MTM, land-cover change over the lifetime of this practice is estimated for a mesoscale watershed in southern West Virginia. To assess hydrologic impacts, we conducted long-term trend analyses to evaluate for systematic changes in hydrology at the mesoscale, and conducted hydrometric and response time modeling to characterize storm-scale responses of a MTM-impacted headwater catchment. Results show a general trend in the conversion of forests to mines, and significant decreases in maximum streamflow and variability, and increases in base-flow ratio attributed to valley fills and deep mine drainage. Decreases in variability are shown across spatial and temporal scales having important implications for water quantity and quality. However, considerable research is necessary to understand how MTM impacts hydrology. In an effort to inform future research, we identify existing knowledge gaps and limitations of our study.
Novel species of fungi described in this study include those from various countries as follows: Australia, Austroboletus asper on soil, Cylindromonium alloxyli on leaves of Alloxylon pinnatum, Davidhawksworthia quintiniae on leaves of Quintinia sieberi, Exophiala prostantherae on leaves of Prostanthera sp., Lactifluus lactiglaucus on soil, Linteromyces quintiniae (incl. Linteromyces gen. nov.) on leaves of Quintinia sieberi, Lophotrichus medusoides from stem tissue of Citrus garrawayi, Mycena pulchra on soil, Neocalonectria tristaniopsidis (incl. Neocalonectria gen. nov.)and Xyladictyochaeta tristaniopsidis on leaves of Tristaniopsis collina, Parasarocladium tasmanniae on leaves of Tasmannia insipida, Phytophthora aquae-cooljarloo from pond water, Serendipita whamiae as endophyte from roots of Eriochilus cucullatus, Veloboletus limbatus (incl. Veloboletus gen. nov.)onsoil. Austria, Cortinarius glaucoelotus onsoil. Bulgaria, Suhomyces rilaensis from the gut of Bolitophagus interruptus found on a Polyporus sp. Canada, Cantharellus betularum among leaf litter of Betula, Penicillium saanichii from house dust. Chile, Circinella lampensis on soil, Exophiala embothrii from rhizosphere of Embothrium coccineum. China, Colletotrichum cycadis on leaves of Cycas revoluta. Croatia, Phialocephala melitaea on fallen branch of Pinus halepensis. Czech Republic, Geoglossum jirinae on soil, Pyrenochaetopsis rajhradensis from dead wood of Buxus sempervirens. Dominican Republic, Amanita domingensis on litter of deciduous wood, Melanoleuca dominicana on forest litter. France, Crinipellis nigrolamellata (Martinique) on leaves of Pisonia fragrans, Talaromyces pulveris from bore dust of Xestobium rufovillosum infesting floorboards. French Guiana, Hypoxylon hepaticolor on dead corticated branch. Great Britain, Inocybe ionolepis on soil. India, Cortinarius indopurpurascens among leaf litter of Quercus leucotrichophora. Iran, Pseudopyricularia javanii on infected leaves of Cyperus sp., Xenomonodictys iranica (incl. Xenomonodictys gen. nov.) on wood of Fagus orientalis. Italy, Penicillium vallebormidaense from compost. Namibia, Alternaria mirabibensis on plant litter, Curvularia moringae and Moringomyces phantasmae (incl. Moringomyces gen. nov.) on leaves and flowers of Moringa ovalifolia, Gobabebomyces vachelliae (incl. Gobabebomyces gen. nov.) on leaves of Vachellia erioloba, Preussia procaviae on dung of Procavia capensis. Pakistan, Russula shawarensis from soil on forest floor. Russia, Cyberlindnera dauci from Daucus carota. South Africa, Acremonium behniae on leaves of Behnia reticulata, Dothiora aloidendri and Hantamomyces aloidendri (incl. Hantamomyces gen. nov.) on leaves of Aloidendron dichotomum, Endoconidioma euphorbiae on leaves of Euphorbia mauritanica , Eucasphaeria proteae on leaves of Protea neriifolia , Exophiala mali from inner fruit tissue of Malus sp., Graminopassalora geissorhizae on leaves of Geissorhiza splendidissima, Neocamarosporium leipoldtiae on leaves of Leipoldtia schultzii, Neocladosporium osteospermi on leaf spots of Osteospermum moniliferum, Neometulocladosporiella seifertii on leaves of Combretum caffrum, Paramyrothecium pituitipietianum on stems of Grielum humifusum, Phytopythium paucipapillatum from roots of Vitis sp., Stemphylium carpobroti and Verrucocladosporium carpobroti on leaves of Carpobrotus quadrifolius, Suttonomyces cephalophylli on leaves of Cephalophyllum pilansii. Sweden, Coprinopsis rubra on cow dung, Elaphomyces nemoreus fromdeciduouswoodlands. Spain, Polyscytalum pini-canariensis on needles of Pinus canariensis, Pseudosubramaniomyces septatus from stream sediment, Tuber lusitanicum on soil under Quercus suber. Thailand, Tolypocladium flavonigrum on Elaphomyces sp. USA, Chaetothyrina spondiadis on fruits of Spondias mombin, Gymnascella minnisii from bat guano, Juncomyces patwiniorum on culms of Juncus effusus, Moelleriella puertoricoensis on scale insect, Neodothiora populina (incl. Neodothiora gen. nov.) on stem cankers of Populus tremuloides, Pseudogymnoascus palmeri fromcavesediment. Vietnam, Cyphellophora vietnamensis on leaf litter, Tylopilus subotsuensis on soil in montane evergreen broadleaf forest. Morphological and culture characteristics are supported by DNA barcodes.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.