Space weather phenomena occur from the Sun to the Earth with damaging impacts on ground-based and space-borne technological infrastructure. The geomagnetic auroral electrojet indices, AU, AL, and AE, have been widely used for monitoring space weather and geomagnetic activities during space storms and substorms. The time series data of solar wind monitored by upstream satellite and ground-based auroral electrojet indices form the input-output system characterizing the dynamic coupling among solar wind, Earth’s magnetosphere, and ionosphere. The data-driven predictions of auroral electrojet indices during geomagnetic storms and substorms face the challenges of capturing the variations of ionospheric electrojet current driven by multiple solar wind variables and are modeled as a coupled complex system with finite and variable memory. The recurrent neural network (RNN) based Long Short-Term Memory (LSTM) machine learning algorithm is well suited to classify, process, and make predictions of the coupled solar wind-magnetosphere-ionosphere system by preserving important information from earlier parts of the coupled time series and carrying it forward. In this study, an RNN-based LSTM model has been built to predict the time series of AE/AL indices with multi-variate solar wind inputs. Both 5-minute and hourly long-term time series data from the NASA OMNI database were used to drive the LSTM model. The coupled time series data are divided into training and testing datasets. The Root-Mean-Square-Error (RMSE) between the predicted and actual AE/AL indices of the testing sets was used to evaluate the roles of the number of layers in the LSTM, memory length of the coupled system, prediction time, and different combinations of solar wind input parameters (magnetic field, velocity, and density). The performance of the LSTM model in predicting AL/AE indices during major geomagnetic storm and substorm events is analyzed. The differences and challenges of applying LSTM to predict 5-min and hourly AE/AL indices are also discussed.

Understanding the nature and behavior of the rocks in boundary zones between tectonic plates is important to improve our understanding of earthquake-associated hazard. Laboratory experiments can derive models that explain material behavior on small scales and under controlled conditions. These models can also be tested on observations of surface motion near plate boundaries: Fitting surface displacements from earthquakes (either shortterm offsets or longterm motion) yields estimates of rock properties for each model. However, using only observations from a single earthquake (from immediately after the quake and/or the subsequent years), may not allows us to confidently distinguish between models. In this study, we investigate the potential of using the displacement timeseries from multiple earthquakes, as well as the period between the quakes, to distinguish between proposed models. We use methods that enable comparison between models and parameters taking into account uncertainties, and perform our assessment on an artificial dataset.

The number of orbiting satellites has increased significantly and in an unrestricted and unregulated manner in recent years. This trend is expected to continue with ongoing plans from the commercial space sector to build mega-constellations of microsatellites. However, the effect of satellite demise during reentry on Earth’s atmosphere has only been lightly studied, and the long-term impact remains unknown.  This poster presents, to our knowledge, the first Reactive Force Field (ReaxFF) Molecular Dynamics (MD) simulation study on the atmospheric chemical mechanisms and byproducts generated by satellite reentry. Simulations are carried out to resolve chemical reactions and byproducts for Aluminum – a typical satellite structure constituent – under reentry conditions. MD simulation results are used to estimate the presence of oxides and predict the accumulated increase of reentry byproducts in the mesosphere when compared with that from meteoroids entering atmosphere and other natural sources. A methodology to estimate the residence time of these substances in the atmosphere is also presented so as to evaluate the polluting potential of mega-constellations reentry events.

Identifying areas of the land surface and surface water features (e.g., rivers, lakes, etc.) likely to contribute groundwater recharge to a public groundwater supply well is typically a first step toward source water protection. Identifying the contributions from these areas is important for assessing contamination sources, developing land use management strategies, and mitigating groundwater risk for drinking water supply. Simple analytical solutions that employ Darcy’s Law are unable to account for surface water boundary conditions within the flow system. Therefore, capture zone delineation is typically performed using three-dimensional, fully distributed numerical models that require considerable numbers of parameters, stratigraphic data, and user expertise. However, advanced analytical solutions exist that can provide approximations to such solutions using few parameters.In this work, the R Shiny web platform is developed to create an open-source application to allow Internet users to visualize potential flow systems near wells in the vicinity of surface water features. Assumptions include homogeneous stratigraphy and aquifer thickness, a steady state flow field, and relatively simple aquifer geometry. The web tool is currently being developed with a focus on Yukon Territory in northern Canada, where most of the population relies on groundwater, but less work has been done on the analysis of well vulnerability and source water protection than in southern Canada. The results are intended for estimation and education purposes and will be compared with numerical model results for some sites with pre-existing investigations. Abstract ID: 1168805

Imperfect models are often used for forecasting and state estimation of complex dynamical systems, typically by mapping a reference initial state into model phase space, making a forecast, and then mapping back to the reference space. In many cases these mappings are implicit, and forecast errors thus reflect a combination of model forecast errors and mapping errors. Techniques to infer parameterizations and parameters to reduce model bias have been the subject of intense scrutiny; however, we lack a general framework for discovering optimal mappings between system and model attractors. Here we propose a novel Machine Learning paradigm for inferring cross-attractor transformations (CATs) that minimize forecast error. CATs are pairs of transformations from the phase space of a reference system to the phase space of a model and vice versa that serve as a bridge between the attractors of a true system and an imperfect model. A computationally efficient analog approximation to tangent linear and adjoint models is developed to enable efficient stochastic gradient descent algorithms to train CAT parameters. Neural networks constructed with a custom analog-adjoint layer permit specification of affine transformations as well as more general nonlinear transformations.

Although pressure plays a vital role in the dynamics of the turbulent plasmas, pressure statistics have not been studied as extensively as other plasma properties. The studies that have focused on pressure are mostly based on hydrodynamic turbulence or have been formulated in the nearly incompressible magnetohydrodynamics (NIMHD) framework. However, less attention has been paid to the scaling properties of pressure in kinetic plasma. In this study, we explore the statistics of magnetic, thermal, and total pressure fluctuation in kinetic collisionless turbulence. A 2.5D kinetic particle-in-cell (PIC) simulation of turbulence is used to investigate pressure balance via the evolution of thermal and magnetic pressure. Further, the behavior of thermal, magnetic, and total pressure structure function and their corresponding spectrum is explored. Finally, we evaluate higher-order total pressure structure functions to discuss intermittency and compare the power exponents with higher-order structure functions of velocity and magnetic fluctuations.

The low viscosity of mafic magmas generally promotes effusive to mildly explosive eruptions; however, highly explosive mafic volcanic activity does occur. Recent models suggest that rapid ascent rates leading to significant disequilibrium drastically impact the rheology leading to fragmentation of the mafic magma. However, the magmatic conditions preceding magma ascent initiation are primarily unknown. There is a rich rock and mineral record for understanding melt generation in the mantle wedge and open system processes stored in the juvenile products of these highly explosive mafic eruptions. This study uses the pyroclasts from the Curacautin ignimbrite, a large volume (3.5–4.5 km3 DRE), mafic (50.4-57.3 wt% SiO2) ignimbrite at Llaima volcano in Chile.Of the 68 ignimbrite samples collected and previously analyzed, we targeted four for small volume (5 mg vs. standard 50 mg) characterization via ICPMS: (1) the top and (2) base of the ignimbrite, and the most (3) depleted and (4) enriched samples. We prepped and analyzed ten randomly selected pyroclasts from the highlighted samples (a total of 40 analyses). This study aimed to determine how sampling (e.g., composite vs. individual and the number of samples) impacts the signal of geological processes in trace element compositions. We used the χ2 distribution to test if the observed standard deviation (corrected for noise by subtracting analytical error) of all 40 samples, as well as each sample set’s ten analyzed pyroclasts, is larger than the analytical error. We identified true variability for at least 13 of the 30 trace elements measured for the base of the ignimbrite and enriched end member sample sets and the combined 40 analyses. For the top of the ignimbrite sample and the depleted end member sample, there is true variability in almost all 30 trace elements. Small volume whole rock geochemical characterization of individual pyroclasts detects previously unseen small-scale geochemical variability in compatible and REE elements and constrains new enrichment end-member compositions for geochemical models. Using principal component analysis, we investigate melt source heterogeneity and fractional crystallization as controls on the observed variability.

Rapid urbanization has caused extensive land use changes, which may impact the local climatology and trigger extreme rainfall events. Wind, another important factor in rainfall, has been examined to determine its role in generating extreme rainfall in conjunction with land use changes. This study evaluates the variability of extreme rainfall in the Dallas Fort-Worth (DFW) area. The study area has been divided into 256 equal sized grids and grouped into 4 clusters. Spatial Synoptic Classification (SSC) technique has been adopted to select days with benign synoptic activity so that the role of large-scale weather patterns is reduced. Extreme rainfall days are defined as those exceeding the 95th percentile threshold of the daily rainfall observed at the DFW airport. After filtering data based on SSC and extreme rainfall criteria, 114 days are selected for the study period from 2000-2016. The Multi-Radar Multi-Sensor rainfall product has been used to examine the variability of extreme rainfall in grids and clusters. The variation in extreme rainfall for different land covers (e.g., high, medium, and low intensity developed land, and open space) has also been examined. The results reveal that the “Urban Core” cluster receives maximum rainfall across the study period. Also, the difference in cumulative maximum rainfall between highly urbanized grids and the lowest urbanized grid is highest in the “Urban Core” cluster. The grid with the highest percentage of “High intensity developed land” received the most rainfall. The wind regime associated with the most rainfall across the study period is from the South-West. Less urbanized grids located downwind of the highest urbanized grids received relatively more rainfall. Urbanization and the magnitude of extreme rainfall is proportional in most of the grids. The role of the Urban Heat Island will also be examined to determine its impact on extreme rainfall variability.  

Mapping global vertical vegetation structure (VS) is critical for the quantification of global carbon stocks. While orbital LIDAR measurements of NASA’s Global Ecosystem Dynamics Investigation (GEDI) mission provide direct estimates of vertical VS, they only cover 4% of the global land surface. Recent works have produced global contiguous maps of canopy height using convolutional neural network (CNN) models and satellite imagery. However, these models faced some challenges estimating high canopy heights (>30 m) and identified some tiling artifacts (Lang et al., 2022; Potapov et al., 2020).We present various methods to address these limitations. First, we remove tiling artifacts by using overlapping tiles when producing maps and removing zero-padding from CNN architectures. Second, we compare the benefits and limitations of a few different methods to improve high canopy height estimation. Among the methods is histogram matching predicted heights to nearby LIDAR measurements. Another is learning a calibration model to correct each pixel based on similar known measurements nearby. Finally, we borrow recent advances in deep depth completion from the autonomous driving field to create an integrated model that uses known values to improve the prediction map. To demonstrate these methods, we map global vertical VS with a CNN model at 1km resolution using observations from Landsat, L-band synthetic aperture radar observations from the Advanced Land Observing Satellite (ALOS) PALSAR-2, and surface topography. For model training, GEDI relative height metrics are filtered and aggregated into 1km grids. We also use measurements from the Ice, Cloud and land Elevation Satellite 2 (ICESat-2), inter-calibrated with GEDI using co-located measurements. Finally, we apply the aforementioned corrective methods to the product, reporting global RMSE and MAE metrics, as well as visual qualitative observations. These results open the path for unsaturated global vertical VS products at higher 500 m, 200 m, and 100 m resolutions.

Identification of global methane (CH4) sources is critical to the quantification and mitigation of this greenhouse gas. Future imaging spectrometer missions, such as Carbon Mapper, will provide global, spatially resolved observations that will make it possible to accurately map methane sources. However, the sheer data volume of these missions make manual source identification infeasible, and expected artifacts in matched filter methane plume identification challenge simple thresholding. Recent works have demonstrated the feasibility of Convolutional Neural Networks (CNNs) for plume detection; however, in the past, these models have suffered from high false positive rates and were limited in their training and evaluation to individual flight campaigns.We have assembled quality-controlled tiled datasets from three Airborne Visible/Infrared Imaging Spectrometer Next Generation (AVIRIS-NG) campaigns: a 2020 California campaign (“COVID”), a 2019 Texas Permian Basin campaign (“Permian”), and another 2018 California campaign (“CACH4”). These datasets are notable for their diversity of surface conditions, spatial resolutions, and source types (oil & gas, energy, waste, livestock). Labeled methane sources in these datasets have been manually verified, and flightlines with systematic artifacts have been filtered out. We trained a GoogLeNet CNN classifier model on each of these campaigns to evaluate intra- and inter- campaign performance. We also trained a model on all three campaigns and evaluated its performance on each dataset. We observed an F1 performance of 0.7 or greater for each model trained and evaluated on its own dataset. We also observed that the model trained on all three datasets often outperforms individual models on multiple metrics. Finally, we converted the model into a fully convolutional network (FCN) for methane plume saliency map generation. We plan to extend this work to datasets acquired by the Global Airborne Observatory (GAO) and prepare a model for deployment for the Carbon Mapper orbital data product pipeline.

Presenting Author: [email protected] roll is the most common coherent noise observed in the land seismic data, mainly characterized by low frequency and low velocities. In this study, we are demonstrating the application of a model-based surface wave attenuation technique for the efficient removal of ground roll from 2D seismic data. The method employs inversion and adaptive subtraction filtering of surface waves. A 1D viscoelastic model is characterized by considering layer thickness, P & Sv wave velocities, density, and P & Sv wave quality factors (Qp & Qs) for each layer, to compute a synthetic seismogram. A linear Radon transform is used to generate dispersion spectra of input shot gather and synthetic shot gather generated from the model. The misfit between the dispersion spectra of input shot gather and synthetic shot gather is given by an objective function ‘J’,\(J\left(m\right)=\ {|\left|s-o\right||}_{2}\), where ‘s’ is the dispersion spectrum of the synthetic data generated by the model, ‘o’ is the dispersion spectrum of the shot gather and \({||.||}_{2}\)is the L2-norm of the difference between ‘s’ and ‘o’. The main focus is to minimize the objective function J. The local minima always seems to be the bottleneck for most of the non-linear inversion problems. To overcome this constraint, we implemented the Genetic Algorithm (GA) method in the surface-wave inversion process using the tool available in a commercial software. GA is a derivative-free search approach towards global minima, for the solution of the inversion problem. Once an optimized model is achieved from GA, synthetic data is generated and is adaptively subtracted from the shot gather. As the 1D model obtained cannot represent any lateral variations of physical parameters in the earth, adaptive subtraction and GA are implemented several times to mitigate this limitation.The proposed technique is applied to process a 2D shot gather data (Fig. 1-a) accessed from the SEG wiki Open data set collection of oz Yilmaz-40 shots. In this data, ground roll is masking the reflection events in the shot gather. After implementation of the proposed technique, it is observed that the ground roll is efficiently attenuated and reflections became more prominent by restoring the actual amplitudes, as shown in Fig. (1-b). The dispersion spectrum of both shot gathers is shown in Fig. 1 (c-d), it is observed that the ground roll present in the low-frequency zone (0-15 Hz) is attenuated. The results of the study suggest that the proposed method is quite efficient for surface wave attenuation.

Animal behaviour such as dispersal and migration ensure their survival in the landscape. It has been established in the past few decades that wildlife conservation and study of their movement in the wilderness is vital for sustainable ecosystem. Thus, identification of regions having high movement permeability for planning and maintenance of functional wildlife corridors has turn out to be a fundamental requirement for habitat management. This study emphases on movement of big cats-Bengal Tiger (Panthera tigris tigris) and Leopard (Panthera pardus fusca) in the protected area of Rajaji National Park situated in Uttarakhand State of India. The National park is a designated tiger reserve with large amount of tigers and leopards at its disposal. Here, Circuitscape was used to generate connectivity map of the study area. The results were validated using occurrence points downloaded from GBIF. The habitat suitability and resistance of the landscape was estimated based on literature review and expert opinion survey. Since, both the species have comparable ecological niche, similar habitat parameters were used for generation of resistance map of the species. Occurrence points for the species were downloaded from GBIF. 60% of the points were used as nodes or focal points where species presence is recorded whereas 40% of the points were used in validation of the connectivity paths. Results depicts the current density map of the study area highlighting areas with high connectivity for the species.

The interpretation of ERT data requires inversion, which is challenging due to its nonlinear and ill­posed nature. We aim to solve the inverse problem using a machine learning technique that approximates the complex non-linearity using the data provided without specifying any geophysical or mathematical equations. Specifically, we make use of CNN with U-Net architecture to estimate the true subsurface resistivity model from observed apparent resistivity data (pseudosection). U-Net is selected because it works well when both input and output data are in the form of images, and in the present work, input (pseudosection) and output (subsurface resistivity model) have spatial correspondence and existing local patterns. For ERT inversion, U-Net learns by minimizing pixel mean squared error between the true and predicted subsurface resistivity images. The ERT dataset was generated by simulating a dipole­dipole survey over a 400m x 100m earth containing two anomalous blocks with varying resistivity, size, and position in a variable homogeneous background resistivity. We used a finite element approach with a triangular mesh grid for forward modelling (using pyGIMLi) and added Gaussian noise to the calculated pseudosection so that simulated data resembles the field data. The U-Net model was trained on 4000 samples (20% of training samples used for validation) and tested on 1000 samples. The architecture consisted of an encoder and decoder subnetwork connected by a bridge section. The encoder depth was 7, and each block in the encoder subnetwork consisted of two sets of convolutional and ReLU layers, followed by a max pooling layer. The decoder subnetwork consisted of a transposed convolution layer for upsampling, followed by two sets of convolutional and ReLU layers. The weights of all the convolutional layers were initialized using the He Normal initializer. We used Adam optimizer with a learning rate of 0.0001. The U-Net model was allowed to train for 100 epochs with a batch size of 128. The trained U-Net model gave good predictions of the anomalous blocks and background resistivity with a root mean square misfit of 2.86% on test data and 2% on training data. The results confirm the efficacy and effectiveness of the proposed U-Net architecture in predicting the subsurface resistivity variation from the DC resistivity data.

A document by Balaji Senapati. Click on the document to view its contents.

Typhoon In-Fa hit continental China in July 2021 and caused an unprecedented rainfall amount, making it a typical case to examine the ability of numerical models in forecasting landfalling typhoons. The record-breaking storm was simulated using a 3-km-resolution weather research and forecast (WRF) model with spectral bin microphysics scheme (BIN) and two-moment seven-class bulk parameterization scheme (BULK). The simulations were then separated into three different typhoon landfall periods (i.e., pre-landfall, landfall, and post-landfall). At present, the ability of WRF and other mesoscale models to accurately simulate the typhoon precipitation hydrometeors is still limited. To evaluate the performances of BIN and BULK schemes of WRF model in simulating the condensed water in Typhoon In-Fa, the observed microwave brightness temperature and radar reflectivity from the core observatory of Global Precipitation Mission (GPM) satellite are directly used for validation with the help of a satellite simulator. It is suggested that BIN scheme has better performance in estimating the spatial structure, overall amplitude, and precise location of the condensed water in typhoons before landfall. During typhoon landfall, the performance of BIN scheme in simulating the structure and location of the condensate is close to that of BULK scheme, but the condensate intensity prediction by BIN scheme is still better; BULK scheme performs even better than BIN scheme in the prediction of condensate structure and location after typhoon landfall. Both schemes seem to have poorer performances in simulating the spatial structure of precipitation hydrometeors during typhoon landfall than before/after typhoon landfall. Moreover, BIN scheme simulates more (less) realistic warm (cold) rain processes than BULK scheme, especially after typhoon landfall. BULK scheme simulates more cloud water and larger convective updraft than BIN scheme, and this is also reported in many model studies comparing BIN and BULK schemes.

A document by PERUMALA RAJU. Click on the document to view its contents.

A document by Andrea Corpolongo. Click on the document to view its contents.

A document by Rachel Lamb. Click on the document to view its contents.