Search

Abstract: Over the last half century, remote sensing has transformed hydrologic science. Whereas early efforts were devoted to observation of discrete variables, we now consider spaceborne missions dedicated to interlinked global hydrologic processes. Furthermore, cloud computing and computational techniques are accelerating analyses of these data. How will the hydrologic community use these new resources to better understand the world's water and related challenges facing society? In this commentary, we suggest that optimizing the benefits of remote sensing for advancing hydrologic research will happen by integrating multidisciplinary and multisensor data, leveraging commercial satellite measurements, and employing data assimilation, cloud computing, and machine learning. We provide several recommendations to these ends.

Microwave brightness temperature observations from the NASA Soil Moisture Active Passive (SMAP) mission and gauge‐based precipitation data over the United States are used to reconstruct the soil water loss function and then historical (1979–2019) hydrological fluxes in the form of evapotranspiration (ET) and drainage (D) are quantified. Over the period of study, with the exception of snowy and hyper‐arid regions, we observe a correlation of R2 > 0.6 between SMAP‐precipitation derived drainage estimates and streamflow measurements from the U.S. Geological Survey (USGS). There is a bias between estimated drainage and USGS streamflow with an underestimation of about 1 mm day−1 in southwest United States to 3 mm day−1 in parts of the eastern United States. SMAP‐derived sensitivities of drainage and ET partitioning with respect to precipitation anomalies are also calculated. In parts of the Great Plains the drainage partitioning exhibits a near‐linear response, while in the southeast United States, the response is nonlinear. Partitioning also is examined for 6 four‐digit hydrologic unit basins wherein year‐to‐year variations in drainage partitioning are shown to be key mediators in translating precipitation anomalies into streamflow anomalies. Observation‐driven drainage and ET estimates are obtained without relying on full hydrologic and Land Surface Models (LSMs). This independence (isolation from model parameterization assumptions) provides a path toward using satellite‐derived landscape hydrological diagnostics to assess hydrologic models and LSMs as well as to guide their further development. Key Points: A large amount of hydrologic information is encoded within satellite remote sensing observations of soil moistureSMAP‐era soil moisture and precipitation information, without reliance on models, can characterize landscape hydrological partitioningHistorical remote sensing and precipitation‐based drainage estimates closely track streamflow measurements across the United States [ABSTRACT FROM AUTHOR]

Land–atmosphere feedbacks occurring on daily to weekly time scales can magnify the intensity and duration of extreme weather events, such as droughts, heat waves, and convective storms. For such feedbacks to occur, the coupled land–atmosphere system must exhibit sufficient memory of soil moisture anomalies associated with the extreme event. The soil moisture autocorrelation e-folding time scale has been used previously to estimate soil moisture memory. However, the theoretical basis for this metric (i.e., that the land water budget is reasonably approximated by a red noise process) does not apply at finer spatial and temporal resolutions relevant to modern satellite observations and models. In this study, two memory time scale metrics are introduced that are relevant to modern satellite observations and models: the "long-term memory" τ L and the "short-term memory" τ S . Short- and long-term surface soil moisture (SSM) memory time scales are spatially anticorrelated at global scales in both a model and satellite observations, suggesting hot spots of land–atmosphere coupling will be located in different regions, depending on the time scale of the feedback. Furthermore, the spatial anticorrelation between τ S and τ L demonstrates the importance of characterizing these memory time scales separately, rather than mixing them as in previous studies. [ABSTRACT FROM AUTHOR]

The partitioning of incident precipitation into evapotranspiration, runoff, drainage, storage change, and hydrologic fluxes depends on the soil moisture state. With the availability of global remotely sensed soil moisture fields, the functional dependence of each flux on soil moisture may be identifiable. In this study we develop an observation‐driven approach to map key hydroclimatology fields using remotely sensed soil moisture and gauge‐based precipitation data only. National Aeronautics and Space Administration's Soil Moisture Active Passive (SMAP) low‐frequency microwave brightness temperature observations and precipitation fields, from the National Centers for Environmental Prediction, are the sole inputs into an adjoint‐state variational estimation framework. Furthermore, the proposed methodology does not rely on micrometeorological information, or land surface models. The approach is flexible by design so that almost any partitioning pattern can result from estimation and corresponding evapotranspiration and drainage fields can be quantified. Three‐year averaged summer season evapotranspiration estimates are compared with available vapor flux at in situ AmeriFlux eddy‐covariance sites. Basin‐averaged drainage over major U.S. hydrologic units is also compared with U.S. Geological Survey streamgages measurements. The remote sensing‐based estimated hydroclimate fields explain about 70% of the variance in the in situ measurements. This exploratory study adds to the body of evidence emerging in literature that a significant amount of hydrologic information is encoded in the dynamic fields of remotely sensed soil moisture. Observation‐driven hydroclimate data fields that are independent of land surface models can provide valuable insights into the state of the water cycle and guide future development of land surface models. Key Points: A significant amount of hydrologic information is encoded in the dynamic fields of surface soil moistureVariational estimation is used to partition precipitation into evapotranspiration and drainage fluxesObservation‐driven maps of evapotranspiration and drainage hydroclimates explain about 70% of their corresponding in situ variance [ABSTRACT FROM AUTHOR]

This study presents an observation-driven technique to delineate the dominant boundaries and temporal shifts between different hydrologic regimes over the contiguous United States (CONUS). The energy- and water-limited evapotranspiration regimes as well as percolation to the subsurface are hydrologic processes that dominate the loss of stored water in the soil following precipitation events. Surface soil moisture estimates from the NASA Soil Moisture Active Passive (SMAP) mission, over three consecutive summer seasons, are used to estimate the soil water loss function. Based on analysis of the rates of soil moisture dry-downs, the loss function is the conditional expectation of negative increments in the soil moisture series conditioned on soil moisture itself. An unsupervised classification scheme (with cross validation) is then implemented to categorize regions according to their dominant hydrological regimes based on their estimated loss functions. An east-west divide in hydrologic regimes over CONUS is observed with large parts of the western United States exhibiting a strong water-limited evapotranspiration regime during most of the times. The U.S. Midwest and Great Plains show transitional behavior with both water- and energy-limited regimes present. Year-to-year shifts in hydrologic regimes are also observed along with regional anomalies due to moderate drought conditions or above-average precipitation. The approach is based on remotely sensed surface soil moisture (approximately top 5 cm) at a resolution of tens of kilometers in the presence of soil texture and land cover heterogeneity. The classification therefore only applies to landscape-scale effective conditions and does not directly account for deeper soil water storage. [ABSTRACT FROM AUTHOR]

Abstract: The soil water content profile is often well correlated with the soil moisture state near the surface. They share mutual information such that analysis of surface‐only soil moisture is, at times and in conjunction with precipitation information, reflective of deeper soil fluxes and dynamics. This study examines the characteristic length scale, or effective depth Δz, of a simple active hydrological control volume. The volume is described only by precipitation inputs and soil water dynamics evident in surface‐only soil moisture observations. To proceed, first an observation‐based technique is presented to estimate the soil moisture loss function based on analysis of soil moisture dry‐downs and its successive negative increments. Then, the length scale Δz is obtained via an optimization process wherein the root‐mean‐squared (RMS) differences between surface soil moisture observations and its predictions based on water balance are minimized. The process is entirely observation‐driven. The surface soil moisture estimates are obtained from the NASA Soil Moisture Active Passive (SMAP) mission and precipitation from the gauge‐corrected Climate Prediction Center daily global precipitation product. The length scale Δz exhibits a clear east‐west gradient across the contiguous United States (CONUS), such that large Δz depths (>200 mm) are estimated in wetter regions with larger mean precipitation. The median Δz across CONUS is 135 mm. The spatial variance of Δz is predominantly explained and influenced by precipitation characteristics. Soil properties, especially texture in the form of sand fraction, as well as the mean soil moisture state have a lesser influence on the length scale. [ABSTRACT FROM AUTHOR]

Soil moisture information is critical for applications like landslide susceptibility analysis and military trafficability assessment. Existing technologies cannot observe soil moisture at spatial scales of hillslopes (e.g., 100 to 102 m) and over large areas (e.g., 102 to 105 km2) with sufficiently high temporal coverage (e.g., days). Physics-based hydrologie models can simulate soil moisture at the necessary spatial and temporal scales, albeit with error. We develop and test a data assimilation framework based on the ensemble Kaiman filter for constraining uncertain simulated high-resolution soil moisture fields to anticipated remote sensing products, specifically NASA's Soil Moisture Active-Passive (SMAP) mission, which will provide global L band microwave observation approximately every 2-3 days. The framework directly assimilates SMAP synthetic 3 km radar backscatter observations to update hillslope-scale bare soil moisture estimates from a physics-based model. Downscaling from 3 km observations to hillslope scales is achieved through the data assimilation algorithm. Assimilation reduces bias in near-surface soil moisture (e.g., top 10 cm) by approximately 0.05 m3/m3 and expected root-mean-square errors by at least 60% in much of the watershed, relative to an open loop simulation. However, near-surface moisture estimates in channel and valley bottoms do not improve, and estimates of profile-integrated moisture throughout the watershed do not substantially improve. We discuss the implications of this work, focusing on ongoing efforts to improve soil moisture estimation in the entire soil profile through joint assimilation of other satellite (e.g., vegetation) and in situ soil moisture measurements. [ABSTRACT FROM AUTHOR]

An observing system simulation experiment is developed to test tradeoffs in resolution and accuracy for soil moisture estimation using active and passive L-band remote sensing. Concepts for combined radar and radiometer missions include designs that will provide multiresolution measurements. In this paper, the scientific impacts of instrument performance are analyzed to determine the measurement requirements for the mission concept. The ensemble Kalman smoother (EnKS) is used to merge these multiresolution observations with modeled soil moisture from a land surface model to estimate surface and subsurface soil moisture at 6-km resolution. The model used for assimilation is different from that used to generate "truth." Consequently, this experiment simulates how data assimilation performs in real applications when the model is not a perfect representation of reality. The EnKS is an extension of the ensemble Kalman filter (EnKF) in which observations are used to update states at previous times. Previous work demonstrated that it provides a computationally inexpensive means to improve the results from the EnKF, and that the limited memory in soil moisture can be exploited by employing it as a fixed lag smoother. Here, it is shown that the EnKS can be used in large problems with spatially distributed state vectors and spatially distributed multiresolution observations. The EnKS-based data assimilation framework is used to study the synergy between passive and active observations that have different resolutions and measurement error distributions. The extent to which the design parameters of the EnKS vary depending on the combination of observations assimilated is investigated. [ABSTRACT FROM AUTHOR]

Land surface data assimilation problems are often limited by the high dimensionality of states created by spatial discretization over large high-resolution computational grids. Yet field observations and simulation both confirm that soil moisture can have pronounced spatial structure, especially after extensive rainfall. This suggests that the high dimensionality of the problem could be reduced during wet periods if spatial patterns could be more efficiently represented. After prolonged drydown, when spatial structure is determined primarily by small-scale soil and vegetation variability rather than rainfall, the original high-dimensional problem can be effectively replaced by many independent low-dimensional problems that can be solved in parallel with relatively little effort. In reality, conditions are continually varying between these two extremes. This is confirmed by a singular value decomposition of the replicate matrix (covariance square root) produced in an ensemble forecasting simulation experiment. The singular value spectrum drops off quickly after rainfall events, when a few leading modes dominate the spatial structure of soil moisture. The spectrum is much flatter after a prolonged drydown period, when spatial structure is less significant. Deterministic reduced-rank Kalman filters can achieve significant computational efficiency by focusing on the leading modes of a system with large-scale spatial structure. But these methods are not well suited for land surface problems with complex uncertain inputs and rapidly changing spectra. Local ensemble Kalman filters are suitable for such problems during dry periods but give less accurate results after rainfall. The most promising option for achieving computational efficiency and accuracy is to develop generalized localization methods that dynamically aggregate states, reflecting structural changes in the ensemble. [ABSTRACT FROM AUTHOR]

Triangulated irregular networks (TIN) can form the basis for multiple-resolution representations in distributed hydrogeomorphic simulations over complex basins. Current methods for deriving TIN meshes depend primarily on surface slope without considering other terrain attributes significant to the watershed response such as the specific basin area. As an alternative, we present a methodology for combining a hydrogeomorphic or landscape index with an unstructured triangulated mesh. Landscape indices provide a concise method for describing steady-state terrain processes by isolating the dominant physical factors. The mesh-generation algorithm results in an adaptive discretization that resembles the spatial pattern of the landscape index with a high resolution retained in areas expected to impact the basin response. We compare the proposed algorithm with a slope-preserving method as a means for initializing the terrain representation in two TIN-based hydrogeomorphic models. Through three case studies in saturation-excess runoff, transport-limited soil erosion and shallow landslide simulation, we assess the distributed model sensitivity to the triangulated terrain algorithm. Model comparisons reveal that the process-based triangulations focus the distributed simulation in regions anticipated via a steady-state index to affect the transient watershed response. [ABSTRACT FROM AUTHOR]

Distributed hydrologic models typically incorporate topographic data through the use of raster-based digital elevation models. The resampling of high-resolution grid data required to effectively use distributed models, however, can result in the distortion of terrain and hydrographic properties. In this study, we present a geographic information system approach for deriving multiple resolution meshes that conserve physiographic features while significantly reducing the number of computational nodes in a distributed hydrologic model. We utilize triangulated irregular networks (TINs) which serve to integrate information on the surface topography, hydrographic features and land surface characteristics into an adaptive representation of a basin. We discuss three approaches for constructing TIN models for hydrologic applications: (1) Traditional, (2) hydrographic and (3) hydrological similarity TINs. We focus on the generation of triangulated terrain models using the concept of hydrological similarity provided through a topographic or wetness index. This new method embeds an estimate of the steady-state hydrologic response directly into the basin terrain model. Through a series of case studies, we demonstrate the advantages of the multiple resolution approaches over a range of terrain characteristics, basin scales and elevation data products. Finally, we discuss the implications of TIN terrain representation for watershed simulation with the TIN-based Real-Time Integrated Basin Simulator model. [ABSTRACT FROM AUTHOR]

This study presents various aspects of the continuous simulation capabilities of a fully-distributed, triangulated irregular network (TIN) hydrologic model. The TIN-based Real-time Integrated Basin Simulator (tRIBS) is calibrated and verified for the Baron Fork at Eldon, Illinois River at Watts, and Blue River at Blue over the period 1993–2000. Computational effort is significantly reduced by simulating complex watersheds using a multiple resolution mesh to represent terrain. Model performance is assessed by comparing streamflow predictions to observations at the basin outlet and interior gauging stations. In addition, simulation results describing the distributed basin response to atmospheric forcing are discussed, including the spatial and temporal variability of runoff, surface soil moisture, evaporative flux, and groundwater table position. By modeling the land-surface water and energy states and fluxes over the computational domain in an efficient manner, the potential for utilizing fully-distributed models at the scales of operational hydrologic forecasting is realized. Through the spatially-explicit approach, high-resolution remote sensing data describing surface properties, topography, rainfall, and soil moisture can be integrated directly into a predictive hydrologic model. A greater degree of physical interpretation of hydrological estimation can thus be added to existing methods of operational forecasting. [Copyright &y& Elsevier]