There is a growing need to address water security issues in Canada in relation to global warming effects and water management issues. Stable water isotopes (𝛿2H and 𝛿18O) have become a popular tool to investigate regional and country level public water distribution systems to identify regional water resource issues and their underlying causes. In this study, we built the first national level maps of isotopic values in tap water across Canada using 576 tap water samples that were collected during the summer season. Observed isotopic values in tap water follow similar patterns as precipitation with highest isotopic values in the coastal and Great Lakes regions and a progressive decrease as continentality increases with lowest values found in the Rocky Mountains. We classified the tap water samples based on their supply sources including groundwater, river and lake. The isotopic values of tap water sourced from groundwater and river are well-predicted by those of local annual precipitation whereas those sourced from lakes are poorly predicted. To explain this difference, we constructed a series of water balance models to predict the 𝛿2H and 𝛿18O variability of surface water across Canada. We used a digital elevation model to accumulate the isotopic composition of local precipitation weighed by precipitation (total or effective) either monthly, seasonally or annually. These water balance models strongly improved the predictability of isotopic values of tap waters sourced from rivers and lakes (particularly the annual average and monthly weighted summer models, with or without accounting for evapotranspiration). Conversely, these models did not improve predictability of the isotopic values in tap water sourced from groundwater. We argue that isotope values of tap water sourced from surface water reflect the accumulation and evapotranspiration of precipitation in catchments whereas groundwater represents an annual average of the isotopic composition of local precipitation. Regionally, we find that winter precipitation and snow melt and glaciers melt from the Rockies contribute to a large part of tap water sources around the Rockies making them vulnerable to water balance changes with ongoing climate warming. In eastern Canada water management processes appear to create some significant evaporative water losses whereas in central Canada dry continental climate causes significant evaporative water loss. Our study highlights the cycling of water from precipitation to tap water across Canada for different tap water supply with implications for water managements, contamination risks and vulnerability to climate change. Our models and databases also provide an isotopic baseline for regionally important water resources monitoring as well as for human forensic cases.

COVID-19’s impact on society and our daily habits has been unprecedented. With a decrease in vehicular traffic and industrial production, a decrease in local emissions was expected to occur. In order to capture any trends in ambient trace gas concentrations, approximately one thousand whole air samples were collected in intervals across the United States from April to July 2020 as part of the NASA Student Airborne Research Program (SARP). These samples were then analyzed by the UCI Rowland-Blake Lab using multi-column gas chromatography for over one hundred unique trace gases, including methane, non-methane hydrocarbons, and halocarbons, as described in Colman et al. (2001) and Barletta et al. (2002). Initial samples collected in April coincided with the peak of stay-at-home/social distancing orders in most states while samples collected later in the spring and early summer reflect the easing of these measures and initial state reopenings. Overall trends in emissions over time in select metropolitan areas will be discussed and compared to trends observed across the entire United States.

Majority of currently applied environmental models relay on linear relations between environmental endmembers. In this research, a detailed and comprehensive comparison between linear (L) and non-linear (NL) models are presented. The L and NL models are realized in a framework of Artificial Neural Networks (ANN). The evolution process of the ANN-L and ANN-NL models is based on estimation of fractional snow cover through data-fusion between high resolution (IKONOS) and medium resolution (Landsat TM/ETM+) remotely sensed images. The statistical measure values of R2, RMSE, MAE, Accuracy, Precision, Recall, and Specificity indicate better performance of the ANN-NL model in comparison to the ANN-L in estimation of ANN Landsat-FSC. The presented results, strongly indicate that to fully capture, untangle, and characterize internal environmental relations high-sensitivity non-linear models are required. Non-linear relations are particularly visible the complex in alpine-forested environments.

Launched in October 2018, the ESA/JAXA BepiColombo mission is currently in cruise to reach Mercury in late 2025. The Mercury Orbiter Radioscience Experiment (MORE) is one of the 16 instruments hosted on board the spacecraft. Testing general relativity is among the primary objectives of MORE. Superior conjunction experiments (SCE) will be performed during the interplanetary trajectory, with the aim of obtaining an accurate estimate of the post-Newtonian parameter γ. This is allowed by MORE advanced radio tracking system which provides precise range and Doppler data almost at all solar elongation angles, thus enabling an accurate measure of the relativistic time delay and frequency shift undergone by the signal when the spacecraft is in a superior solar conjunction (SSC). The rst BepiColombo SCE will take place in March 2021, and others will follow during the cruise phase. The nal objective is to place new limits to the accuracy of the general relativity as a theory of gravity in the weak eld limit, improving previous result from the Cassini SCE (Bertotti et al. 2003), which was able to determine that γ-1=(2.1±2.3)×10-5. Because of the proximity to the Sun, the spacecraft will undergo severe solar radiation pressure acceleration, and the effect of the random uctuations of the solar irradiance may become a major concern. We address the problem of a realistic estimate of the outcome of the SCE of BepiColombo, by including the effects of solar irradiance random variations in the dynamical model. We analyzed the experiment under different assumptions on the ranging system performances, observation coverage and solar activity showing their impact on the attainable result. We propose a numerical method to mitigate the impact of the variable solar radiation pressure on the scienti c result. Our simulations show that, exploiting data from multiple SSCs, the accuracy obtainable in the relativistic time delay measurement is 13×10-6 for a strong solar activity, and 6×10-6 for weak irradiance uctuations. We found that the latter result can be obtained by the rst SSC alone if the plasma noise calibration works until the impact parameter reaches 6 solar radii.

Over the past decade, parts of western Canada have seen a rise in clustered seismic activity coinciding with the growing use of a hydrocarbon reservoir stimulation technique known as hydraulic fracturing. This recent upsurge has the potential to increase the local seismic hazard, particularly in affected areas characterized by a sparser tectonic environment. It is therefore critically important to assess and characterize the space, time and magnitude distributions of induced earthquakes from a statistical standpoint, in order to develop a better understanding of triggering processes and improve forecasting models. In this study, the nearest-neighbour distance method was used to analyze the distribution of space-time inter-event distances across the Western Canada Sedimentary Basin from a regional perspective. Additionally, the epidemic type aftershock sequence model and the Gutenberg-Richter relation were used to compare the structuring and magnitude scaling of several seismic clusters induced by different human operations. The results demonstrate that a transformation in the regional distribution of inter-earthquake distances occurred after 2009, where an emergent subpopulation of abnormally tightly clustered events became distinguishable from both natural and prior-induced seismicity. Several distinctions were also revealed between earthquake clusters occurring near different anthropogenic operations, including a higher proportion of tightly clustered events near hydraulic fracturing treatments which were largely swarm-like in nature.

Evapotranspiration from land surface and vegetation plays an important role as a source of precipitation at continental scale in South America, and more so over the Andes cordillera. We evaluate the local recycling of moisture in a complex orography region located in the Central Andes of Colombia (CAC; 4.5N to 7.5N, 78.6W to 71.4W), comprising the East, Central and West Andean ranges and the two inter-Andean valleys of the Magdalena and Cauca Rivers. To this end, we apply the offline atmospheric moisture tracking model, WAM-2layers (Water Accounting Model-2layers). The model input data comes from the ERA-Interim Reanalysis spanning the period 1980-2016 and at 0.125 x 0.125 degrees resolution. We estimate the spatial distribution of the evaporation recycling ratio εc (source moisture zones) and the precipitation recycling ratio ρc (sink moisture zones) at annual and seasonal timescales. According with our results, up to 63% of the average annual evapotranspiration returns as local precipitation in the CAC. Some specific zones of the Western Cordillera and the eastern hillside of the Central Cordillera are “hot spots” for local moisture recycling given the high values of both εc and ρc. At seasonal timescales, there is more activity of the identified source moisture zones during March-April-May (MAM) and June-July-August (JJA). The higher moisture recycling activity in those seasons interconnects two source regions of moisture: the eastern piedmont of the Eastern Cordillera with the eastern piedmont of the Central Cordillera. Both piedmonts also show high intensity rainfall rates. Finally, our work confirms the fundamental role of local recycled moisture to enhance the midnight and early morning peak of the diurnal cycle of precipitation in the CAC during JJA.

Rapid urban growth has led to large population densities in foreland basin regions, and therefore a rapid increase in the number of people exposed to hazard from earthquakes in the adjacent mountain ranges. It is well known that ground shaking is amplified in sedimentary basins. However, questions remain regarding the main controls on this effect. It is, therefore, crucial to identify the main controls on earthquake shaking in foreland basins as a step towards mitigating the earthquake risk posed to these regions. We model seismic-wave propagation (using SW4 finite difference code) from range-front thrust-faulting earthquakes in a foreland-basin setting. The basin geometry (depth and width) and source characteristics (fault dip and source-to-basin distance) were varied, and the resultant ground motion was calculated. We find that the source depth determines the amount of near-source ground shaking and the basin structure controls the propagation of this energy into the foreland basin. Of particular importance is the relative length scales of the basin depth and dominant seismic wavelength (controlled by the source characteristics), as this controls the amount of dispersion of surface-wave energy, and so the amplitude and duration of ground motion. The maximum ground motions occur when the basin depth matches the dominant wavelength set by the source. Basins that are shallow compared with the dominant wavelength result in low-amplitude and long-duration dispersed waveforms. However, the basin structure has a smaller effect on the ground shaking than the source depth and geometry, highlighting the need for understanding the depth distribution and dip angles of earthquakes when assessing earthquake hazard in foreland-basin settings.

Venus is comparable to the Earth in size and overall distance from the Sun, but is outside of a defined “habitable zone” in which water can exist in liquid form. The deep atmosphere of Venus provides a hostile environment for life. However, recent work suggests that the atmosphere at cloud altitudes, which includes regions with temperatures and pressures similar to those at the Earth’s surface, could provide possible locations for microscopic life. Since the transit method strongly favors observations of planets close to their host stars, Venus analogs may be common in exoplanet observations. With over 100 Earth-sized exoplanets observed to date, it is important to be able to identify the characteristics of Venus analogs. Thick layers of clouds, such as those which enshroud Venus, are the regions most likely to be observed on a Venus-like exoplanet. The cloud layers on Venus display a wide range of wave-related features including global scale and smaller scale gravity waves, Rossby and Kelvin waves, streak-like structures, irregular dark regions, and vortices, and show considerable variations with altitude. It is important to characterize these temporal and spatial variations to understand the dynamics of the cloud-level atmosphere on Venus or Venus-like exoplanets. Observed thermal phase curves of exoplanets may show longitudinal variations, which could indicate inhomogeneities in cloud cover, day-night differences, variations in composition, or the influence of atmospheric waves, which may shift the phase curve relative to the substellar point. To understand and interpret observations of Venus or Venus-like exoplanets at cloud altitudes, simulations have been performed using a Venus middle atmosphere general circulation model. In this investigation we simulate Rossby and Kelvin waves and modify forcing parameters, including the resolution and dissipation, and compare the results with observations from Venus probes, Venus Express and the Akatsuki mission.

Stream confluences are ubiquitous features in freshwater networks, have distinct hydrogeomorphic characteristics relative to upstream tributaries and downstream reaches, and serve as junctions of previously independent streams. Confluences may enhance or disrupt biological processes. How ecosystem functions (e.g., carbon metabolism, nutrient removal) change at confluences remains a knowledge gap in our understanding of the processes controlling water quality at the network-scale. To test how carbon and nutrient cycling may differ between confluences and their tributaries, we estimated dissolved organic carbon (DOC) and PO43- uptake in October 2018 and July 2019 in two tributary reaches as well as downstream of their confluence mixing zone using pulse injections of roasted barley leachate (a standardized, colored DOC source), K2HPO4, and NaCl (a non-bioreactive tracer). We hypothesized that biological processes would be enhanced at confluences due to the delivery and mixing of different microbial communities and/or carbon and nutrient sources. We calculated PO43- and DOC uptake velocities (vf-PO4, vf-DOC) and compared them across sites and season. In October 2018, vf-PO4 in each tributary was 10.2 and 4.9 mm/min while vf-DOC was 0.84 and 0.38 mm/min. vf-PO4 downstream (6.6 mm/min) was lower than vf-PO4 predicted from a mixing model of upstream vf-PO4 and proportional flow contributions of tributaries (10.1 mm/min), suggesting in-stream PO43- uptake was suppressed as a result of confluence mixing. Conversely, vf-DOC downstream (0.94 mm/min) was higher than vf-DOC predicted from a mixing model (0.75 mm/min). This difference in measured and predicted vf-DOC was supported by bioassay experiments, which found enhanced DOC uptake downstream of the mixing zone. DOC uptake within the confluence mixing zone was spatially heterogeneous (0.00 to 0.19 day-1) and varied more within mixing zone transects than among the two tributary reaches. Ongoing analyses are comparing uptake estimates among seasons. Our results suggest that DOC and PO43- uptake at confluences cannot be estimated from tributary DOC and PO43- uptake alone. A critical next step in this work is to identify the mechanisms behind confluence-derived changes in carbon metabolism and nutrient removal across freshwater networks.

Muography is an imaging technique using the directional flux of atmospheric muons to indirectly study a suitable object of geophysical or strategic interest. In this work we present the current status of some projects that are being carried out by our groups looking for potential muography applications in Argentina. These developments range from the application of this technique on large structures, such as volcanoes or dams; exploration of minerals in shallow mines; and for the study of the muon flux inside underground laboratories, such as the incoming ANDES underground laboratory. Development and implementation of novel muon detection techniques are required to achieve the spatial and temporal resolution needed for some of these projects, and new prototypes are being built in Buenos Aires, Argentina, including, e.g., sub pixel spatial resolution on cheaper, lighter and faster detectors. In this work we present the current status of our prototypes and an analysis of possible application targets at the argentinian volcanoes Copahue and Peteroa and in the ANDES underground laboratory. These results are based on logistic and technical criteria, such as the calculation of the distance traveled by the muons depending on the site and the trajectories of the muons through layers of different types of rocks.

Electric field measurements inside thunderstorms are essential to our understanding of thunderstorm charge structure, electrification, and lightning initiation. However, most existing measurements have been made by single instruments carried aloft by weather balloon, thus providing measurements made at a single point that moves through the storm on a timescale of tens of minutes. It is therefore difficult to interpret such data, since a change in observed field strength may be due to motion of the balloon into a region with different field or due to overall evolution of the storm's electrical structure with time. Separation of such spatial and temporal variability requires simultaneous measurements at multiple locations within the storm. This can be accomplished with a single weather balloon by carrying multiple independent electric field dropsondes aloft and releasing them one at a time, separated by short time intervals. The balloon payload design is optimized for low mass and use of off-the-shelf components whenever possible, releasing each dropsonde by a hot wire cut-down mechanism. Each dropsonde spins as it falls, measuring electric field as it rotates and sends data to a ground station in real-time. The dropsondes are designed to fall and rotate stably by use of aerodynamic simulations, with internal components robustly connected along the axis of the instrument to ensure the desired balance and alignment of the principal axes of the moment of inertia. The telemetry transmitters use simple low-cost low-power all-in-one transmitter chips. The telemetry ground station receives signals simultaneously from all dropsondes by a single software-defined radio receiver. Robust long-range communication is enabled by use of spread spectrum techniques and error correcting codes.

We examine long-term energetic heavy ion measurements including three planets’ magnetospheres, focusing on Fe ions (specifically, but not exclusively, Fe+) in and near Earth’s magnetosphere. We compare Fe data to that of other energetic ion species with masses greater than C (carbon) and consider the relationship(s) of energetic Fe ion measurements at the three planets to internal (ionospheres, exospheres, moons, rings, and trapped radiation) and external (solar wind and interplanetary dust) source candidates. Fe+ has been observed at Earth and Saturn, but not yet at Jupiter, as our observations there were brief. The measurements are from two functionally identical charge-energy-mass ion spectrometers: one on Geotail (~87-212 keV/e), orbiting Earth at ~9-30 Re; and the other on Cassini (~83-167 keV/e), in interplanetary space, during Jupiter flyby, and at ~4-20 Rs on its constantly varying orbits around Saturn. These ion spectrometers efficiently separate energetic light and heavy ions by mass, as well as lower charge state ions from higher charge state ions by mass-per-charge. Energetic low charge state ions often derive from magnetospheric sources, while energetic high charge state ions most often derive from the solar wind. We also enlist heavy ion measurements closer to the Earth from AMPTE/CCE which are used for C and Fe radiation-belt- modeling content, consideration, and estimation.

Recent observations and simulations have revealed a wide variety of plasma processes and multiscale structures at the dayside magnetopause. In this presentation, we focus on the origins and evolution of the plasma populations observed in the magnetopause boundary layers. We present the results of Particle-In-Cell (PIC) simulations encompassing large volumes of the dayside magnetosphere. The implicit PIC code used in the study was initialized from a global MHD state of the magnetosphere for southward interplanetary field conditions. Three-dimensional plots of the perpendicular slippage indicates that reconnection occurs over most of the dayside magnetopause. However, the simulation reveals that the reconnection region has a much more filamentary structure than the X-line expected from the extrapolation of 2D models and that multiscale structures thread the reconnection outflow. In particular, the simulation indicates the formation of multiple layers of electrons with significant field-aligned velocities within the main magnetopause current layer. We use velocity distribution functions at different locations in the reconnection outflow to characterize the origins and evolution of the electron and ion populations of the magnetosheath and magnetospheric boundary layers.

In-situ novelty-based target selection of scientifically interesting (“novel”) surface features can expedite follow-up observations and new discoveries for the Mars Science Laboratory (MSL) rover and other planetary exploration missions. This study aims to identify which methods perform best for detecting novel surface features in MSL Navcam images for follow-up observations with the ChemCam instrument, as a complement to the existing Autonomous Exploration for Gathering Increased Science (AEGIS) onboard targeting system. We created a dataset of 6630 candidate targets within Navcam grayscale images acquired between sols 1343-2578 using the Rockster algorithm. These were the same target candidates considered by AEGIS, chosen to enable direct comparison to past AEGIS target selections. We employed five novelty detection methods, namely Discovery via Eigenbasis Modeling of Uninteresting Data (DEMUD), Isolation Forest, Principal Component Analysis (PCA), Reed-Xiaoli (RX) detector, and Local RX. To evaluate the algorithm selections, a member of the MSL science operations team independently identified candidate targets that represented example scenarios of novel geology that we would desire an algorithm to identify, such as layered rocks, light-toned unusual textures, and small light-toned veins. We compared these methods to selections made by AEGIS and a random baseline. Initial experiments for three scenarios showed that Local RX most frequently prioritized novel targets, followed by DEMUD and AEGIS. Our next steps in this study include evaluating input feature representations other than pixel intensities (e.g., Histogram of Oriented Gradients features), performing additional experiments to evaluate novel target prioritization performance, and selecting target candidates in Mastcam color images.

For catchment-scale streamflow estimation, recently, the preference is shifting towards the use of innovative remote sensing (RS)-based approaches at the remote gauging stations. In this context, this study evaluates the performances of three RS-based models designed with the near-infrared (NIR) bands of Landsat images at the Jenapur outlet of the Brahmani River basin in eastern India. These RS-based models are designed using the spectral behavior of land (C) and water (M) pixels in the NIR region of the electromagnetic spectrum in the presence and absence of water surrounding the streamflow gauging station. Further, the computed pixel ratio (C/M) is used as a parameter for the discharge estimation, in which four years (2009-2013) of Landsat images are used during calibration and three years (2014-2016) of these images are used during validation. Model-I uses the C/M method in which a box-matrix is conceptualized to analyze the optimal location of the land pixel (C0); and subsequently, the time series of C/M is calibrated with the in-situ discharge (Q) time series. The best pixel ratio (C0/M) time series is preprocessed with an exponential smoothing filter to derive the best filtered-pixel ratio (C0/M*) time series, which is used in the regression model to estimate the river discharge. Model-II corresponds to the multi-pixel ratio (MPR) method, where a 3×3 window is used to calculate the average reflectance of both the C and M pixels within the box-matrix, and subsequently, to obtain the best pixel (C0ʹ/Mʹ) ratio as in the case of Model-I to develop the spectral relationship between C0ʹ/Mʹ and Q time series. Model-III uses both the C/M and water width-based function to estimate the streamflow. The performance evaluation of the models is carried out using the Nash-Sutcliffe efficiency, Percentage bias, and Mean absolute error, which reveals that the model performance varies in the order: Model-III > Model-II > Model-I. This proposed RS-based discharge estimation model framework has the potential to be used in many world- rivers with varying cross-sections.

Urban Green Spaces (UGSs) are proving to be most important part of urban area of a city. These green spaces not only provide psychological comfort to humans but also affect heat impact on city to a vast level. In a developing country like India, where urban growth is happening in a very fast and haphazard manner a little consideration is given to Green Spaces in city. Such a study has been conducted over Lucknow metro-city of Uttar Pradesh state of India. In this study, the relation between Land Use\Land Cover (LULC) and Land surface temperature (LST) has been tried to find. Using Landsat-8 OLI data (Band 3, 4 & 5) and Maximum Likelihood Classification algorithm, 6 Land Use classes are obtained, which are “Built-up”, “Vegetation”, “Shrub”, “Water”, “Fallow Land” and “Other”. In addition to these 4 Land Cover Indices namely Normalized Difference Built-up Index (NDBI), Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI) and Normalized Difference Barren Index (NDBaI) are also generated for same areas. LST is obtained using TIRS data (Band 10 & 11) and Split Window Algorithm by Radiative Transfer Equation. Ancillary data is used for digitization of 5 assembly constituencies (ACs) of Lucknow parliamentary constituency (PC) of Lucknow district namely, Lucknow Cantonment, Lucknow North, Lucknow East, Lucknow West and Lucknow Central has been done. River and Canal passing through these areas, are also considered in Urban Green Spaces as per Urban Atlas code 14100. 250m radius buffer is generated around Built-up pixels for analysis of impact of UGSs on LST. Cantonment and East Lucknow area having highest amount of UGSs in terms of “Vegetation”, “Shrub” & “Water” pixels due to presence of Forest area in both ACs. It is found that LST is positively related with all indices except for NDVI with strong negative correlation and R2 of 0.47 and highest R2 of 0.53 with NDBI. Among all 5 ACs best correlation between all 4 LC and LST values is found in Lucknow East AC with R2 > 0.64 for NDVI, NDWI and NDBI. Lucknow East AC is having least LST but there is very little difference between LST values of Built-up pixels having minimum UGSs present in 250m radius buffer around built-ups and Built-up pixels having no UGSs around. AC Lucknow Central is having 1 °C difference in LST values of such different Built-ups.

Multicollector noble gas mass spectrometry has yielded significant improvements in the precision of 40Ar/39Ar age determinations over the last decade [1,2]. We report initial results from the new Isotopx NGX mass spectometer at the WiscAr Laboratory that is equipped with 9 Faraday detectors and 1 ion counting multiplier. The NGX utilizes Isotopx’s ATONA® amplifier technology, enabling measurements spanning a dynamic range from below 10−16 A to above 10−9 A [3]. Moreover, the WiscAr NGX offers the unique opportunity to evaluate instrument performance using both a conventional, and a newer low temperature, Nier-type ion source, which allows for trap current variation. We have performed a series of tests to: (1) assess optimal measurement and integration times for blanks, baselines, and air aliquots of various sizes, (2) quantify the NGX sensitivity via measurements of first principle 40Ar/39Ar standards fused with a CO2 laser, and (3) compare the sensitivity and effects on instrument background between the conventional and new low temperature Nier-type sources. Initial results suggest that for most samples optimal precision is achieved when using 10 second integration times for measurements lasting 400-600 seconds. Measurements of low 36Ar signals are optimized with longer integration and measurement times. In addition to measurement optimization experiments, we performed a comparative analysis of both fusion and incremental heating data obtained using the WiscAr Nu Instruments Noblesse, MAP 215-50, and the NGX for a variety of geologic materials which range in age from Pleistocene to Permian. Pychron software [4] controls the entire NGX analytical system, including data collection and reduction. Careful experiment design in this automated system, when informed by optimized counting practices and knowledge of collector sensitivity, can lead to high precision 40Ar/39Ar dates. [1] Jicha et al. (2016), Chemical Geology 431, 54–66 [2] Mark et al. (2009), Geochemistry, Geophysics, Geosystems 10 [3] Cox et al. (2020), Geochron. Discuss, 2020-1. [4] Ross J., 2019, NMGRL/pychron v18.2, doi: 10.5281/zenodo.3237834.

Large igneous provinces (LIPs) are often associated with mass extinctions and are thus vital for life evolution on Earth. However, the precise relation between LIPs and their impacts on biodiversity is enigmatic as they can be asynchronous. If the environmental impacts are primarily related to sill emplacement, the structure of LIPs’ magma storage system becomes critical as it dictates the occurrence and timing of mass extinction. Here we use surface wave tomography to image the lithosphere under the Permian Emeishan Large Igneous Province (ELIP) in southwestern China. We find a NE-trending zone of high shear-wave velocity (Vs) and negative radial anisotropy (Vsv > Vsh) in the crust and lithosphere and interpret it as a mafic-ultramafic, dike-dominated magma storage system on the hidden hotspot track of the ELIP. An area of less-negative radial anisotropy, on the hotspot track but away from the eruption center, reflects an elevated proportion of sills emplaced at the incipient stage of the ELIP. Liberation of poisonous gases and mercury by the sills explains why the mid-Capitanian global biota crisis preceded the peak ELIP eruption by 2-3 million years.