Non-crossing nonlinear regression quantiles by monotone composite quantile regression neural network,with application to rainfall extremes

The goal of quantile regression is to estimate conditional quantiles for specified values of quantile probability using linear or nonlinear regression equations. These estimates are prone to “quantile crossing”, where regression predictions for different quantile probabilities do not increase as probability increases. In the context of the environmental sciences, this could, for example, lead to estimates of the magnitude of a 10-year return period rainstorm that exceed the 20-year storm, or similar nonphysical results. This problem, as well as the potential for overfitting, is exacerbated for small to moderate sample sizes and for nonlinear quantile regression models. As a remedy, this study introduces a novel nonlinear quantile regression model, the monotone composite quantile regression neural network (MCQRNN), that (1) simultaneously estimates multiple non-crossing, nonlinear conditional quantile functions; (2) allows for optional monotonicity, positivity/non-negativity, and generalized additive model constraints; and (3) can be adapted to estimate standard least-squares regression and non-crossing expectile regression functions. First, the MCQRNN model is evaluated on synthetic data from multiple functions and error distributions using Monte Carlo simulations. MCQRNN outperforms the benchmark models, especially for non-normal error distributions. Next, the MCQRNN model is applied to real-world climate data by estimating rainfall Intensity–Duration–Frequency (IDF) curves at locations in Canada. IDF curves summarize the relationship between the intensity and occurrence frequency of extreme rainfall over storm durations ranging from minutes to a day. Because annual maximum rainfall intensity is a non-negative quantity that should increase monotonically as the occurrence frequency and storm duration decrease, monotonicity and non-negativity constraints are key constraints in IDF curve estimation. In comparison to standard QRNN models, the ability of the MCQRNN model to incorporate these constraints, in addition to non-crossing, leads to more robust and realistic estimates of extreme rainfall.     

An algorithm to calculate the gravity anomaly of sedimentary basins with exponential density-depth relationships

We derive wavenumber domain expressions to calculate the gravity anomaly of a body with irregular bounding surfaces and an exponential density‐depth relationship. We apply the method to sedimentary basins, which commonly have this type of geometry and density distribution. The mathematical formulation also allows the exponential density‐depth relationship to be measured from an arbitrary irregular surface rather than the top surface. Using this arrangement, the gravity anomaly of exhumed sedimentary basins can be predicted if the amount of eroded section can be estimated. The corresponding inverse algorithms are also derived. Examples of the use of the forward algorithms, from the Galicia Interior Basin and the Central Irish Sea Basin, are used to illustrate these methods.     

The Gove relict iron meteorite from Arnhem Land,Northern Territory,Australia

On February 24, 1979, a deeply oxidized mass of iron meteorite was excavated from bauxite at an open cut mine on the Gove Peninsula, Northern Territory, Australia. The meteorite, measuring 0.75–1 m in diameter and of unknown total weight, was found at coordinates 12°15.8′S, 136°50.3′E. On removal from the ground, the meteorite is reported to have disintegrated rapidly. A preliminary analysis at the mine laboratory reportedly gave 8.5 wt% Ni. A modern analysis of oxidized material gave Ni = 32.9, Co = 3.67 (both mg g^?1), Cr = 168, Cu = 195, Ga = 22.5, Ge = <70, As = 4.16, W = 1.35, Ir = 10.5, Pt = 21.2, Au = 0.672 (all μg g^?1), Sb = <150, and Re = 844 (both ng g^?1). Competent fragments of oxidized material retain a fine to medium Widmanstätten pattern with an apparent average bandwidth of 0.5 mm (range 0.2–0.9 mm in plane section). Primary mineralogy includes rare γ–taenite and daubréelite, and secondary minerals produced by weathering include awaruite (with up to 78.5 wt% Ni) and an, as yet, unnamed Cu‐Cr‐bearing sulfide with the ideal formula CuCrS₂ that is hitherto unknown in nature. Deep weathering has masked many of the features of the meteorite; however, the analysis normalized to the analyses of fresh iron meteorites favors chemical group IIIAB. The terrestrial age of the meteorite is unknown, although it is likely to be in the Neogene (2.5–23 Ma), which is widely accepted as the major period of bauxite formation in the Northern Territory of Australia. Gove is the second authenticated relict meteorite found in Australia.     

Evolution of an englacial volcanic ridge: Pillow Ridge tindar, Mount Edziza volcanic complex, NCVP, British Columbia, Canada

Glaciovolcanic deposits are critical for documenting the presence and thickness of terrestrial ice-sheets, and for testing hypotheses about inferred terrestrial ice volumes based on the marine record. Deposits formed by the coincidence of volcanism and ice at the Mount Edziza volcanic complex (MEVC) in northern British Columbia, Canada, preserve an important record for documenting local and possibly regional ice dynamics. Pillow Ridge, located at the northwestern end of the MEVC, formed by ice-confined, fissure-fed eruptions. It comprises predominantly pillow lavas and volcanic breccias of alkaline basalt composition, with subordinate finer-grained volcaniclastic deposits and dykes. The ridge is presently  4 km long,  1000 m in maximum width, and  600 m high. Fifteen syn- and post-eruptive lithofacies are recognized in excellent exposures along the glacially dissected western side of the ridge. We recognize five lithofacies associations: (1) poorly sorted tuff breccia and dykes, (2) proximal pillow lava, dykes and tuff breccia, (3) distal pillow lava, poorly sorted conglomerate and well-sorted volcanic sandstone, (4) interbedded tuff, lapilli tuff, and tuff breccia units, and (5) heterolithic volcanogenic conglomerate and sandstone. Given the abundance of pillow lavas and the lack of surrounding topographic barriers capable of impounding water, we agree with Souther [Souther, J.G., 1992. The late Cenozoic Mount Edziza volcanic complex. Geol. Soc. Can. Mem., vol. 420. 320 pp] that the bulk of the edifice formed while confined by ice, but have found evidence for a more complex and variable eruption history than that which he proposed. Preliminary estimates of water-ice depths derived from FTIR analyses of H₂O give ranges of 300 to 680 m assuming 0 ppm CO₂, and 857 to 1297 m assuming 25 ppm CO₂. Variations in depth estimates among samples may indicate that water/ice depths changed during the evolution of the ridge, which is consistent with our interpretations for the origins of different lithofacies associations. Given that the age of the units are likely to be ca. 0.9 Ma [Souther, J.G., 1992. The late Cenozoic Mount Edziza volcanic complex. Geol. Soc. Can. Mem., vol. 420. 320 pp], Pillow Ridge may be the best documentation of a regional high stand of the Cordilleran Ice Sheet (CIS) in the middle Pleistocene, and an excellent example of the lithofacies and stratigraphic complexities produced by variations in water levels during a prolonged glaciovolcanic eruption.     

A preliminary investigation of boundary layer effects on daytime atmospheric CO<Subscript>2</Subscript> concentrations at a mountaintop location in the Rocky Mountains

Observations of CO₂ concentration at a mountaintop in the Colorado Rockies in summer show a large diurnal variability with minimum CO₂ concentrations found between 10:00 and 18:00 MST. Simulations are performed with a mesoscale model to examine the effects of atmospheric structure and large-scale flows on the diurnal variability. In the simulations initialized without large-scale winds, the CO₂ minimum occurs earlier compared to the observations. Upslope flows play an important role in the presence of this early (pre-noon) minimum while the timing and magnitude of the minimum depend only weakly on the temperature structure. An increase in large-scale flow has a noticeable impact on the diurnal variability with a more gradual decrease in daytime CO₂ concentration, similar to summer-averaged observations. From the idealized simulations and a case study, it is concluded that multi-scale flows and their interactions have a large influence on the observed diurnal variability.     

Temporal Interpolation of Spatially Dynamic Polygons

In this study we compare two methods of interpolating simple metrics from polygons that change over time. Conventional methodology involves computing values of the metrics at points in time where the polygon is observed, and interpolating between these observed values to estimate values at other points in time. An alternative method applies interpolation techniques to the locations of vertices composing polygons are interpolated between observations and metrics are computed from the interpolated polygon shapes. We conducted a Monte Carlo simulation of both techniques using simulated polygons composed of varying numbers of vertices, which were allowed to move randomly over varying numbers of discrete time periods. We observed differences in the success of the two methods when applied to three metrics: Area, Perimeter, and Area/Perimeter Ratio. We found the two techniques to perform differently depending upon the metric being interpolated. Full results are presented and possible reasons why the techniques succeeded or failed under various conditions are discussed.     

Nonlinear pushover analysis of infilled concrete frames

Six reinforced concrete frames with or without masonry infills were constructed and tested under horizontal cyclic loads. All six frames had identical details in which the transverse reinforcement in columns was provided by rectangular hoops that did not meet current ACI specifications for ductile frames. For comparison purposes, the columns in three of these frames were jacketed by carbon-fiber-reinforced-polymer （CFRP） sheets to avoid possible shear failure. A nonlinear pushover analysis, in which the force-deformation relationships of individual elements were developed based on ACI 318, FEMA 356, and Chen＇s model, was carried out for these frames and compared to test results. Both the failure mechanisms and impact of infills on the behaviors of these frames were examined in the study. Conclusions from the present analysis provide structural engineers with valuable information for evaluation and design of infilled concrete frame building structures.     

Signature of explosive volcanic eruptions in the sediments of a high-altitude Swiss lake

The analysis of 125 years of well-dated varved sediments in Lake Silvaplana, located at 1,791 m a.s.l. in the Upper Engadine region of south-eastern Switzerland, reveals that 7 out of the 8 climatically relevant explosive volcanic eruptions between A.D. 1880 and 2004 were followed by distinct peaks in median grain-size. Although the underlying mechanisms are yet unclear, an analysis of local meteorological data suggests that this phenomenon is unlikely to be related to any change in air temperature associated with the eruptions, but instead may be related to an increase in autumn precipitation subsequent to the eruptions that led to the erosion and fluvial transport of particles larger than normal.     

Groundwater Availability as Constrained by Hydrogeology and Environmental Flows

Groundwater pumping from aquifers in hydraulic connection with nearby streams has the potential to cause adverse impacts by decreasing flows to levels below those necessary to maintain aquatic ecosystems. The recent passage of the Great Lakes‐St. Lawrence River Basin Water Resources Compact has brought attention to this issue in the Great Lakes region. In particular, the legislation requires the Great Lakes states to enact measures for limiting water withdrawals that can cause adverse ecosystem impacts. This study explores how both hydrogeologic and environmental flow limitations may constrain groundwater availability in the Great Lakes Basin. A methodology for calculating maximum allowable pumping rates is presented. Groundwater availability across the basin may be constrained by a combination of hydrogeologic yield and environmental flow limitations varying over both local and regional scales. The results are sensitive to factors such as pumping time, regional and local hydrogeology, streambed conductance, and streamflow depletion limits. Understanding how these restrictions constrain groundwater usage and which hydrogeologic characteristics and spatial variables have the most influence on potential streamflow depletions has important water resources policy and management implications.