Machine Learning and Remote sensing method to determine the relationship between Climate and Groundwater Recharge. Adya Aiswarya Dash1, Abhijit Mukherjee1,2,3. 1Department of Geology and Geophysics, Indian Institute of Technology Kharagpur, West Bengal 721302, India 2School of Environmental Science and Engineering, Indian Institute of Technology Kharagpur, West Bengal 721302, India 3Applied Policy Advisory for Hydrogeoscience (APAH) Group, Indian Institute of Technology Kharagpur, West Bengal 721302, India Abstract Through machine learning and remote sensing, a high-end model with a finer resolution for groundwater recharge has been developed for the region of South-East Asia. The groundwater recharge coefficient can be found by the application of Random Forest regression followed by the implication of the water budget method to calculate the Groundwater Recharge values. Climatic factors such as precipitation and actual evapotranspiration to map Groundwater Recharge has been framed with a sophisticated machine learning method to be considered as a scale predicting model. A comprehensive visualization of the dataset has been done; the accuracy of the model is noted through random forest regression. Thus, the model can be used for various regions of the dataset specifically for the area where there is a lack of reach for data. It can be successfully used to form a sophisticated end-to-end ML model. Keywords: Machine Learning, Remote Sensing, Groundwater Recharge, Climate science.

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.

Sustainably managing resources in a transboundary freshwater basin is a complex problem, particularly when considering the compounding impacts of climate change, hydropower development, and evolving water governance paradigms. In this study, we used a mixed methods approach to analyze potential impacts of climate change on regional hydrology, the ability of dam operation rules to keep downstream flow within acceptable limits, and the present state of water governance in Laos, Vietnam, and Cambodia. Our results suggest that future river flows in the 3S river system could move closer to natural (i.e., pre-development) conditions during the dry season and experience increased floods during the wet season. This anticipated new flow regime in the 3S region would require a shift in the current dam operations, from maintaining minimum flows to reducing flood hazards. Moreover, our Governance and Stakeholders survey assessment results revealed that existing water governance systems in Laos, Vietnam, and Cambodia are ill-prepared to address such anticipated future water resource management problems. Our results indicate that the solution space for addressing these complex issues in the 3S river basins will be highly constrained unless major deficiencies in transboundary water governance, strategic planning, financial capacity, information sharing, and law enforcement are remedied in the next decade. This work is part of an ongoing research partnership between the National Aeronautical and Space Agency (NASA) and the Conservation International (CI) dedicated to improving natural resources assessment for conservation and sustainable management.

Earth Observations (EO) systems aim to monitor nearly all aspects of the global Earth environment. Observations of Essential Water Variables (EWVs) together with advanced data assimilation models, could provide the basis for systems that deliver integrated information for operational and policy level decision making that supports the Water-Energy-Food-Nexus (EO4WEF), and concurrently the UN Sustainable Development Goals (SDGs), and UN Framework Convention on Climate Change (UNFCCC). Implementing integrated EO for GEO-WEF (EO4WEF) systems requires resolving key questions regarding the selection and standardization of priority variables, the specification of technologically feasible observational requirements, and a template for integrated data sets. This paper presents a concise summary of EWVs adapted from the GEO Global Water Sustainability (GEOGLOWS) Initiative and consolidated EO observational requirements derived from the GEO Water Strategy Report (WSR). The UN-SDGs implicitly incorporate several other Frameworks and Conventions such as The Sendai Framework for Disaster Risk Reduction; The Ramsar Convention on Wetlands; and the Aichi Convention on Biological Diversity. Primary and Supplemental EWVs that support WEF Nexus & UN-SDGs, and Climate Change are specified. The EO-based decision-making sectors considered include water resources; water quality; water stress and water use efficiency; urban water management; disaster resilience; food security, sustainable agriculture; clean & renewable energy; climate change adaptation & mitigation; biodiversity & ecosystem sustainability; weather and climate extremes (e.g., floods, droughts, and heat waves); transboundary WEF policy.

Low marine clouds are a major source of uncertainty in cloud feedbacks across climate models and in forcing by aerosol-cloud interactions. The evolution of these clouds and their response to aerosol are sensitive to the ambient environmental conditions, so it is important to be able to determine different responses over a representative set of conditions. Here, we propose a novel approach to encompassing the broad range of conditions present in low marine cloud regions, by building a library of observed environmental conditions. This approach can be used, for example, to more systematically test the fidelity of Large Eddy Simulations (LES) in representing these clouds. ERA5 reanalysis and various satellite observations are used to extract and derive macrophysical and microphysical cloud-controlling variables (CCVs) such as SST, estimated inversion strength (EIS), subsidence, and cloud droplet number concentrations. A few locations in the stratocumulus (Sc) deck region of the Northeast Pacific during summer are selected to fill out a phase space of CCVs. Thereafter, Principal Component Analysis (PCA) is applied to reduce the dimensionality and to select a reduced set of components that explain most of the variability among CCVs in order to efficiently select cases for LES simulations that encompass the observed CCV phase space. From this phase space, 75-100 cases with distinct environmental conditions will be selected and used to initialize 2-day LES modeling to provide a spectrum of aerosol-cloud interactions and Sc-to-Cumulus transition under observed ambient conditions. Such a large number of simulations will help create statistics to assess how well the LES can simulate the cloud lifecycle when constrained by the ‘best estimate’ of the environmental conditions, and how sensitive the modeled clouds are to changes in these driving fields.

Earth Observations (EO) systems aim to monitor nearly all aspects of the global Earth environment. Observations of Essential Water Variables (EWVs) together with advanced data assimilation models, could provide the basis for systems that deliver integrated information for operational and policy level decision making that supports the Water-Energy-Food-Nexus (EO4WEF), and concurrently the UN Sustainable Development Goals (SDGs), and UN Framework Convention on Climate Change (UNFCCC). Implementing integrated EO for GEO-WEF (EO4WEF) systems requires resolving key questions regarding the selection and standardization of priority variables, the specification of technologically feasible observational requirements, and a template for integrated data sets. This paper presents a concise summary of EWVs adapted from the GEO Global Water Sustainability (GEOGLOWS) Initiative and consolidated EO observational requirements derived from the GEO Water Strategy Report (WSR). The UN-SDGs implicitly incorporate several other Frameworks and Conventions such as The Sendai Framework for Disaster Risk Reduction; The Ramsar Convention on Wetlands; and the Aichi Convention on Biological Diversity. Primary and Supplemental EWVs that support WEF Nexus & UN-SDGs, and Climate Change are specified. The EO-based decision-making sectors considered include water resources; water quality; water stress and water use efficiency; urban water management; disaster resilience; food security, sustainable agriculture; clean & renewable energy; climate change adaptation & mitigation; biodiversity & ecosystem sustainability; weather and climate extremes (e.g., floods, droughts, and heat waves); transboundary WEF policy.

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.

Earth Observations (EO) systems aim to monitor nearly all aspects of the global Earth environment. Observations of Essential Water Variables (EWVs) together with advanced data assimilation models, could provide the basis for systems that deliver integrated information for operational and policy level decision making that supports the Water-Energy-Food-Nexus (EO4WEF), and concurrently the UN Sustainable Development Goals (SDGs), and UN Framework Convention on Climate Change (UNFCCC). Implementing integrated EO for GEO-WEF (EO4WEF) systems requires resolving key questions regarding the selection and standardization of priority variables, the specification of technologically feasible observational requirements, and a template for integrated data sets. This paper presents a concise summary of EWVs adapted from the GEO Global Water Sustainability (GEOGLOWS) Initiative and consolidated EO observational requirements derived from the GEO Water Strategy Report (WSR). The UN-SDGs implicitly incorporate several other Frameworks and Conventions such as The Sendai Framework for Disaster Risk Reduction; The Ramsar Convention on Wetlands; and the Aichi Convention on Biological Diversity. Primary and Supplemental EWVs that support WEF Nexus & UN-SDGs, and Climate Change are specified. The EO-based decision-making sectors considered include water resources; water quality; water stress and water use efficiency; urban water management; disaster resilience; food security, sustainable agriculture; clean & renewable energy; climate change adaptation & mitigation; biodiversity & ecosystem sustainability; weather and climate extremes (e.g., floods, droughts, and heat waves); transboundary WEF policy.

As all kinds of physics-based and data-driven models are emerging in the fields of hydrologic and hydraulic engineering, Bayesian model averaging (BMA) is one of the popular multi-model methods used to account for the various uncertainty sources in the flood modeling process and generate robust ensemble predictions based on multiple competitive candidate models. The reliability of BMA parameters (weights and variances) determines the accuracy of BMA predictions. However, the uncertainty in the BMA parameters with fixed values, which are usually obtained from the Expectation-Maximization (EM) algorithm, has not been adequately investigated in BMA-related applications over the past few decades. Given the limitations of the commonly used EM algorithm, the Metropolis-Hastings (M-H) algorithm, which is one of the most widely used algorithms in the Markov Chain Monte Carlo (MCMC) method, is proposed to estimate the BMA parameters and quantify their associated uncertainty. Both numerical experiments and the one-dimensional HEC-RAS models are employed to examine the applicability of the M-H algorithm with multiple independent Markov chains. The performances of the EM and M-H algorithms in the BMA analysis are compared based on the daily water stage predictions from 10 model configurations. The results show that the BMA weights estimated from both algorithms are comparable, while the BMA variances obtained from the M-H MCMC algorithm are closer to the given variances in the numerical experiment. Moreover, the normal proposal distribution used in the M-H algorithm can yield narrower distributions for the BMA weights than those from the uniform prior. Overall, the MCMC approach with multiple chains can provide more information associated with the uncertainty of BMA parameters and its prediction performance is better than the default EM algorithm in terms of multiple evaluation metrics as well as algorithm flexibility.

Evaluation of the performance of hydrologic and hydraulic models is a crucial step in the modeling process. Considering the limitations of single statistical metrics, such as the Nash Sutcliffe efficiency (NSE), the Kling Gupta efficiency (KGE), and the coefficient of determination (R2), which are widely used in the evaluation of model performance, an evaluation framework that incorporates multiple criteria and based on the generalized likelihood uncertainty estimation (GLUE) is proposed to demonstrate the uncertainty in the evaluation criteria and hence to quantify the overall uncertainty of flood models in a comprehensive way. This framework is applied to the one-dimensional HEC-RAS models of six reaches located in States of Indiana and Texas of the United States to quantify the uncertainty associated with the channel roughness and upstream flow input. Specifically, the effects of different prior distributions of the uncertainty sources, multiple high-flow scenarios, and various types of measurement errors (white noise, positive bias, and negative bias) in observations on the evaluation metrics are investigated by using the bootstrapping method and Monte Carlo simulations. The results show that the model performances based on the uniform and normal priors are comparable. The distributions of all the evaluation metrics in the framework are significantly different for the flood model under different high-flow scenarios, and it further indicates that the metrics are essentially random statistical variables. Additionally, the white-noise error in observations has the least impact on the metrics, while the positive and the negative biases would have opposite impacts, which depends on whether the model overestimated or underestimated the hydrologic variable.

Natural hazards such as floods, hurricanes, heatwaves, and wildfires cause significant economic losses (e.g., agricultural and property damage) as well as a high number of fatalities. Natural hazards are often driven by univariate or multivariate hydrometeorological drivers. Therefore, it is crucial to understand how and which hydrometeorological variables (i.e., drivers) combine to contribute to the impacts of these hazards. Additionally, when multiple drivers are associated with a hazard, traditional univariate risk assessment approaches are insufficient to cover the full spectrum of impact-relevant conditions originating from different combinations of multiple drivers. Based on historical socioeconomic loss data, we develop an impact-based approach to assess the influence of different hydrometeorological drivers on the impacts caused by different hazard event types. We use the Spatial Hazard Events and Losses Database for the United States (SHELDUS™) to identify the historical hazard events that caused socioeconomic impacts (property and crop damage, injuries, and fatalities) in our case study area, Miami-Dade County, in south Florida. For 9 different hazard types, we obtained data for 13 hydrometeorological drivers from historical in-situ observations and reanalysis products corresponding to the timing and locations of the hazard events found in the SHELDUS database. The relative importance of each hazard driver in generating impacts and the frequency of multiple drivers was then assessed. We found that many high-impact events were caused by multiple hydrometeorological drivers (i.e., compound events). For example, 61% of the recorded flooding events were compound events rather than univariate hazards and these contributed 99% of total property damage and 98.2% of total crop damage in Miami-Dade County. For several hazards, such as hurricanes/tropical storms and wildfires, all the events that caused damage are classified as compound events in our framework. Our findings emphasize the benefit of including socioeconomic impact information when analyzing hazard events, as well as the importance of analyzing all relevant hydrometeorological drivers to identify compound events.

The nucleation and triggering of basal microseisms, or icequakes, at the bottom of glaciers as the ice flows over it can grant us valuable insights about deformation processes that occur at the bed. The collaborative efforts of Penn State University and the British Antarctic Survey (BAS) during the 2018/2019 austral summer enabled the deployment of several seismic arrays over 3 months in the Rutford Ice Stream in West Antarctica for monitoring natural source seismicity. Using the earthquake detection and location software QuakeMigrate, we generated unique high-resolution icequake catalogs, particularly at Rutford’s grounding line. Our data showed an unprecedented number of detected events which we used to resolve key topographical features and characteristics at the bed like sticky spots, and how they related to the continuous ice loading-slipping process at the bed. To properly quantify relations between events, we performed rigorous testing via manual event inspection at each array to determine a trigger threshold that aims to balance event coverage with artefact minimization. To handle the massive amounts of incoming seismic data and subsequent located icequakes, we also created a systematic data processing pipeline, and used machine learning clustering algorithms to resolve inter- & intra-clusters spatial and temporal relations. We present our pre-processing methods on handling similarly large datasets and present findings from our seismic data in combination with other data sources, like GPR and tidal gauge data, that improves our understanding of ice flow dynamics in the region.

We present a cost-efficient tilt sensor that was originally developed by our team at Dartmouth College to study ice deformation as part of the Jarvis Glacier Project, and we showcase our successful initial run that includes the development, deployment, and data collection processes. In this case study, we installed our tilt sensor system in two boreholes drilled close to the lateral shear margin of Jarvis Glacier in Alaska and successfully collected over 16 months of uninterrupted borehole deformation data in a harsh polythermal glacial environment. The data included gravity and magnetic data that we used to track the orientation of our sensors in the boreholes over time, and the resultant kinematic measurements enabled us to compute borehole deformation. While our sensors were applied under polythermal thermal regime conditions, we present use cases for our sensors in a variety of glacier thermal regimes including Athabasca glacier, a temperate glacier in Canada, and in Antarctic regions with similar polythermal regimes such as ice streams and outlet glaciers. Sensors embedded in our tilt sensors can be modified to suit different needs, and the tilt sensor can also be modified for different boreholes and glacier conditions. Our goal is to improve the accessibility of borehole geophysics research mainly through supporting production efforts of our sensor for various research needs. With an established sensor development plan, successful applications in the field, and years of experience, our team is open to potential research collaborations with researchers who are interested in using our tilt sensors. Our team is working with Polar Research Equipment, a Dartmouth alumni founded company that specializes in the development of polar research tools, that will serve as a commercial resource for researchers who may require support during the development process or mass-production of our cost-efficient (~20% the price of other commercial versions) yet effective tilt sensors.

Oceanographic structure is often represented as a collection of vertical profiles, i.e. temperature, salinity, and/or biogeochemical values at various depths. These profiles contain information about water mass structures and the boundaries between them, which are consequences of the integrated effects of water mass formation, advection, and destruction. In recent years, researchers have applied various unsupervised profile classification methods in an attempt to identify a set of “profile types” and the spatially coherent regimes associated with them. These efforts have identified a number of regimes that are consistent with existing oceanographic knowledge, and they have also identified previously under-appreciated structural differences. However, as this application area matures, questions remain about the strengths and limitations of these methods as applied to oceanography. A key question is “under what circumstances does unsupervised profile classification produce interpretable and scientifically useful knowledge?” Here, I explore the mechanisms and parameters of various unsupervised learning approaches, in particular Gaussian Mixture Modeling, in an attempt to clarify the conditions under which unsupervised learning produces robust, interpretable, and trustworthy understanding. As with pattern classification approaches in general, there is a tradeoff between interpretability and accuracy (the ability of the method to represent the full underlying structure of the system). As a case study, I explore an unsupervised profile classification application in the Weddell Gyre. I show that, using a combination of statistical guidance, expert judgment, and traditional oceanographic analysis, we can, in some cases, increase the interpretability of a profile classification model with acceptable losses in accuracy. The goal is to elucidate the conditions under which unsupervised learning can be fully integrated into the oceanographic knowledge generation process, both by confronting existing understanding and by highlighting new avenues for exploration.

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. Speci¦cally, 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 ¦nite 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 ¦eld 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 mis¦t of 2.86% on test data and 2% on training data. The results con¦rm the e¨cacy and effectiveness of the proposed U-Net architecture in predicting the subsurface resistivity variation from the DC resistivity data.

Bentonite is a fine-grained geologic material consisting mainly of montmorillonite clay. It presents a low permeability, a high swelling pressure, and a strong capacity to retain radionuclides that make it an important component in current efforts to design engineered barrier systems for the isolation of radioactive waste. In these barriers, the thermal gradient generated by radioactive decay is expected to lead to coupled thermal-hydrologic-mechanical-chemical (THMC) processes that may impact barrier performance. However, constitutive relations characterizing the THMC coupled properties of bentonite in variable temperature, aqueous chemistry, and dry density conditions remain incompletely understood. Here, we use high-performance molecular dynamics (MD) simulations to gain insight into the THMC constitutive relations of compacted montmorillonite clay. Specifically, we report large-scale MD simulations of water-saturated clay assemblages containing 27 montmorillonite particles performed using the codes GROMACS and LAMMPS (Fig. 1). Simulations were carried out using the replica-exchange MD (REMD) technique, with 96 replicas of the system with a wide range of temperatures up to 100 °C. In addition, simulated systems were progressively dehydrated to examine a range of dry densities. Results were analyzed to determine a series of properties including hydraulic conductivity, water and ion self-diffusivity, heat capacity, thermal expansion, and swelling pressure as a function of temperature, dry density, and the type of exchangeable cations (Na, K, Ca). Finally, simulation predictions were validated and refined by benchmarking against experimental results and previous MD simulation predictions. This research provides new insight into the coupled THMC properties of clay barrier systems and advances efforts to predict the performance of engineered clay barriers over a long timescale.

Elevated ozone (O3) pollution in the warm season is an emerging environmental concern affecting global highly urbanized megacities. In southwestern China, full characterization of causes for O3 pollution has been stymied by limited observations and the dominant factors that influence O3 variability on a long-term basis still lack understanding. Herein, we identified O3 variations and inferred trends in precursor emissions in Chengdu over 2013–2020 based on extensive ambient measurements, emission inventory, and satellite data. Numerical models were used to investigate the changes in meteorological variability and biogenic emissions. Trends of O3 in urban areas show deterioration (+14.0% yr−1) between 2013 and 2016 followed by a slight decrease over 2017–2020, while O3 levels in rural areas generally show a downward trend (−2.9% yr−1) during 2014–2020. Both emission inventory (−3.7% yr−1) and OMI satellite columns (−4.5% yr−1) depict strong decline trends in NOx emissions, while satellite HCHO columns exhibit a flattened downward trend of VOC emissions (−1.8% yr−1), which caused rural areas shifted from VOCs-limited to transitional or NOx-limited regime since 2016. Considering metropolitan Chengdu remains VOCs-limited regime over time, the existing regulatory framework involving simultaneous NOx and VOCs control would result in evident O3 improvements in the near future. Despite benefits from anthropogenic emission reductions, we demonstrate that meteorological conditions and enhanced biogenic emissions over the warm season could partially or even fully offset effects attributed to emission changes, making the net effects obscure. This finding provides robust evidence of reductions in NOx and VOCs emission and informs effective O3 mitigation policies for megacities which undergo similar emission pathways in Chengdu.

Recent years have witnessed devastating weather extremes, so it is no surprise that people have started wondering, “what’s up with the weather?” The potential role of climate change in making these extremes worse has gained traction not only in the scientific community but in public discourse a well. While scientists have made considerable progress on statistical “climate attribution”—a way of assessing the probability that climate change is influencing the character of some extreme weather events—public understanding has not kept pace. However, members of society do not need to know everything about climate and its related changes to make decisions. Rather, building climate literacy across society is paramount to better inform those decisions. There are many facets of climate literacy and many ways to help society become comfortable with the concepts. While riddled with challenges, one way of explaining the statistical methods around climate attribution is through carefully designed spinner boards. This is just one demonstration of how to explain a difficult subject, and if properly developed and used, this method holds the potential to significantly improve climate literacy.