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

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.

The ILRS contribution to ITRF2020 is a time series of weekly/bi-weekly SINEXs with station position estimates and EOP, from 7-day arcs (1993.0 – 2021.0) and 15-day arcs over 1983.0- 1993.0. Each solution was obtained as the combination of loosely constrained individual solutions from the seven ILRS Analysis Centers: ASI, BKG, DGFI, ESA, GFZ, JCET and NSGF. Everyone followed strict standards agreed within the ILRS Analysis Standing Committee (ASC) and used SLR data from LAGEOS, LAGEOS-2, Etalon-1 and Etalon-2, (LAGEOS-only from 1983 to 1992). The ILRS ASC devised an innovative approach in handling systematic errors in the network, never before utilized. After a 5-year pilot-project documented in Luceri et al., (2019). The Station Systematic Error Monitoring PP (SSEM), delivered a series of long-term mean bias estimates for each station, the time intervals of applicability and their statistics. They were derived from freely adjusted station position and EOP solutions for the period 1993.0 to 2020.5, using the latest satellite CoG model. The simultaneous estimation of the station heights and measurement biases resulted in a self-consistent set of weekly bias estimates for each site. Breaks and “jumps” were used to decne the periods of applicability and to calculate the mean bias and its standard deviation. The mean biases were pre-applied in the re- analysis, limiting the remaining jitter of the bias to negligible level. This approach strengthened the estimation process without a compromise of the cnal results’ accuracy. As a result, the ILRS contribution to ITRF2020 minimized the scale difference between SLR and VLBI to below 2 mm (ITRF2014 ~9 mm). We present an overview of the procedures, models, the improvement over previous ILRS products, focusing especially on the Core ILRS sites, and an overview of how the new model has been implemented in support of the ILRS ogcial products.

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.

The Gravity Recovery and Climate Experiment (GRACE) data help to determine the total water storage anomalies (TWS) across the global scale. The various other important components such as Groundwater storage (GWS) and evapotranspiration for the region of South –East Asia have been determined. With the study of the gravity variation across the globe the long-term changes in the hydrological cycle can be determined which can be related to climate science or the influence of anthropogenic activities. The variation between the Groundwater storage (GWS) and the Total water storage (TWS) of the study area has been calculated for the pre and post-monsoon season of the study area. The variation between groundwater storage and total water storage can be visualized through geospatial analysis. Therefore, the regions with a substantial decrease in water storage can be related to various climate and anthropogenic factors hence implying a sustainable use of groundwater as a resource. Keywords: Machine Learning, Remote Sensing, Groundwater Recharge, Climate science.

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.

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.