Accurate estimates of the soil water balance components are critical for optimizing irrigation water use in agricultural fields. Estimates are normally obtained using simple water balance models and for representative areas, not taking into consideration the within variability of soil properties. In this study, we used the MOHID-Land distributed process-based model to compute the variability of the soil water balance components in a 23ha almond field located in southern Portugal, at a resolution of 5m. The main objective was the possible assessment of management zones for improving water productivity in that water-scarce region. An electromagnetic induction survey was carried out first to obtain electromagnetic conductivity images which provided the spatial distribution of the real soil electrical conductivity (ff) with depth. The spatial distribution of ff was then correlated to soil particle size distribution using an in-situ calibration. Afterward, pedotransfer functions were applied to define the soil hydraulic parameters necessary to run the distributed model and map the within soil variability at the field scale. Irrigation data was monitored on-site, at two locations, while weather data was extracted from a local meteorological station. The distributed modeling approach included the definition of potential evapotranspiration fluxes computed from the product of the reference evapotranspiration obtained according to the FAO56 Penman-Monteith equation and a crop coefficient for each stage of almond’s growing season, the variable-saturated flow using the Richards equation, and root zone water stress following a macroscopic approach. Modeling results were used to present the maps of the variability of the seasonal actual crop transpiration and soil evaporation, the mean soil moisture, seasonal runoff, and seasonal percolation. Then, management zones for improving irrigation water use in the studied almond field were proposed.

Environmental justice and equity should include access to clean water for all. It is expensive to drill borehole wells, typically over $10,000 US dollars, and so organizations working to provide wells in developing countries have typically installed community wells at some common gathering place. This requires that many users must walk long distances to access these water sources. This limits the quantity of water available to a family, and also creates vulnerabilities for the family member, usually a woman or child, sent for the water since the journey is often made early in the morning or at night in the dark. I have been drilling wells with a Kenyan team since 2010 using a simple, manual percussion hydraulic method developed by WaterForAllinternational.org whereby we can install a well generally for less than $200 US dollars excluding labor. Through their own participation in the drilling process, this low-cost enables families to pay for and drill their own well. In this way, they gain access to a much larger supply of water at or close to home, and eliminate the need and vulnerability associated with walking long distances to procure water for their family. Both the drilling apparatus and the cased well, including the pump, is constructed from materials available off-the-shelf at local hardware stores. Over the years I have made several modifications to the pump design, other infrastructure, and manufacturing process to improve the longevity, simplicity, and interchangeability of the final product. The drilling method is primarily applicable to aquifers lying above bedrock and it is feasible to drill wells to a depth of several hundred feet. The greatest challenge in the endeavor is earning the trust and cultivating the participation of the local community. This presentation will address the drilling process, the well infrastructure, and some socio-cultural aspects of the project.

Holistic approaches are needed to investigate the capacity of current water resource operations and infrastructure to sustain water supply and critical ecosystem health under projected drought conditions. Drought vulnerability is complex, dynamic, and challenging to assess, requiring simultaneous consideration of changing water demand, use and management, hydrologic system response, and water quality. We are bringing together a community of scientists from the U.S. Geological Survey, National Center for Atmospheric Research, Department of Energy, and Cornell University to create an integrated human-hydro-terrestrial modeling framework, linking pre-existing models, that can explore and synthesize system response and vulnerability to drought in the Delaware River Basin (DRB). The DRB provides drinking water to over 15 million people in New York, New Jersey, Pennsylvania, and Delaware. Critical water management decisions within the system are coordinated through the Delaware River Basin Commission and must meet requirements set by prior litigation. New York City has rights to divert water from the upper basin for water supply but must manage reservoir releases to meet downstream flow and temperature targets. The Office of the Delaware River Master administers provisions of the Flexible Flow Management Program designed to manage reservoir releases to meet water supply demands, habitat, and specified downstream minimum flows to repel upstream movement of saltwater in the estuary that threatens Philadelphia public water supply and other infrastructure. The DRB weathered a major drought in the 1960s, but water resource managers do not know if current operations and water demands can be sustained during a future drought of comparable magnitude. The integrated human-hydro-terrestrial modeling framework will be used to identify water supply and ecosystem vulnerabilities to drought and will characterize system function and evolution during and after periods of drought stress. Models will be forced with consistent input data sets representing scenarios of past, present, and future conditions. The approaches used to unify and harmonize diverse data sets and open-source models will provide a roadmap for the broader community to replicate and extend to other water resource issues and regions.

Evidence based on sparse tree-ring data suggests a severe sustained drought occurred in the 2nd century CE that could have rivaled medieval period droughts in the Colorado River basin (Gangopadhyay et al. 2022). Most of these tree-ring data have been used in gridded drought reconstructions (Cook et al., 2010) which extend back to 1 CE over an area that includes the intermountain western US. However, the 2nd century drought has not been highlighted in prior studies given the sparseness of the data available for this time period. A new reconstruction of Colorado River flow based on these data documents a notably severe and sustained drought over much of the 2nd century (Gangopadhyay et al. 2022). While this reconstruction suggests that the drought exceeds the severity and duration of any drought in the past 2000 years, a complete assessment of the 2nd century drought is challenging due to the sparseness of data. In this poster presentation, we describe the tree-ring data available, along with other proxy data that provide evidence for the 2nd century drought and support its severity. In our conclusions, we discuss outstanding questions and thoughts for further work.

Important progress has been made in recent years in characterizing surface soil moisture (SSM) at regional scales, through remote sensing estimates and the implementation of new in situ networks. Each of these sources of information has intrinsic features, such as the dynamic range of the SSM and the temporal frequency of acquisition. Another relevant factor is the period of data availability. Improving the knowledge of the limitations and biases of these features is crucial to increase the potential and the consistency of data sources validations. As a case of study we considered an agricultural area in the Argentinean Pampas, characterized by a sub-humid climate with a marked seasonal dynamic. It also holds a synchronized cropping rhythm and is subject to flooding and waterlogging that can last from days to months. The features mentioned above and considering that the region is almost devoid of irrigation, offer a natural laboratory that is distinguished by a wide dynamic range of SSM conditions. In this context, we analyze and expose different sources of SSM data gaps over long periods of time, using information from in situ stations and from the SMOS and SMAP satellite systems, during 2015-2019. We found SMAP data gaps resulting from the filtering of high SSM signals that are not spurious but typical for this flood-prone region. Reports from national institutions and comparison with other data sources allowed us to identify that high soil water content in the same period in which the data gaps occurred. In a different way, the SMOS register has a low-frequency range of data due to radio frequency interference over the study area. This data gap occurs during a long-anomalously wet period and it is relevant to take it into account when analyzing SMOS data for the full period. Our study shows the importance of using multiple sources of information and the relevance of examining the availability of data.

Covid- 19 dominantly impacted the Indian agricultural sector. During the period of COVID-19 the southwest monsoon covered a major part of the country, thus resulting in an increase of 9 percent coverage in rainfall than the usual average period. Due to the good amount of rainfall the area under cultivation during the kharif season stood above 4.8% than the previous year. During, the initial lockdown period the agriculture has not been much affected and an increase in migration resulted an increase in people employed in agriculture. Through regression analysis the relationship between the yield and rainfall has been determined. The R2 values have been calculated and the spatial relationship between them has been established. Regions with higher R2 values have been found to be more dominantly affected by Covid-19, though in certain areas strong R2 has shown a weaker spatial relationship owing to certain other factors and policies taken by the Government. Therefore, regression analysis can be used as a suitable method to study the relationship of rainfall and agricultural yield during Covid-19. Keywords: Agriculture, Regression Analysis, Spatial relationship, Rainfall, Covid-19.

Since first diagnosed in the early 1990s, chronic kidney disease of unknown etiology (CKDu) has markedly increased in the North Central Province in the dry zone of Sri Lanka. CKDu has been identified as a global health issue in more than a dozen countries in Asia, South America, and the Middle East. It has been reported that out of these countries, Sri Lanka is the most affected, with the highest cases of CKDu patients and mortality rates. In Sri Lanka, the disease primarily affects male paddy (rice) farmers from low socioeconomic levels. A major river diversion scheme completed in the 70s feeds water from wet zones to ancient tanks that rely on rainwater only. The drinking water for the CKDu affected farming communities comes from the irrigation canals, shallow regolith water table aquifers recharged by canal seepage and precipitation, and deep-bored wells. Many contributing factors and hypotheses have been presented and discussed in the literature. Out of these multiple factors, the suspected environmental exposure pathways are through water (potable water and food) and air (unprotected pesticide spraying). Extensive data on water quality have been collected to develop, test, and support hypotheses on the role of water on the disease. However, no systematic investigations have been conducted to identify, study and analyze how pathways develop through the water storage and distribution systems from sources to the receptors where human exposure occurs. This study proposes a systems-based framework to conduct such analysis using numerical models of the intergraded surface and subsurface system. The models will simulate the fate and transport of naturally occurring toxins and agrichemicals and their geo-bio-chemicals transformation products. These models should incorporate characterization parameters of the surface water storage and distribution system and hydrogeologic data for shallow and deep aquifers, water quality data, epidemiological data, and climate drivers. Innovations methods to use the downscaled climate and regional hydrological model simulations to evaluate exposure pathways at local scales (e.g., villages) under different climate scenarios.

Human pressures on the coastal zones and oceans have increased considerably in the last decades. Human activities constitute the greatest threat to the coastal and marine environment, generating considerable quantities of plastic waste. Currently, it is widely recognized that the increase of marine-related activities has adversely affected the coastal environment as well as the associated ecosystems. Our study focuses on marine litter and specifically on the floating part of it which is frequently composed of plastic materials. Floating litter tends to accumulate on beach-dune ecosystems, already characterized by multiple anthropogenic pressures and environmental factors. In addition, litter items may be trapped by coastal dune vegetation or saltmarsh. Successively, the degradation of marine litter will cause the entering of secondary microplastics. Most of the previous studies are based on monitoring activities and aim to identify the origin and destination of litter in order to manage the fate and transport issues. Therefore, it is important to develop modeling and monitoring tools to detect and prevent marine debris dispersal in coastal environments. We applied field sampling and UAVs (Unmanned Aerial Vehicles) survey over a complex geomorphic set up in the Po River Delta (Italy). Our field data are implemented into a high-resolution hydro-morphodynamic numerical model for validation. Then, we are able to project into different scenarios of plastic debris accumulation in the coastal zone. Our preliminary results show an accumulation of floating debris in coastal dunes vegetation mainly driven by alongshore currents and wave set up in the nearshore area. Then, wind-dominated directions and magnitude disperse plastic debris in embryo dunes and back-barrier marshes. Specific cleaning operations are therefore needed. Considering that coastal management scenarios and decisions rely on numerical models that can predict best practices for coastal sustainability, our results might help local agencies and stakeholders to manage coastal environments.

Land evapotranspiration (ET) and lake evaporation are important water budget components, representing the main processes of energy and water exchange between the earth and the atmosphere, and thus can influence the regional-scale hydrological cycles. Based on several long-term and comprehensive land-atmosphere interaction measurements over the Tibetan Plateau, the total amounts of land ET and lake evaporation are estimated by a combination of satellite products and meteorological data, the results show that: (1) the total ET amount has an average annual value of 1.238±0.058×103 km3. The trends of annual ET amount show high variable in spatial distributions, with an increasing trend in the east plateau and a decreasing trend in the west plateau. (2) As for the lake surface, lake ice phenology are clearly presented by MODIS 8-day snow cover products, and they show large spatial variability in the duration of ice-free season. The estimated Bowen ratio and evaporation amounts show acceptable accuracies, and display opposite spatial distributions, with the latter being higher in the southern part than in the northern part. On the TP, a lake with a higher elevation, a smaller area and a larger latitude mostly corresponds to a shorter ice-free season (a lower total net radiation), a larger Bowen ratio and finally a lower evaporation amount. The multi-year average evaporation amounts are listed, with the total water evaporated from lake surface being approximately 29.4±1.2 km3 year-1 for the studied 75 lakes and 51.7±2.1 km3 year-1 for all Plateau lakes included. (3) To further explore the land/lake-atmosphere interaction processes in detail over data-limited regions of the TP and supported by the “Third Pole Environment (TPE) program, 16 comprehensive observation and research stations have been constructed over all kinds of landscapes and in different regions of the TP in 2021. These data have provided significance for future research on plateau- and regional-scale water budget, hydrological cycle and water resources management.

Because climate change is both a physical and social phenomenon, personal experience has been considered the first step to entail how individuals perceive climate change risk and which actions can be promoted to reduce their vulnerability. Considering that agriculture is affected by climate change in several ways, farmers can provide first-hand observations of climate change impacts and suggest better adaptation options. However, modeling farmers’ behavior is a non-trivial task: personal experience is well recognized as a complex non-linear, multi-variate process due to the high heterogeneity and uncertainties in human cognition and decision-making processes. Furthermore, individual understandings of climate change are always contextualized within broader considerations, meaning that farmers are not ‘blank slates’ receiving information about climate change, but that information is always and inevitably filtered through values and worldviews. Despite the burgeoning of research on climate change, information about farmers’ awareness and risk perception is not geographically homogenized and varies substantially among countries and regions. For example, studies from Global North regions are scarce and emphasize how farmers characterize themselves rather than how they perceive and react to climate change. Drawing on farmers’ surveys in the Lombardy region (Italy), we provide an empirical study to pre-test the triple-loop analysis of farmers’ behavior regarding climate change: awareness, perceived impacts, and adaptation measures and barriers. Applying descriptive statistics and considering socio-economic data and farm characteristics, we address two main research questions: 1) What are farmers’ perceptions of climatic impacts and which responses do they promote? 2) How do personal experience and attitude change is conditioning farmers’ adaptation capacity? Obtained results from accurate bottom-up knowledge on farmers’ behavior may increase policy-makers’ and managers’ understanding of climate change and re-think local policies, which is essential to address agricultural risks in climate change hotspots.

River systems represent important drivers of carbon loading, utilization and storage. However, underlying controls of hydrological transport and biological degradation on fluvial dissolved organic carbon (DOC) have yet to be revealed. Here, we explored spatiotemporal variability of DOC concentrations, components and sources, as well as its biodegradability in a headwater tributary of the Yangtze. We found that temporal rainfall stimulated terrestrial inputs and increased terrigenous DOC abundance. Hydrological transport was accompanied by biological generation and utilization of DOC, resulting in reduced labile components and accumulated recalcitrant components from tributaries to the main stem. Biodegradable DOC (BDOC) notably responded to temperature gradients over a 56-day laboratory incubation. Riverine DOC component, molecular weight and source highly predicted its biodegradation. Particularly, partial refractory (ultraviolet humic-like) fractions contributed to biological degradation of DOC, which was incompletely degraded from high-molecular to low-molecular weight compounds. The findings hope to supplement a new understanding of carbon fate under global change.

Forest management can enhance forest resiliency against natural disturbances such as fire, drought, or disease. Mechanical thinning, followed by a prescribed burn, is a useful technique to achieve a desired forest structure, usually maximizing large tree basal area or decreasing fuel loads, meant to protect against wildfire or reduce water stress in the western US. Changing forest structure can alter ecosystem function by reducing competition and exposing soil, modifying microclimates and creating suitable conditions for shrubs and grasses to encroach. Typically, forest treatments are expected to make the remaining trees more productive through competitive release, and an open canopy helps the understory to thrive. This enhanced plant water use often contradicts the expected result of increased streamflow following thinning. In mountainous terrain, water yield is further complicated by hillslope-scale processes of subsurface lateral flow and groundwater recharge. This research seeks to understand how management-derived forest structure influences hillslope-scale forest regrowth and water yield. We apply a spatially-distributed ecohydrologic model (RHESSys) to an experimental hillslope in the Sierra Nevada, CA. We incorporate multi-temporal Lidar observations and U.S. Forest Service Forest Inventory & Analysis (FIA) survey data to estimate post-thinning regrowth in treated plots in the watershed, which is used to verify RHESSys accuracy of vegetation regrowth. Then, we run long-term virtual thinning experiments to understand how the combination of thinning and prescribed burns in upslope and riparian sites separately and concurrently influences regrowth and water fluxes in these sites. We expect that an intermediate forest density will yield the most co-benefits in terms of carbon sequestration and water yield. However, these patterns will likely be modified along a hillslope, such that riparian forest stands will be less sensitive to the competitive release that thinning provides, whereas dense upslope forests will be highly sensitive to treatment since they are more water-limited. Water yield is likely to be confounded by multiple factors, including topography, whether a burn follows thinning to remove understory fluxes, and interactions between upslope thinning and processes of lateral flow and groundwater recharge when increased riparian water use compensates for additional upslope subsidies.

In this presentation, we highlight undergraduate research approaches and projects at Smith College, the largest college for women in the United States. We share lessons learned along with current challenges to spark conversation and improvement. We comprise a hydrologist (Guswa), environmental engineer (Ismail), and aqueous geochemist (Rhodes), and we investigate the effects of landscape, land management, and natural infrastructure on water quality and water resources. Rhodes and her students carryout field work and laboratory analyses to determine the impacts of development on water chemistry, and a recent project investigates the fate and transport of road salt in a calcareous fen in western Massachusetts. Ismail and her students conduct laboratory experiments to assess the efficacy of filter-feeding organisms to improve water quality in natural systems. A recent project determined how environmental conditions affect the uptake of bacteria by zooplankton. Guswa and his students use models to understand the interactions among climate, landscape, and water resources, and a recent project explores the effects on peak flows of a set of plausible land-use futures for New England in 2060. As undergraduates, students join these projects with limited relevant coursework and research experience. We find that undergraduate engagement is best facilitated by activities that are skill- or technique-based (such as making careful measurements) rather than those based on a deep understanding of theory. Additionally, multiple scales of involvement (e.g., newer students attending group meetings and more senior students designing experiments and serving as peer mentors) allow students to explore potential interests and possibly persist to richer levels of involvement.

Flash droughts have recently gained significant attention due to their severe economic and ecological impacts. Despite extensive and growing research on flash drought processes, predictability, and trends, there is still no standard quantitative definition that encompasses all flash drought characteristics and pathways. This has motivated efforts to define, inventory, monitor, and forecast flash drought events. In our recent studies of flash droughts over the United States, we have introduced the Soil Moisture Volatility Index definition (SMVI) to inventory the onset dates and severity of flash across the Contiguous United States (CONUS) for the period 1979-2018. Post to an extended evaluation and comparison to other flash drought definitions and independent vegetation and crop datasets for seminal flash drought events, the SMVI has proved effectiveness in capturing flash drought onset in both humid and semi-arid regions. Using our SMVI inventory of flash droughts, we examine and classify flash droughts events based on multiple land surface and atmospheric conditions that may represent predictable drivers using a K-means-based clustering methodology. We found that there are three distinct classes of flash drought that can be diagnosed in our inventory. The first defined class of events are the “dry and demanding” droughts, showing high anomalies of evaporative demand and low soil moisture levels; The second are “evaporative” events, which develop under conditions of high demand and when elevated evapotranspiration accelerates soil drying, and a third class that we refer to as “stealth” events, which may be challenging to predict based on precursor atmospheric conditions due to the lack of a clear atmospheric signal with the observed modest anomalies. The contrasting meteorological and surface process signatures of the three classes do, however, indicate that events identified as “flash drought” using a reasonable definition, including events that have been widely reported as seminal flash droughts, represent a diversity of onset and intensification processes. Our results suggest that recognizing this diversity is critical to advance our understanding and ability to predict these events.

Land surface features such as elevation, soils, land use, and vegetation fluctuate on scales ranging from millimeters to hundreds of kilometers. The state of the land surface and many hydrological processes vary accordingly. Land surface temperature (LST) is a crucial factor determining the interactions between the land surface and the atmosphere (i.e., energy, water, and carbon fluxes). Decades of global satellite remote sensed LST fields are now available, constituting an unprecedented opportunity to understand better the factors influencing hydrological variability from regional to global scales. An important under-researched aspect regarding variability, at least over continental extents, is determining the scales for which hydrological variations are spatially and temporally related. These scales would serve as indicators for the required time and spatial resolution for observational systems. This presentation will address this gap in understanding across scales through a comprehensive analysis of spatial and temporal correlation lengths of LST across the contiguous United States (CONUS). Correlation lengths (CLs) are measures of the stationarity of a property distribution both in space and time. They reveal the scales of variability for fields thus, contributing to estimating the stationarity of the property. Temporal correlation lengths (tCLs) express the property changes in time for a fixed location, providing a measure of the persistence or variability of the time series. On the other hand, spatial correlation lengths (sCLs) depict the spatial patterns of the property over a predefined area by representing the distance for which variations are spatially related. As part of our evaluation, we will analyze derived fields of tCLs and sCLs for the ~2x2 km2 GOES-16 LST hourly product over CONUS. A 0.25-degree regular grid over CONUS will be defined, and an hourly time step between 2017 and 2021 will be used for the analysis. The obtained CLs will be assessed in terms of the time of the day and season. Additionally, we propose a comparison of well-known spatiotemporal influencing factors of LST such as land cover, surface thermal properties, topography, incoming solar radiation, and meteorological conditions.

Water is scarce in semi-arid and arid areas where urban irrigation consumes a large portion of city water. It is important to manage and conserve water properly to meet the growing demand, especially in the summer season. The purpose of this research is to identify the major contributors to streamflow in the semi-arid Denver metropolitan area, CO, USA, and analyze the temporal variation of streamflow sources using two-year data. In this study, water-stable isotopes (δ18O and δ2H) were used as the tracer to identify the contribution of different sources such as tap water, precipitation, or irrigation water in Denver urban streamflow. Stream and tap water were sampled every other week and precipitation samples were collected once a month. There were 13 urban and 6 grassland streams, and tap water was collected from 6 different water providers. The USGS real-time streamflow data and BaseflowSeparation function in R package ‘EcoHydRology’ were used to select the baseflow condition in the streams for sampling. Picarro L2130i Laser Water Isotope Analyzer was used for oxygen and hydrogen isotope (δ18O and δ2H) analysis of stream, tap, and precipitation water. Results showed that precipitation samples were heavier in earlier summertime than the late summer and 2019 showed greater variability than 2021. Tap samples showed temporal and spatial variability in δ18O and δ2H values. Less variability in tap isotopic data could be observed in 2021 than in 2019. Centennial Water and Sanitation District showed a decreasing trend. In the δ18O vs. δ2H plot, stream and tap water followed local meteoric water line (LMWL) and global meteoric water line (GMWL) well. But precipitation sample exhibited a slight deviation from the LMWL and GMWL. The similar isotopic range in tap water and stream water supported that tap water was the main source of water during the summertime. In the future, the percent contribution of different sources will be evaluated. Furthermore, the effect of reduced urban irrigation by using efficient irrigation, landscape, or conservation techniques will be analyzed to achieve water security for sustainable urban development.

The Maranhão reservoir (~180 hm3) is mainly used for energy production and irrigation of 15360 ha in the Sorraia Valley, in southern Portugal. The necessity to save water in the Autumn/Winter rainfall season to be used for irrigation of Spring/Summer crops creates a sensible management situation regarding the control of floods and their impact on downstream areas in years with extreme precipitation events. Streamflow forecast is then essential to improve the reservoir’s management regarding water storage. This study addresses the estimation of the daily streamflow in the watershed draining to the Maranhão reservoir (2311 km3) following two different approaches. Firstly, the physically based hydrological model, MOHID-Land, was calibrated/validated for estimating daily streamflow in the study area following physically processes and using a finite volume approach, which require considerable amount of input data. Secondly, a data-driven model composed of an artificial neural network (ANN) was used with the same purpose. This ANN model was selected from a previous work where different models (multi-layer perceptron, convolutional neural network, and long short-term memory model) were tested with different scenarios for the combination of input variables. Both models were optimized considering the observed streamflow in the Ponte Vila Formosa station, which drains 30% of the Maranhão watershed, being after applied to the entire domain. The results for the Maranhão watershed were then validated considering a mass balance considering the reservoir’s outflow, level, and water consumed values. The ANN model had a better performance predicting streamflow than the physically based model, and with less calibration effort. However, the physically based model can give much more information about the entire system. Lastly, it is expected that a physically based model can correctly estimate the streamflow for extreme events not considered in the calibration and validation datasets, but for the ANN models this question should be carefully addressed since data-driven models are event-based.

Most remote sensing-based surface energy balance (SEB) models are limited by data availability and physical constraints to fully capture the non-linear and temporally varying nature of atmospheric, biophysical, and environmental controls on evapotranspiration (ET). As such, currently, no single SEB model is considered to work best under all conditions particularly in irrigated croplands where surface moisture conditions could change dramatically in a short amount of time. Hence, irrigation water management based on a single remotely sensed ET model is often required to cope with model limitations and data latency issues, which could lead to unsustainable and unreliable accounting of water use over time. The recent inception of ensemble-based ET modeling takes the advantage of the strengths of the several SEB models under different conditions and is found to perform better as compared to an Individual model. Yet, challenges remain in how high-temporal ET outputs from different models are accurately assembled in a way that yields the most reliable estimates of ET across any environmental and surface conditions. Specifically, existing simple or Bayesian average and machine learning-based ensemble approaches have not been able to optimally utilize the comprehensive suite of existing SEB models and the availability of multiple remotely sensed datasets. Here, we discuss the utility of convolutional neural networks (CNNs) to assemble the outputs from a host of SEB models that can robustly capture the non-linear dynamics of ET under all conditions. We will also discuss the advantage and potential limitations of using the CNN-based ensemble ET modeling framework with respect to the individual, simple or Bayesian average, and other machine learning approaches and their implications for use in allocating water use across critically dry regions. Several ensemble models will be trained using eddy covariance flux data globally and will be evaluated based on their ability to estimate ET from MODIS and Landsat sensors with both individual and fused products and minimal weather inputs. The results can provide useful insights into how multiple datasets and SEB models could be optimally utilized to accurately monitor crop water status and support sustainable water resource management in drylands.