Global climate model projections suggest that 21st century climate change will bring significant drying in the midlatitudes. Recent glacier modeling suggests that runoff from glaciers will continue to provide substantial freshwater in many drainage basins, though the supply will generally diminish throughout the century. In the absence of dynamic glacier ice within global climate models (GCMs), a comprehensive picture of future basin-scale water availability for human and ecosystem services has been elusive. Here, we leverage the results of existing GCMs and a global glacier model to compute the effect of glacial runoff on the Standardized Precipitation-Evapotranspiration Index (SPEI), an indicator of basin-scale water availability. We find that glacial runoff tends to increase mean SPEI and reduce interannual variability, even in basins with relatively little glacier cover. However, in many basins we find inter-GCM spread comparable to the amplitude of the ensemble mean glacial effect, which suggests considerable structural uncertainty.

Deep learning (DL) methods have shown great promise for accurately predicting hydrologic processes but have not yet reached the complexity of traditional process-based hydrologic models (PBHM) in terms of representing the entire hydrologic cycle. The ability of PBHMs to simulate the hydrologic cycle makes them useful for a wide range of modeling and simulation tasks, for which DL methods have not yet been adapted. We argue that we can take advantage of each of these approaches to couple DL methods into PBHMs as individual process parameterizations. We demonstrate that this is viable by developing DL process parameterizations for turbulent heat fluxes and couple them into the Structure for Unifying Multiple Modeling Alternatives (SUMMA), a modular PBHM modeling framework. We developed two DL parameterizations and integrated them into SUMMA, resulting in a one way coupled implementation (NN1W) which relies only on model inputs and a two-way coupled implementation (NN2W), which also incorporates SUMMA-derived model states. Our results demonstrate that the DL parameterizations are able outperform calibrated standalone SUMMA benchmark simulations. Further we demonstrate that the two-way coupling can simulate the long-term latent heat flux better than the standalone benchmark. This shows that DL methods can benefit from PBHM information, and the synergy between these modeling approaches is superior to either approach individually.

Knowledge of crustal stress fields is essential for understanding tectonics and earthquake generation. One approach for estimating the crustal stress field is based on the focal mechanisms of earthquakes. This study investigated the focal mechanisms of approximately 110,000 microearthquakes in the area of the Japanese islands that occurred at a depth shallower than 20 km, based on the first-motion polarities picked by a simple neural network model. The model was first trained using a data set of mainly moderate to large earthquakes throughout Japan. Following on, the model was re-trained using a data set of microearthquakes in two regions of Japan. The threshold of the confidence score from the neural network model was chosen to maximize the overall quality of the focal mechanism solutions. The P- and T-axes of the numerous focal mechanism solutions provided more detailed distributions of the crustal stress field. For example, in the Chugoku region, small differences were observed in the trend of P-axes azimuths between the northern and southern areas, spatially corresponding to geodetic observations. The results of this study are useful for revealing the crustal stress field, and, as such, for assessing past and current tectonic activities and potential future earthquake generation.

Drought stress threatens vegetation dynamics across diverse ecosystems. Monitoring how vegetation responds to water stress is vital for ecological conservation. The response of vegetation photosynthesis to water availability variations in Southwest China from 2008 to 2018 is investigated in this study. The solar-induced fluorescence (SIF) derived from GOME-2 is used to characterize photosynthetic changes. We examined the sensitivity of SIF anomaly to standardized precipitation-evapotranspiration index (SPEI) at multiple time scales to evaluate the drought impacts on different ecosystems (i.e. forests, croplands, grasslands, and shrublands). We find that (1) SIF has significant yet weak correlations to SPEI across major ecosystems in Southwest China; (2) Forests are more sensitive to short-term droughts in comparison with other ecosystems. (3) Cropland, grassland, and shrubland are more subjected to long-term droughts compared to forests. Our findings indicate that, in Southwest China, satellite SIF may not be effective in monitoring the impact of drought on vegetation due to its weak response to SPEI. The robustness of using satellite-observed SIF to assess drought’s effects still needs to be further tested with high-resolution SIF data.

The 2016 M7.8 Kaikōura earthquake is one of the most complex earthquakes in recorded history, with significant rupture of at least 21 crustal faults. Using a matched-filter detection routine, precise cross-correlation pick corrections, and accurate location and relocation techniques, we construct a catalog of 33,328 earthquakes between 2009 and 2020 on and adjacent to the faults that ruptured in the Kaikōura earthquake. We also compute focal mechanisms for 1,755 of the earthquakes used as templates. Using this catalog we reassess the rupture pathway of the Kaikōura earthquake. In particular we show that: (1) the earthquake nucleated on the Humps Fault; (2) there is a likely linking offshore reverse fault between the southern fault system and the Papatea Fault, which could explain the anomalously high slip on the Papatea Fault; (3) the faults that ruptured in the 2013 Cook Strait sequence were reactivated by the Kaikōura earthquake and may have played a role in the termination of the earthquake; and (4) no seismicity on an underlying subduction interface is observed beneath almost all of the ruptured region suggesting that if deformation did occur on the plate interface then it occurred aseismically and did not play a significant role in generating co-seismic ground motion.

The Sumatran fault is an arc-parallel dextral strike-slip fault that accommodates much of the right-lateral component of oblique subduction of the Indian-Australian plate beneath the Sunda plate. The 1900-km-long fault is divided into multiple segments, some of which ruptured the surface during the moderate to large historical earthquakes. The northern Sumatran fault in Aceh Province has not ruptured in the past 120 years and is considered a seismic gap. Since 2012, we have mapped the northern Sumatran fault based on the ALOS (Advanced Land Observing Satellite) PRISM (Panchromatic Remote-sensing Instrument for Stereo Mapping) satellite images. We also conducted several campaigns of geologic fieldworks. The Sumatran fault in the study area is composed of the Aceh and Seulimeum segments. Both segments exhibit conspicuous tectonic landforms, including fault scarps, aligned saddles, offset streams, and linear valleys. The Aceh segment does not show clear geomorphic evidence of late Quaternary movement north of 5°27’N, and the Seulimeum segment appears to accommodate much of the right-lateral motion. We conducted a trenching survey of the Seulimeum segment at Lamtamot, where the fault offsets fluvial terraces. We identified geologic evidence of four surface-rupturing events that occurred after AD1265-1365 and before AD1892. The average recurrence interval of the surface-rupturing earthquake is calculated at 130-210 years. At least 120 years, close to the shortest estimated average recurrence interval, have passed since the last faulting event. We estimate that the probability of a massive earthquake on the northernmost Sumatran fault is high.

The advocacy activities necessary to sustain healthy watersheds and improve impaired ones ultimately rely on the democratic process, and therefore depend on a public that values our coastal resources and understands the role that water quality plays in maintaining that value. We contend that an opportunity exists to improve the temporal and spatial density of monitoring by reducing the cost of collecting measurements, while simultaneously fostering an informed and invested public. We envision a distributed water quality monitoring sensor network, composed of low-cost ($1000-$2000) profiling devices we call TideRiders, built and operated by private citizens and local educational organizations and supported by an institution-hosted centralized data and control portal. The TideRider concept engages the public not just in the collection of data but also in the building, deployment, operation, and recovery of these robot sensors. TideRiders will carry a suite of basic water quality instrumentation (temperature, conductivity, and dissolved oxygen), transmit data and accept commands over the cellular network, and can sample surface and bottom waters by surfacing and submerging on a programmable schedule. Operators will harness tidal currents to move their TideRiders deliberately around an embayment, essentially by surfacing in a favorable tide and anchoring on the bottom in an adverse tide. A network of TideRiders deployed in tidally-dominated estuaries like Buzzards Bay and Narragansett Bay could provide basic water quality data at several-hour intervals for weeks at a time by “virtually mooring” in center-bay locations that are otherwise only accessible by boat and therefore typically sampled less frequently than shore stations. We present preliminary field results from a series of prototypes designed and built by students. The prototype devices utilize a novel low-cost semi-passive shallow-water buoyancy engine and were constructed for less than $1000 in parts.

A growing body of literature investigates convective organisation, but few studies to date have sought to investigate how wind shear plays a role in the spatial organization of shallow (trade-wind) convection. The present study hence investigates the morphology of precipitating marine cumulus convection using large-eddy-simulation experiments with zonal forward and backward shear and without shear. One set of simulations includes evaporation of precipitation, promoting for cold-pool development, and another set inhibits evaporation of precipitation and thus cold-pool formation. Without (or with only weak) subcloud-layer shear, conditions are unfavourable for convective deepening, as clouds remain stationary relative to their subcloud-layer roots so that precipitative downdrafts interfere with emerging updrafts. Under subcloud-layer forward shear, where the wind strengthens with height (a condition that is commonly found in the trades), clouds move at greater speed than their roots, and precipitation falls downwind away from emerging updrafts. Forward shear in the subcloud layer appears to promote the development of stronger subcloud circulations, with greater divergence in the cold-pool area downwind of the original cell and larger convergence and stronger uplift at the gust front boundary. As clouds shear forward, a larger fraction of precipitation falls outside of clouds, leading to more moistening within the cold pool (gust front).

Ocean phytoplankton play a critical role in the global carbon cycle, contributing ~50% of global photosynthesis. As planktonic organisms, phytoplankton encounter significant environmental variability as they are advected throughout the ocean. How this variability impacts phytoplankton growth rates and population dynamics remains unclear. Here, we systematically investigated the impact of different rates and magnitudes of sea surface temperature (SST) variability on phytoplankton community growth rates using surface drifter observations from the Southern Ocean (> 30oS) and a phenotype-based ecosystem model. Short-term SST variability (<7 days) had a minimal impact on phytoplankton community growth rates. Moderate SST changes of 3-5oC over 7-21 days produced a large time lag between the temperature change and the biological response. The impact of SST variability on community growth rates was nonlinear and a function of the rate and magnitude of change. Additionally, the nature of variability generated in a Lagrangian reference frame (following trajectories of surface water parcels) was larger than that within an Eulerian reference frame (fixed point), which initiated different phytoplankton responses between the two reference frames. Finally, we found that these dynamics were not captured by the Eppley growth model commonly used in global biogeochemical models and resulted in an overestimation of community growth rates, particularly in dynamic, strong frontal regions of the Southern Ocean. This work demonstrates that the timescale for environmental selection (community replacement) is a critical factor in determining community composition and takes a first step towards including the impact of variability and biological response times into biogeochemical models.

Multi-plant imaging using arrays of low-cost cameras is a successful strategy for capturing affordable high-throughput plant phenotyping data. An imaging platform of this type can enable simultaneous imaging of hundreds to thousands of plants. The resulting datasets enable analysis of dynamic plant growth, development, and environmental responses at high temporal resolution. Full analysis of these datasets requires the identification of individual plants for measurement, but computational separation of individual plants becomes challenging when neighboring plants overlap. Here, we introduce the use of the watershed transform to segment moderately overlapping plants in multi-plant time series datasets. Rather than focusing on segmenting plants in individual images, we utilize information encoded in the entire time series to propagate plant labels from an early time point when individual plants are separate to later time points. In preliminary studies, using this method allowed us increase the analyzable size of the dataset by 28%.

The frequency, size, and intensity of wildfires in California have increased substantially in recent years, leading to widespread mandatory evacuations affecting millions of residents. However, because evacuation orders are implemented by local agencies, there is limited quantitative evidence on the scope of evacuations statewide. In order to improve the understanding of wildfire evacuations, we assembled information on historical evacuation orders for two distinct wildfire-prone regions --- Fresno and Sonoma county --- in California. This data was used to understand how the frequency and extent of evacuations have changed over time. We then combined this information with census data to characterize which populations have been most affected by evacuation orders. Ultimately, our work aims to quantify this important element of wildfire impacts in key regions around California. Collectively, it provides a starting point for a public database of evacuation orders that could be used by researchers and policymakers to better understand dynamics and improve decision-making around wildfire evacuations.

The spatio-temporal variability of bedload transport processes poses considerable challenges for bedload monitoring systems. One such system, the Swiss plate geophone (SPG), has been calibrated in several gravel-bed streams using direct sampling techniques. The linear calibration coefficients linking the signal recorded by the SPG system to the transported bedload can vary between different monitoring stations by about a factor of six, for reasons that remain unclear. Recent controlled flume experiments allowed us to identify the grain-size distribution of the transported bedload as a further site-specific factor influencing the signal response of the SPG system, along with the flow velocity and the bed roughness. Additionally, impact tests performed at various field sites suggested that seismic waves generated by impacting particles can propagate over several plates of an SPG array, and thus potentially bias the bedload estimates. To gain an understanding of this phenomenon, we adapted a test flume by installing a partition wall to shield individual sensor plates from impacting particles. We show that the SPG system is sensitive to seismic waves that propagate from particle impacts on neighboring plates or on the concrete bed close to the sensors. Based on this knowledge, we designed a filter method that uses time-frequency information to identify and eliminate these “apparent” impacts. Finally, we apply the filter to four field calibration datasets and show that it significantly reduces site-to-site differences between calibration coefficients and enables the derivation of a single calibration curve for total bedload at all four sites.

Astronomical variations in tidal magnitude can strongly modulate the severity of coastal flooding on daily, monthly, and interannual timescales. Here, we present a new quasi-nonstationary skew surge joint probability method (qn-SSJPM) that estimates interannual fluctuations in flood hazard caused by the 18.6 and quasi 4.4-year modulations of tides. We demonstrate that qn-SSJPM-derived storm tide frequency estimates are more precise and stable compared with the standard practice of fitting an extreme value distribution to measured storm tides, which is often biased by the largest few events within the observational period. Applying the qn-SSJPM in the Gulf of Maine, we find significant tidal forcing of winter storm season flood hazard by the 18.6-year nodal cycle, whereas 4.4-year modulations and a secular trend in tides are small compared to interannual variation and long-term trends in sea-level. The nodal cycle forces decadal oscillations in the 1% annual chance storm tide at an average rate of ±13.5 mm/y in Eastport, ME; ±4.0 mm/y in Portland, ME; and ±5.9 mm/y in Boston, MA. Currently (in 2020), nodal forcing is counteracting the sea-level rise-induced increase in flood hazard; however, in 2025, the nodal cycle will reach a minimum and then begin to accelerate flood hazard increase as it moves toward its maximum phase over the subsequent decade. Along the world’s meso-to-macrotidal coastlines, it is therefore critical to consider both sea-level rise and tidal non-stationarity in planning for the transition to chronic flooding that will be driven by sea-level rise in many regions over the next century.

Geomagnetic storms, which are governed by the plasma magnetohydrodynamics of the solar-interplanetary-magnetosphere system, entail a formidable challenge for physical forward modeling. Yet, the abundance of high quality observational data has been amenable for the application of data-hungry neural networks to geomagnetic storm forecasting. Previous applications of neural networks to storm forecasting have utilized solar wind observations from the Earth-Sun first Lagrangian point (L1) or closer and have all generated deterministic output without uncertainty estimates. Furthermore, forecasting work has focused on indices that are also sensitive to induced internal magnetic fields, complicating the forecasting problem with another layer of non-linearity. We address these points, presenting neural networks trained on observations from both the solar disk and the L1 point. Our architecture generates reliable probabilistic forecasts over Est, the external component of the disturbance storm time index, showing that neural networks can gauge confidence in their output.

Reducing uncertainty in the global carbon budget requires better quantification of ocean CO2 uptake and its temporal variability. Several methodologies for reconstructing air-sea CO2 exchange from sparse pCO2 observations indicate larger decadal variability than estimated using ocean models. We develop a new application of multiple Large Ensemble Earth system models to assess these reconstructions’ ability to estimate spatiotemporal variability. With our Large Ensemble Testbed, pCO2 fields from 25 ensemble members each of four independent Earth system models are subsampled as the observations and the reconstruction is performed as it would be with real- world observations. The power of a testbed is that the perfect reconstruction is known for each of the 100 original model fields; thus, reconstruction skill can be comprehensively assessed. We find that a commonly used neural-network approach can skillfully reconstruct air-sea CO2 fluxes when and where it is trained with sufficient data. Flux bias is low for the global mean and Northern Hemisphere, but can be regionally high in the Southern Hemisphere. The phase and amplitude of the seasonal cycle are accurately reconstructed outside of the tropics, but longer-term variations are reconstructed with only moderate skill. For Southern Ocean decadal variability, insufficient sampling leads to a 39% [15%:58%, interquartile range] overestimation of amplitude, and phasing is only moderately correlated with known truth (r=0.54 [0.46:0.63]). Globally, the amplitude of decadal variability is overestimated by 21% [3%:34%]. Machine learning, when supplied with sufficient data, can skillfully reconstruct ocean properties. However, data sparsity remains a fundamental limitation to quantification of decadal variability in the ocean carbon sink.

Identifying fluid flow maldistribution in planar geometries is a well-established problem in subsurface science/engineering. Of particular importance to the thermal performance of Engineered (or “Enhanced”) Geothermal Systems (EGS) is identifying the existence of non-uniform (i.e., heterogeneous) permeability and subsequently predicting advective heat transfer. Here, machine learning via a Genetic Algorithm (GA) identifies the spatial distribution of an unknown permeability field in a two-dimensional Hele-Shaw geometry (i.e., parallel-plates). The inverse problem is solved by minimizing the L2-norm between simulated Residence Time Distribution (RTD) and measurements of an inert tracer breakthrough curve (BTC) (C-Dot nanoparticle). Principal Component Analysis (PCA) of spatially-correlated permeability fields enabled reduction of the parameter space by more than a factor of ten and restricted the inverse search to reservoir-scale permeability variations. Thermal experiments and tracer tests conducted at the mesoscale Altona Field Laboratory (AFL) demonstrate that the method accurately predicts the effects of extreme flow channeling on heat transfer in a single bedding-plane rock fracture. However, this is only true when the permeability distributions provide adequate matches to both tracer RTD and frictional pressure loss. Without good agreement to frictional pressure loss, it is still possible to match a simulated RTD to measurements, but subsequent predictions of heat transfer are grossly inaccurate. The results of this study suggest that it is possible to anticipate the thermal effects of flow maldistribution, but only if both simulated RTDs and frictional pressure loss between fluid inlets and outlets are in good agreement with measurements.

Atmospheric rivers (AR) are large and narrow filaments of horizontal water vapor poleward transport. The size of ARs determines how much water vapor is carried from the tropics into higher latitudes, leading to extreme precipitation. Current AR detection and tracking algorithms have resulted in large uncertainty in the estimation of their sizes, with areas varying over several orders of magnitude among different detection methods. We develop and implement five independent methods to reduce the uncertainty and characterize the size of ARs that make landfall over the west coast of North America in the 1980-2017 period. ARs that originate in the northwest Pacific (WP) (100oE-180oE) have larger sizes and are more zonally oriented than those from the northeast Pacific (EP) (180oE-240oE). ARs become smaller through their lifecycle, mainly due to reductions in their width. They also become more meridionally oriented towards the end of their lifecycle. Overall, the size estimation methods proposed in this work provide a range of AR areas (between 7x1011m2 and 1013 m2) that is several orders of magnitude narrower than current methods estimation. This methodology can provide statistical constraints in size and geometry for the AR detection and tracking algorithms; and objective insight into AR size studies under climate change scenarios in the future.

In-situ conjugate electromagnetic ion cyclotron (EMIC) waves observed by the Swarm mission in both hemispheres are presented. A complex and unusual pattern of Alfvénic EMIC wave energy is observed, with a mid-latitude peak close to the source at L=3.3, as well as a secondary lower L-peak. A wave propagation model reveals that the secondary peak at L=1.7 may be explained by wave power being redirected equatorward due to the Buchsbaum resonance, crossing and interfering with the same EMIC wave power propagating equatorwards from the opposite hemisphere. This interference creates a coherent equatorial driver for a low-L field line resonance at the secondary peak, and which is associated with strong shear-to-fast mode coupling in the ionosphere. This behavior complicates the interpretation of low-Earth orbit EMIC data for applications assessing radiation belt loss. Combined LEO observations and modelling enable these novel and localized magnetosphere-ionosphere EMIC wave propagation pathways to be identified.