Boosting for Mineral Prospectivity Modeling: A New GIS Toolbox     

Brandmeier  Melanie  Cabrera Zamora  Irving Gibran  Nykänen  Vesa  Middleton  Maarit 《Natural Resources Research》2020,29(1):71-88

With an increasing demand for raw materials, predictive models that support successful mineral exploration targeting are of great importance. We evaluated different machine learning techniques with an emphasis on boosting algorithms and implemented them in an ArcGIS toolbox. Performance was tested on an exploration dataset from the Iberian Pyrite Belt (IPB) with respect to accuracy, performance, stability, and robustness. Boosting algorithms are ensemble methods used in supervised learning for regression and classification. They combine weak classifiers, i.e., classifiers that perform slightly better than random guessing to obtain robust classifiers. Each time a weak learner is added; the learning set is reweighted to give more importance to misclassified samples. Our test area, the IPB, is one of the oldest mining districts in the world and hosts giant volcanic-hosted massive sulfide (VMS) deposits. The spatial density of ore deposits, as well as the size and tonnage, makes the area unique, and due to the high data availability and number of known deposits, well-suited for testing machine learning algorithms. We combined several geophysical datasets, as well as layers derived from geological maps as predictors of the presence or absence of VMS deposits. Boosting algorithms such as BrownBoost and Adaboost were tested and compared to Logistic Regression (LR), Random Forests (RF) and Support Vector machines (SVM) in several experiments. We found performance results relatively similar, especially to BrownBoost, which slightly outperformed LR and SVM with respective accuracies of 0.96 compared to 0.89 and 0.93. Data augmentation by perturbing deposit location led to a 7% improvement in results. Variations in the split ratio of training and test data led to a reduction in the accuracy of the prediction result with relative stability occurring at a critical point at around 26 training samples out of 130 total samples. When lower numbers of training data were introduced accuracy dropped significantly. In comparison with other machine learning methods, Adaboost is user-friendly due to relatively short training and prediction times, the low likelihood of overfitting and the reduced number of hyperparameters for optimization. Boosting algorithms gave high predictive accuracies, making them a potential data-driven alternative for regional scale and/or brownfields mineral exploration.

  相似文献   

Fluid injection monitoring using electrical resistivity tomography — five years of injection at Ketzin,Germany          下载免费PDF全文

P. Bergmann  C. Schmidt‐Hattenberger  T. Labitzke  F.M. Wagner  A. Just  C. Flechsig  D. Rippe 《Geophysical Prospecting》2017,65(3):859-875

Between the years 2008 and 2013, approximately 67 kilotons of CO₂ have been injected at the Ketzin site, Germany. As part of the geophysical monitoring programme, time‐lapse electrical resistivity tomography has been applied using crosshole and surface‐downhole measurements of electrical resistivity tomography. The data collection of electrical resistivity tomography is partly based on electrodes that are permanently installed in three wells at the site (one injection well and two observation wells). Both types of ERT measurements consistently show the build‐up of a CO₂‐related resistivity signature near the injection point. Based on the imaged resistivity changes and a petrophysical model, CO₂ saturation levels are estimated. These CO₂ saturations are interpreted in conjunction with CO₂ saturations inferred from neutron‐gamma loggings. Apart from the CO₂–brine substitution response in the observed resistivity changes, significant imprints from the dynamic behaviour of the CO₂ in the reservoir are observed.  相似文献   

High temperature gas–solid reactions in calc–silicate Cu–Au skarn formation; Ertsberg,Papua Province,Indonesia     

Richard?W.?HenleyEmail authorView author&#;s OrcID profile  Frank?J.?Brink  Penelope?L.?King  Clyde?Leys  Jibamitra?Ganguly  Terrance?Mernagh  Jill?Middleton  Christian?J.?Renggli  Melanie?Sieber  Ulrike?Troitzsch  Michael?Turner 《Contributions to Mineralogy and Petrology》2017,172(11-12):106

The 2.7–3 Ma Ertsberg East Skarn System (Indonesia), adjacent to the giant Grasberg Porphyry Copper deposit, is part of the world’s largest system of Cu–Au skarn deposits. Published fluid inclusion and stable isotope data show that it formed through the flux of magma-derived fluid through contact metamorphosed carbonate rock sequences at temperatures well above 600° C and pressures of less than 50 MPa. Under these conditions, the fluid has very low density and the properties of a gas. Combining a range of micro-analytical techniques, high-resolution QEMSCAN mineral mapping and computer-assisted X-ray micro-tomography, an array of coupled gas–solid reactions may be identified that controlled reactive mass transfer through the ~ 1 km³ hydrothermal skarn system. Vacancy-driven mineral chemisorption reactions are identified as a new type of reactive transport process for high-temperature skarn alteration. These gas–solid reactions are maintained by the interaction of unsatisfied bonds on mineral surfaces and dipolar gas-phase reactants such as SO₂ and HCl that are continuously supplied through open fractures and intergranular diffusion. Principal reactions are (a) incongruent dissolution of almandine-grossular to andradite and anorthite (an alteration mineral not previously recognized at Ertsberg), and (b) sulfation of anorthite to anhydrite. These sulfation reactions also generate reduced sulfur with consequent co-deposition of metal sulfides. Diopside undergoes similar reactions with deposition of Fe-enriched pyroxene in crypto-veins and vein selvedges. The loss of calcium from contact metamorphic garnet to form vein anhydrite necessarily results in Fe-enrichment of wallrock, and does not require Fe-addition from a vein fluid as is commonly assumed.  相似文献   

Diatom-inferred depth models in 8 Canadian boreal lakes: inferred changes in the benthic:planktonic depth boundary and implications for assessment of past droughts     

Kathleen R. Laird  Melanie V. Kingsbury  C.F. Michael Lewis  Brian F. Cumming 《Quaternary Science Reviews》2011,30(9-10):1201-1217

Assessment of past drought in boreal regions, a region predicted to be strongly affected by climate warming, can provide insights into future availability of water. However, limited instrumental data and paleoclimatic data are available for this assessment. To address this lack of data in the boreal region of northwest Ontario, a regional study of lakes in the Winnipeg River Drainage Basin was initiated. Diatom-inferred (D-I) depth models were developed based on surface samples collected along a depth gradient within 8 small boreal lakes. Weighted-averaging and modern analog approaches provided robust within-lake depth models for each of the study lakes, with bootstrapped r² values ranging from 0.90 to 0.98, and root-mean-squared-errors of prediction (RMSEP) between 1.1 and 2.5 m. Large differences in the estimated depth optima for three representative, but common diatom species across our 8 study lakes suggested that within-lake calibration datasets are more appropriate for inferring past drought based in depth models than a regional multi-lake calibration dataset, and that light and related variables are controlling factors governing the maximum depth of benthic taxa. A down-core application of the D-I depth models on a near-shore core from Meekin Lake, retrieved near the present-day ecotone between the benthic and planktonic diatom assemblages indicated highly similar trends in inferred depth (r = 0.96). The models have significant correlations with other metrics of changes in depth including diatom species richness (r = 0.74–0.78) and evenness (r = 0.76–0.8), thereby allowing a check on the strength and direction of the depth inferences down-core. Near-shore cores located near the benthic:planktonic transition is a sensitive region that can provide estimates of past droughts in lakes where such inferences have been difficult to estimate.  相似文献   

A high-resolution Late Glacial to Holocene record of environmental change in the Mediterranean from Lake Ohrid (Macedonia/Albania)     

Jack H. Lacey  Alexander Francke  Melanie J. Leng  Christopher H. Vane  Bernd Wagner 《International Journal of Earth Sciences》2015,104(6):1623-1638

  相似文献   

Reconstruction of the Surface-Layer Vertical Structure from Measurements of Wind,Temperature and Humidity at Two Levels     

Juan Carlos Bergmann 《Boundary-Layer Meteorology》2011,139(1):161-165

A comparison of sensible heat flux derived from ultrasonic-anemometer measurements and profile-derived values shows strong systematic bias in the stable branch. One of the two effects, large sonic-based values combined with small profile-based values, can be explained as due to katabatic flows down the terrain slope. The other effect, large profile-based values combined with small sonic-based values, can only partially be explained as due to intermittent turbulence. The remainder occurs at small stability and indicates a possible systematic kink in ultrasonic-anemometer eddy-correlation values. Evaluation of the temperature-profile parameter θ _* contributes to further clarification by reducing the systematic bias and confirming the katabatic-flow explanation. Also, the intermittent-turbulence influence is confirmed by the θ _* evaluation, but the combination of large profile-based values with small sonic-based values does not vanish outside the more clearly defined intermittent-turbulence region. This is associated with a strongly increased occurrence of opposite-sign sonic values. Neutral profile values are frequently associated with considerably unstable sonic values of θ _*.  相似文献   

Bacterial communities are sensitive indicators of contaminant stress     

Sun MY  Dafforn KA  Brown MV  Johnston EL 《Marine pollution bulletin》2012,64(5):1029-1038

With many environments worldwide experiencing at least some degree of anthropogenic modification, there is great urgency to identify sensitive indicators of ecosystem stress. Estuarine organisms are particularly vulnerable to anthropogenic contaminants. This study presents bacterial communities as sensitive indicators of contaminant stress. Sediments were collected from multiple sites within inner and outer zones of three heavily modified and three relatively unmodified estuaries. Bacterial communities were censused using Automated Ribosomal Intergenic Spacer Analysis and analysed for a suite of metal and PAH contaminants. Shifts in both bacterial community composition and diversity showed strong associations with sediment contaminant concentrations, particularly with metals. Importantly, these changes are discernable from environmental variation inherent to highly complex estuarine environments. Moreover, variation in bacterial communities within sites was limited. This allowed for differences between sites, zones and estuaries to be explained by variables of interest such as contaminants that vary between, but not within individual sites.  相似文献   

A sedimentary record of the rise and fall of the metal industry in Bergslagen,south central Sweden     

Dan Hammarlund  Anson W. Mackay  David M. J. Fallon  Gemma Pateman  Luisa C. Tavio  Melanie J. Leng  Neil L. Rose 《Journal of Paleolimnology》2008,39(4):463-475

Detailed stratigraphic analyses of sediments deposited in Lake Botjärnen, a small boreal forest lake in the shield terrain of central Sweden, clearly reflect progressively increasing human impact on terrestrial and aquatic ecosystems following settlement and establishment of an iron industry in the 17th century. Rising frequencies of pollen and spores from light-demanding plants provide evidence of extensive forest clearance for charcoal and timber production, which peaked in the early 20th century. An associated increase in catchment erosion is reflected by changing carbon and nitrogen elemental content and carbon–isotope composition of sediment organic matter and by increased magnetic susceptibility of the sediments. Records of air-borne pollutants (lead, zinc and sulphur) can be correlated to the development of local and regional mining and metal industry as inferred from historical accounts. Rapid recession of the iron industry led to re-forestation and recovery of the aquatic nutrient status to pre-industrial conditions over the past 100 years. The chronology of the sediment succession, which is based on ²¹⁰Pb and ¹³⁷Cs radionuclide data in combination with radiocarbon dating, is confirmed by historical lead pollution trends established for the region.  相似文献   

Organic matter in deepwater sediments of the Northern Gulf of Mexico and its relationship to the distribution of benthic organisms   总被引：1，自引：0，他引：1  

John W. Morse  Melanie J. Beazley 《Deep Sea Research Part II: Topical Studies in Oceanography》2008,55(24-26):2563

Sediment samples were selected from 28 sites across the deep (212–3527 m) northern Gulf of Mexico (GOM) as part of the Deep Gulf of Mexico Benthos (DGoMB) program, and analyzed for geochemical parameters related to organic carbon (OC) distribution and characteristics. The results of this study indicate that the OC content of sediments in the deep northern GOM is controlled by several factors; including water depth, overlying water productivity, sediment carbonate content, sediment oxygen exposure time, OC sources, and regional influences. The best correlation between sediment OC content, on a CaCO₃-free basis, and other parameters examined was an inverse correlation of OC with water depth. The OC/SA ratio had a wide range of values and, along with variable sources of sedimentary OC, indicated that the organism-available concentration of metabolizable organic matter may not be simply related to sedimentary OC content. This was perhaps reflected in the observation that benthic macrofaunal and meiofaunal biomass abundances were well correlated with sedimentary OC, but the abundance of bacteria in sediments was not.  相似文献   

Shell growth and oxygen isotopes in the topshell Osilinus turbinatus: resolving past inshore sea surface temperatures     

Marcello A. Mannino  Kenneth D. Thomas  Melanie J. Leng  Hilary J. Sloane 《Geo-Marine Letters》2008,28(5-6):309-325

Shells of the rocky shore intertidal gastropod Osilinus turbinatus (von Born), often abundant in archaeological deposits in the Mediterranean region, are a potential source of data on palaeotemperature, palaeoseasonality and archaeological seasonality. To evaluate this species as a climate archive, investigations of annual patterns of shell growth and of monthly variations in oxygen isotopes in shell carbonates were made on different populations in NW Sicily. Mark-recapture experiments at San Vito lo Capo and Mazzaforno show that O. turbinatus grows almost continuously throughout the year but at different rates in different seasons. Around 75% of the yearly shell growth occurs in the autumn and winter. On average, larger/older shells produce narrower annual growth increments than smaller/younger ones. Conspicuous growth lines in larger/older shells show that growth stops during the hottest part of the summer. Oxygen isotope analyses on monthly collected shells of O. turbinatus from three shores (Cala Grande, Monte Cofano and Mazzaforno) show that the isotope values record temperature variations through the year. In all the datasets, surface seawater temperatures (SSTs) calculated from δ¹⁸O_SHELL mostly underestimate measured SSTs, offsets being generally greater in summer. Minimum annual offsets range from 0.0°C to 0.7°C and maximum annual offsets from 3.1°C to 8.7°C. δ¹⁸O_SHELL values fail to record temperatures higher than 25°C. Careful selection of shells to be analysed can reduce offsets between δ¹⁸O_SHELL temperature estimates and measured SSTs for many parts of the year, except the hottest. Allowing for this, shells of O. turbinatus offer good potential as climate archives and for archaeological studies of seasonal patterns of human foraging for shellfish.  相似文献