Geoscience applications have been using sophisticated machine learning methods to model complex phenomena. These models are described as black boxes since it is unclear what relationships are learned. Models may exploit spurious associations that exist in the data. The lack of transparency may limit user’s trust, causing them to avoid high performance models since they cannot verify that it has learned realistic strategies. EXplainable Artificial Intelligence (XAI) is a developing research area for investigating how models make their decisions. However, XAI methods are sensitive to feature correlations. This makes XAI challenging for high-dimensional models whose input rasters may have extensive spatial-temporal autocorrelation. Since many geospatial applications rely on complex models for target performance, a recommendation is to combine raster elements into semantically meaningful feature groups. However, it is challenging to determine how best to combine raster elements. Here, we explore the explanation sensitivity to grouping scheme. Experiments are performed on FogNet, a complex deep learning model that uses 3D Convolutional Neural Networks (CNN) for coastal fog prediction. We demonstrate that explanations can be combined with domain knowledge to generate hypotheses about the model. Meteorological analysis of the XAI output reveal FogNet’s use of channels that capture relationships related to fog development, contributing to good overall model performance. However, analyses also reveal several deficiencies, including the reliance on channels and channel spatial patterns that correlate to the predominate fog type in the dataset, to make predictions of all fog types. Strategies to improve FogNet performance and trustworthiness are presented.
The climatological mean barotropic vorticity budget is analyzed to investigate the relative importance of surface wind stress, topography and nonlinear advection in dynamical balances in a global ocean simulation. In addition to a pronounced regional variability in vorticity balances, the relative magnitudes of vorticity budget terms strongly depend on the length-scale of interest. To carry out a length-scale dependent vorticity analysis in different ocean basins, vorticity budget terms are spatially filtered by employing the coarse-graining technique. At length-scales greater than 10o (or roughly 1000 km), the dynamics closely follow the Topographic-Sverdrup balance in which bottom pressure torque, surface wind stress curl and planetary vorticity advection terms are in balance. In contrast, when including all length-scales resolved by the model, bottom pressure torque and nonlinear advection terms dominate the vorticity budget (Topographic-Nonlinear balance), which suggests a prominent role of oceanic eddies, which are of Ο(10-100) km in size, and the associated bottom pressure anomalies in local vorticity balances at length-scales smaller than 1000 km. Overall, there is a transition from the Topographic-Nonlinear regime at scales smaller than 10o to the Topographic-Sverdrup regime at length-scales greater than 10o. These dynamical balances hold across all ocean basins; however, interpretations of the dominant vorticity balances depend on the level of spatial filtering or the effective model resolution. On the other hand, the contribution of bottom and lateral friction terms in the barotropic vorticity budget remains small and is significant only near sea-land boundaries, where bottom stress and horizontal friction generally peak.
Most of the ocean’s kinetic energy is contained within the mesoscale eddy field. Models that do not resolve these eddies tend to parameterize their impacts through down-gradient transport of buoyancy and tracers, aiming to reduce the large-scale available potential energy and spread tracers. However, the parameterizations used in the ocean components of current generation Earth System Models (ESMs) rely on an assumption of a flat ocean floor even though observations and high-resolution modelling show that eddy transport is sensitive to the potential vorticity gradients associated with a sloping sea floor. We show that buoyancy diffusivity diagnosed from idealized eddy-resolving simulations is indeed reduced over both prograde and retrograde bottom slopes (topographic wave propagation along or against the mean flow, respectively) and that the reduction can be skilfully captured by mixing length parameterization by introducing the topographic Rhines scale as a length scale. This modified ‘GM’ parameterization enhances the strength of thermal wind currents over the slopes in coarse-resolution, non-eddying, simulations. We find that in realistic global coarse-resolution simulations the impact of topography is most pronounced at high latitudes, enhancing the mean flow strength and reducing temperature and salinity biases. Reducing buoyancy diffusivities further with a mean-flow dependent eddy efficiency factor has notable effects also at lower latitudes and leads to reduction of global mean biases.
This study investigates the response of the semidiurnal tide (SDT) to the 2013 major sudden stratospheric warming (SSW) event using meteor radar wind observations and mechanistic tidal model simulations. In the model, the background atmosphere is constrained to meteorological fields from the Navy Global Environmental Model - High Altitude analysis system. The solar (thermal) and lunar (gravitational) SDT components are forced by incorporating hourly global temperature tendency fields from the ERA5 forecast model, and by specifying the M2 and N2 lunar gravitational potentials, respectively. The simulated SDT response is compared against meteor wind observations from the CMOR (43.3◦N, 80.8◦W), Collm (51.3◦N, 13.0◦E), and Kiruna (67.5◦N, 20.1◦E) radars, showing close agreement with the observed amplitude and phase variability. Numerical experiments investigate the individual roles of the solar and lunar SDT components in shaping the net SDT response. Further experiments isolate the impact of changing propagation conditions through the zonal mean background atmosphere, non-linear wave-wave interactions, and the SSW-induced stratospheric ozone redistribution. Results indicate that between 80-97 km altitude in the northern hemisphere mid-to-high latitudes the net SDT response is driven by the solar SDT component, which itself is shaped by changing propagation conditions through the zonal mean background atmosphere and by non-linear wave-wave interactions. In addition, it is demonstrated that as a result of the rapidly varying solar SDT during the SSW the contribution of the lunar SDT to the total measured tidal field can be significantly overestimated.
We propose that the observed enrichment of sulfur at the surface of Mercury (up to 4 wt.%) is the product of silicate sulfidation reactions with a S-rich reduced volcanogenic gas phase. Here, we present new experiments on the sulfidation behavior of olivine, diopside, and anorthite. We investigate these reaction products, and those of sulfidized glasses with Mercury compositions previously reported, by mid-IR reflectance spectroscopy. We investigate both the reacted bulk materials as powders, as well as cross-sections of the reaction products by in-situ micro-IR spectroscopy. The mid-IR spectra confirm the presence of predicted reaction products including quartz. The mid-IR reflectance of sulfide reaction products, such as CaS (oldhamite) or MgS (niningerite), is insufficient to be observed in the complex run products. However, the ESA/JAXA BepiColombo mission to Mercury will be able to test our hypothesis by investigating correlated abundances of sulfides with other reaction products such as quartz.
Extensive regions in the permafrost zone are projected to become climatically unsuitable to sustain permafrost peatlands over the next century, suggesting transformations in these landscapes that can leave large amounts of permafrost carbon vulnerable to post-thaw decomposition. We present three years of eddy covariance measurements of CH4 and CO2 fluxes from the degrading permafrost peatland Iskoras in Northern Norway, which we disaggregate into separate fluxes of palsa, pond, and fen areas using information provided by the dynamic flux footprint in a novel ensemble-based Bayesian deep neural network framework. The three-year mean CO2-equivalent flux is estimated to be 106 gCO2 m-2 yr-1 for palsas, 1780 gCO2 m-2 yr-1 for ponds, and -31 gCO2 m-2 yr-1 for fens, indicating that possible palsa degradation to thermokarst ponds would strengthen the local greenhouse gas forcing by a factor of about 17, while transformation into fens would slightly reduce the current local greenhouse gas forcing.
Recent studies show that stoichiometric elemental ratios of marine ecosystems are not static at Redfield proportions but vary systematically between biomes. However, the wider Atlantic Ocean is under-sampled for particulate organic matter (POM) elemental composition, especially as it comes to phosphorus. Thus, it is uncertain how environmental variation in this region translates into shifts in C:N:P. To address this, we analyzed hydrography, genomics, and POM concentrations from 877 stations on the meridional transects AMT28 and C13.5, spanning the Atlantic Ocean. We observed nutrient-replete, high-latitude ecosystem C:N:P to be significantly lower than the oligotrophic gyres. Latitudinal and zonal differences in elemental stoichiometry were linked to overall nutrient supply as well as N vs. P limitation. C:P and N:P were generally higher in the P-stressed northern region compared to southern hemisphere regions. We also detected a zonal difference linked to a westward deepening nutricline and a shift from N to P limitation. We also evaluated possible seasonal changes in C:N:P across the basin and predicted these to be limited. Overall, this study confirms latitudinal shifts in surface ocean POM ratios but reveals previously unrecognized hemisphere and zonal gradients. This work demonstrates the importance of understanding how regional shifts in hydrography and type of nutrient stress shape the coupling between Atlantic Ocean nutrient and carbon cycles.
Offshore wind energy deployment in the US is expected to increase in the years to come, with proposed wind farm sites located in regions with high-risk for tropical cyclones. Yet, the wind turbine design criteria outlined by the International Electrotechnical Commission for extreme events may not account for the severe wind conditions in tropical cyclones, even the weaker storms that are likely to reach mid-Atlantic wind resource areas. To evaluate if current design standards capture the extreme conditions of these storms, we perform idealized large-eddy simulations of five tropical cyclones (two category-1, two category-2, and one category-3 storms) using the Weather Research and Forecasting model. Wind conditions near the eyewall of category-1, category-2 and category-3 storms can exceed current design standards for offshore wind turbines. Hub-height winds can exceed design criteria for Class I and Class T turbines for 50-year recurrence periods. Moreover, wind speed shear across the turbine rotor layer is larger than assumed in design specifications. Vertical variations in wind direction across the turbine rotor layer are also large for tropical cyclones of all intensity levels, suggesting design standards should include veer, which can amplify loads in wind turbines.
Shallow cloud decks residing in or near the boundary layer cover a large fraction of the Southern Ocean (SO) and play a major role in determining the amount of shortwave radiation reflected back to space from this region. In this article, we examine the macrophysical characteristics and thermodynamic phase of low clouds (tops < 3 km) and precipitation using ground-based ceilometer, depolarization lidar and vertically-pointing W-band radar measurements collected during the Macquarie Island Cloud and Radiation Experiment (MICRE) from April 2016-March 2017. During MICRE, low clouds occurred ~65% of the time on average (slightly more often in austral winter than summer). About 2/3 of low clouds were cold-topped (temperatures < 0°C); these were thicker and had higher bases on average than warm-topped clouds. 83-88% of cold-topped low clouds were liquid phase at cloud base (depending on the season). The majority of low clouds had precipitation in the vertical range 150 to 250 meters below cloud base, a significant fraction of which did not reach the surface. Phase characterization is limited to the period between April 2016 and November 2016. Small-particle (low-radar-reflectivity) precipitation (which dominates precipitation occurrence) was mostly liquid below-cloud, while large-particle precipitation (which dominates total accumulation) was predominantly mixed/ambiguous or ice phase. Approximately 40% of cold-topped clouds had mixed/ambiguous or ice phase precipitation below (with predominantly liquid phase cloud droplets at cloud base). Below-cloud precipitation with radar reflectivity factors below about -10 dBZ were predominantly liquid, while reflectivity factors above about 0 dBZ were predominantly ice.
A large portion of Central-Western Asia is made up of contiguous closed basins, collectively termed as the Asian Endorheic Basins (AEB). As these retention basins are only being replenished by the intermittent precipitation, increasing droughts in the region and a growing demand for water have been presumed to jointly contributed to the land degradation. To understand the impact of climate change and human activities on dryland vegetation over the AEB, we conducted trend and partial correlation analysis of vegetation and hydroclimatic change from 2001 to 2021 using multi-satellite observations, including vegetation greenness, total water storage anomalies (TWSA) and meteorological data. Here we show that much of the AEB (65.53%) exhibited a greening trend over the past two decades. Partial correlation analyses indicated that climatic factors had varying effects on vegetation productivity as a function of vegetation types and aridity. In arid AEB, precipitation dominated the vegetation productivity trend. Such a rainfall dominance gave way to TWSA dominance in the hyper-arid AEB. We further showed that the decoupling of rainfall and hyper-arid vegetation greening was largely due to a significant expansion (17.3%) in irrigated cropland across the hyper-arid AEB. Given the extremely harsh environment in the hyper-arid AEB, our results therefore raised the concerns on the ecological and societal sustainability in this region, where a mild increase in precipitation might not be able to catch up the rising evaporative demand and water consumption resulted from global warming and irrigation intensification.
Crop models are often used to project future crop yield under climate and global change and typically show a broad range of outcomes. To understand differences in modeled responses, we analysed modeled crop yield response types using impact response surfaces along four drivers of crop yield: carbon dioxide (C), temperature (T), water (W), and nitrogen (N). Crop yield response types help to understand differences in simulated responses per driver and their combinations rather than aggregated changes in yields as the result of simultaneous changes in various drivers. We find that models’ sensitivities to the individual drivers are substantially different and often more different across models than across regions. There is some agreement across models with respect to the spatial patterns of response types but strong differences in the distribution of response types across models and their configurations suggests that models need to undergo further scrutiny. We suggest establishing standards in model evaluation based on emergent functionality not only against historical yield observations but also against dedicated experiments across different drivers to analyze emergent functional patterns of crop models.
Supervolcanic eruptions induced abrupt global cooling (roughly at a rate of ~1ºC/year lasting for years to decades), such as the prehistoric Yellowstone eruption released, by some estimates, SO2 about 100 times higher than the 1991 Mt. Pinatubo eruption. An abrupt global cooling of several ºC, even if only lasting a few years, would present immediate and drastic stress on biodiversity and food production - posing a global catastrophic risk to human society. Using a simple climate model, this paper discusses the possibility of counteracting supervolcanic cooling with the intentional release of greenhouse gases. Although well-known longer-lived compounds such as CO2 and CH₄ are found to be unsuitable for this purpose, select fluorinated gases (F-gases), either individually or in combinations, may be released at gigaton scale to offset most of the supervolcanic cooling. We identify candidate F-gases (viz. C4F6 and CH3F) and derive radiative and chemical properties of ‘ideal’ compounds matching specific cooling events. Geophysical constraints on manufacturing and stockpiling due to mineral availability are considered alongside technical and economic implications based on present-day market assumptions. The consequences of F-gas release in perturbing atmospheric chemistry are discussed in the context of those due to the supervolcanic eruption itself. The conceptual analysis here suggests the possibility of mitigating certain global catastrophic risks via intentional intervention.
The Southern Hemisphere storm track is a key component of the Earth's global circulation patterns, with a prominent role in the movement of heat and momentum across the mid-latitudes, and a controlling influence over the behaviour of synoptic eddies. Storm track characteristics are expected to change with anthropogenic forcings, leading to changes in regional weather patterns, and impacting communities through its influence on extreme events. We document projected storm track climatologies at the end of the 21st century under high- and low-emissions scenarios using the CMIP6 generation of models. We find previously described projections -- the poleward migration of the storm track, and intensification of storm activity -- persist in CMIP6. We explore projections of spatio-temporal variability of the Southern Annular Mode, and consequences for the storm track.
Oceanic lee waves play an important role in dissipating wind-driven ocean circulations and powering turbulent diapycnal mixing. Here we investigate impacts of the greenhouse warming on global energy conversion into lee waves using a linear theory of lee wave generation and output from a high-resolution (0.1° for the ocean) coupled global climate model. The global energy conversion rate into lee waves under the historical (1930s) climate condition is estimated to be 193.0±3.0 GW. Under the high carbon emission scenario, this conversion rate is projected to decrease by about 20% by the end of 21st century, due to weakened bottom large-scale mean flows, mesoscale eddies and stratification. The decrease of the conversion rate is widespread and particularly pronounced in the Gulf Stream and Drake Passage. Our results suggest significant response of oceanic lee waves to the greenhouse warming, with implications for future changes of global ocean circulations and climate.
Unlike the United States, Nigeria's installed overall electricity capacity is 12.8 GW, while the operational capacity is estimated to be 3.9 GW which is well below the current demand of 98 GW. This results in a consumer power demand shortfall of 94.1 GW across the country. As a result of this wide gap between demand and generation, only about 45% of Nigeria's citizens have access to electricity. In this paper, a comparative feasibility analysis of the utilization of a photovoltaic system with energy storage for residential application is presented. The comparative analysis is conducted to compare the feasibility of using a solar Farm with an energy storage system between the US and Nigeria. This analysis is carried out using a model developed by IREQ Hydro-Quebec Research Institute. The results are shown in phasor form to analyze the energy stored, solar intensity, and also enable the community in making informed decisions regarding reducing grid dependency.
Growing evidence indicates that a selected group of global-scale waves from the lower 3 atmosphere constitute a significant source of ionosphere-thermosphere (IT, 100-600 km) 4 variability. Due to the geometry of the magnetic field lines, this IT coupling occurs mainly at low 5 latitudes (< 30 •) and is driven by waves originating in the tropical troposphere such as the diurnal 6 eastward propagating tide with zonal wave number s =-3 (DE3) and the quasi-3-day ultra-fast 7 Kelvin wave with s =-1 (UFKW1). In this work, over 2 years of simultaneous in situ ion densities 8 from Ion Velocity Meters (IVMs) onboard the Ionospheric Connection Explorer (ICON) near 9 590 km and the Scintillation Observations and Response of the Ionosphere to Electrodynamics 10 (SORTIE) CubeSat near 420 km, along with remotely-sensed lower (ca. 105 km) and middle 11 (ca. 220 km) thermospheric horizontal winds from ICON's Michelson Interferometer for Global 12 High-resolution Thermospheric Imaging (MIGHTI) are employed to demonstrate a rich spectrum 13 of waves coupling these IT regions. Strong DE3 and UFKW1 topside ionospheric variations are 14 traced to lower thermospheric zonal winds, while large diurnal s = 2 (DW2) and zonally symmetric 15 (D0) variations are traced to middle thermospheric winds generated in situ. Analyses of diurnal 16 tides from the Climatological Tidal Model of the Thermosphere (CTMT) reveal general agreement 17 near 105 km, with larger discrepancies near 220 km due to in situ tidal generation not captured 18 by CTMT. This study highlights the utility of simultaneous satellite measurements for studies of IT 19 coupling via global-scale waves. 20