A coupled snow physics-radiative transfer forward-inversion modeling system was applied over snow-covered terrain in Grand Mesa and the Senator Beck Basin, Colorado, USA to estimate vegetation contributions to the total backscatter from the ground-snow-vegetation system from dual-frequency airborne SnowSAR (X- and Ku-bands) and Sentinel-1 C-band measurements. A simplified but comprehensive first-order microwave emission model (MEMLS-V) was iteratively inverted by a global optimizer using simulated annealing to retrieve unknown parameters and backscatter components from double-bounce, snowpack volume, and snow-ground interface. The retrieved parameters demonstrate high correlation with the observed SnowSAR signal dynamics tied to vegetation and snowpack heterogeneities. This suggests that the forward-inversion system accounting for complex multiple scattering within the ground-snow-vegetation system reliably regulated compensation effects of vegetation and backscatter at the snow-ground interface. The findings have useful practical implications for retrieving large-scale SWE in the northern hemisphere boreal forests from satellite-based radar measurements.

A physical-statistical framework to estimate Snow Water Equivalent (SWE) and Snow depth (SD) from SAR measurements was implemented and applied to SnowSAR flight-line data collected during the SnowEx’2017 field campaign in Grand Mesa, Colorado, USA and averaged to 90 m resolution. The physical (radar) model is used to describe the relationship between snowpack conditions and volume backscatter. The statistical model is a Bayesian inference model that seeks to estimate the joint probability distribution of volume backscatter measurements, SWE and SD and physical model parameters. To reduce the number of physical parameters, the snowpack is represented by two layers only. Retrievals compare well with pit observations with good performance in deep snow and residual errors less than 8% for SnowSAR incidence angles > 30°.

Snow water equivalent (SWE) is an important characteristic of a terrestrial hydrological cycle that needs to be retrieved in any global snow satellite mission. Many retrieval algorithms have been proposed based on microwave backscattering of snow packs. And X and Ku bands have been a focus on many of these past and future missions. In this paper we analyse the airborne X(9.6 GHz) and Ku (17.2 GHz) band data of the SnowSAR 2017 campaign and the University of Massachusetts InSAR Ku (13.3 GHz) band data using the bi-continuous dense media radiative transfer (DMRT) model. In-situ measurements of density, temperature and specific surface area (SSA) from the snow pits are used as physical parameters and are used in estimating the numerical parameters ($\zeta$) and$b$which characterizes the model. The background effects such as rough surface scattering are also removed from the airborne data and only the volume scattering is analyzed. Overcoming limitations in other models such as the sticky sphere model, the bi-continuous media model gives a more realistic representation of snow microstructure and has a weaker frequency dependence.

Sentinel-1 C-band SAR observations over the high-altitude Senator Beck Basin in Colorado show spatial patterns strongly modulated by local topography. An uncalibrated coupled snow hydrology-radiative transfer model driven by downscaled atmospheric reanalysis data (30-m, 15-min) was applied to simulate the spatial and temporal evolution of the backscattering behavior of the seasonal snowpack without vegetation. Model simulations were corrected to capture the impact of complex topography on the viewing geometry. The moisture-attenuated background backscattering field was estimated from a small number of averaged summertime measurements under snow-free conditions. The comparison between the results shows good agreement (25°). Localized higher discrepancies at lower elevations and relatively flat terrain can be unambiguously attributed to underestimation of snowfall in the reanalysis data.

An ensemble of land surface models and forcing data was developed to assess variability in SWE estimation over North America. In this study, the ensemble output was used to assess how SWE uncertainty impacts streamflow estimation. The analysis was conducted by major basins of North America over the 2009-2017 time period.

This study investigates the sensitivity of multispectral reflectivity to changing snow correlation lengths. Mätzler's ice-lamellae radiative transfer model was implemented and tested to evaluate the reflectivity of snow correlation lengths at multiple frequencies from the ultraviolet (UV) to the microwave bands. The model reveals that, in the UV to infrared (IR) frequency range, the reflectivity and correlation length are inversely related, whereas reflectivity increases with snow correlation length in the microwave frequency range. The model further shows that the reflectivity behavior can be mainly attributed to scattering rather than absorption for shallow snowpacks. The largest scattering coefficients and reflectivity occur at very small correlation lengths (~10-5m) for frequencies higher than the IR band. In the microwave range, the largest scattering coefficients are found at millimeter wavelengths. For validation purposes, the ice-lamella model is coupled with a multilayer snow physics model to characterize the reflectivity response of realistic snow hydrological processes. The evolution of the coupled model simulated reflectivities in both the visible and the microwave bands is consistent with satellite-based reflectivity observations in the same frequencies. The model results are also compared with colocated in situ snow correlation length measurements (Cold Land Processes Field Experiment 2002-2003). The analysis and evaluation of model results indicate that the coupled multifrequency radiative transfer and snow hydrology modeling system can be used as a forward operator in a data-assimilation framework to predict the status of snow physical properties, including snow correlation length.

The microwave response of an idealized snowpack is evaluated for a change of the liquid water content$(LWC)$ranging from 0.0$ \hbox{m}^{3}\ \hbox{m}^{-3}$to 0.1$\hbox{m}^{3}\ \hbox{m}^{-3}$, a realistic range for a dry to a wet snowpack transition. Next, the microwave radiometric behavior of a snowpack as a function of$LWC$is investigated using a coupled snow hydrology-microwave emission model that consists of a direct combination of multi-layer snow hydrology model and a forward model of microwave emission based on the simulated multi-layer snowpack. The microwave response during the transition is characterized by an initial increase of the brightness temperature$(Tb)$followed by a monotonic attenuation of$Tb$with a linear increase of$LWC$. Thus, the microwave response$Tb$to the increase in$LWC$exhibits a convex shape. The early amplification of$Tb$is caused by a sharp increase of the absorption coefficient attributed by a relatively small amount of$LWC$within the snowpack. This peak of$Tb$can be explained by the decrease in layer reflectivity and transmissivity in the Microwave Emission Model of Layered Snowpacks. The decrease of snow layer transmissivity indicates that the snowpack operates as an opaque medium in the microwave spectrum for small values of$LWC$. However, when$LWC$continues to increase, the increase of interface reflectivity at the atmosphere-snowpack interface begins to suppress the increasing$Tb$trend. As a result, the$Tb$subsequently decreases because the interface transmissivity, a complement of reflectivity, decreases asymptotically. This arched (convex) behavior of the$Tb$signal with increasing$LWC$is also exhibited by the Special Sensor Microwave Imager and the ground-based microwave radiometer observations concurrent with the presence of$LWC$in the simulated snowpack. For the validation of the$Tb$response to the$LWC$, simulations using a coupled snow hydrology-microwave emission model are compared against both passive microwave satellite and ground-based radiometer observations during the 2002–2003 Cold Land Processes Experiment (CLPX). Two Meso-cell Study Areas from CLPX are selected to evaluate the coupled model performances of$Tb$and snowpack physical properties in response to changes in$LWC$.

The objective of this work is to develop a framework for monitoring snow water equivalent (SWE) and snowpack radiometric properties (e.g., surface emissivity and reflectivity) and microwave emissions in remote regions where ancillary data and ground-based observations for model calibration and/or data assimilation are lacking. For this purpose, an existing land surface hydrology model (LSHM) with single-layer (SL) snow physics was coupled to a microwave emission model (MEMLS). The coupled model (MLSHM-SL) predicts microwave emissions at various frequencies and polarizations as well as snowpack radiometric properties (e.g., emissivity) based on snowpack density, temperature, snow depth, and volumetric liquid water content simulated by the hydrology model with atmospheric forcing obtained from either observations, or the analysis of weather forecasts. The MLSHM-SL was evaluated in prognostic observing system simulation (OSS) mode for two case-studies: 1) a multi-year simulation of snowpack radio-brightness behavior at Valdai, Russia compared against Scanning Multichannel Microwave Radiometer (SMMR) observations at three frequencies (18, 21, and 37 GHz, V, and H polarizations) over six years, 1978-1983; and 2) an intercomparison of simulated and observed brightness temperatures for the Special Sensor Microwave/Imager (SSM/I) and the Advanced Microwave Scanning Radiometer-EOS (AMSR-E) during the 2002-2003 snow season as part of the Cold Land Processes Field Experiment (CLPX) in Colorado. In the case of Valdai, the model captures well the mass balance as well as radiometric behavior of the snowpack during both accumulation and melt, with significantly best skill for vertical polarization (10-16 K differences in error statistics as compared to horizontal polarization), particularly in the winter season January-March (dry snow conditions). Larger biases were detected for intermittent snowpack conditions at the beginning of the fall season due to uncertainty in fractional snow cover and snow wetness at the spatial scale of the SMMR. Similar results were obtained for the OSS of SSM/I and AMSR-E for CLPX, though differences between vertical and horizontal polarization error statistics are more modest (~ 2-4 K). Error statistics are lower for AMSR-E V-pol at 19 and 37 GHz. MLSHM-SL predicted snowpack physical properties (bulk snow density and SWE) compare well against CLPX snowpit observations during the accumulation season with residuals smaller than 10% of observed values. Moreover, the MLSHM-SL simulations in full prognostic mode, and without calibration from the beginning through the end of the snow season, are as skillful as MEMLS with specified physical attributes from snow pit observations. This indicates that the MSLSHM-SL can be used independently as a physically based estimator of SWE in remote regions, and in a data-assimilation framework to provide a physical basis to the interpretation of satellite-based observations of snow.

A multilayer formulation of snow hydrological processes implemented in an existing snow hydrology-emission model (MLSHM-ML) was applied in observing system simulation mode (OSS) to two very different climatic and physiographic regions (Valdai, Russia and Colorado, USA) for both wet and dry snow regimes, and over multiple years. The results were evaluated against ground-based observations of snowpack physical properties and microwave radiometric observations from the Scanning Multichannel Microwave Radiometer (SMMR), Special Sensor Microwave/Imager (SSM/I), and Advanced Microwave Scanning Radiometer-EOS (AMSR-E) observations at 18-19-, 22-23-, and 36-37-GHz vertical and horizontal polarizations (V-pol and H-pol, respectively). Whereas snow water equivalent (SWE) results are similar to the results obtained with single-layer physics when the snow is dry, the multilayer physics have a better skill at capturing the overall temporal evolution of bulk density, snow temperature, and snow depth during the accumulation season, and at the onset and throughout the melting season. However, snow density profiles overestimate density at the bottom of the snowpack, consistent with the lack of an explicit representation of depth hoar in the rearrangement of mass and grain size distribution in the snowpack. Regarding the radiometric behavior, the multilayer SMMR OSS for Valdai shows improved results for nighttime simulations (descending SMMR paths, 11 P.M. LST) and H-pol (~ 3-5 K decrease in error statistics), particularly at 37 GHz. For daytime simulations (ascending SMMR paths, 11 A.M. LST), there are modest improvements at 18 (~ 1 K) and 37 GHz (~ 2-3 K) for H-pol, and generally loss of skill for V-pol at all frequencies. Systematic improvements at nighttime but not during daytime suggest that surface heterogeneities, including subgrid scale variability of transient melting, play an important role on surface emissivity. This is the case for cold land process experiment in Colorado, where spatial variability in fractional forest cover, geology, and complex topography explains the modest differences between the single and multilayer SSM/I OSS for H-pol, whereas significant gains (~ 4-8 K decrease in error statistics) were attained for V-pol at 37 GHz only. For AMSR-E, the multilayer OSS looses skill for H-pol, and only bias and mean absolute error improve for V-pol at all frequencies. These somewhat mixed results suggest that representation of snow stratigraphy alone is not sufficient to improve the OSS ability to describe the nonlinear interactions among hydrologic and electromagnetic processes. Chief among these are the temporal evolution of snow correlation length with depth and the representation of subgrid scale variability constrained by the spatial resolution and inherent uncertainty of the meteorological forcing. Nevertheless, the multilayer OSS improved performance at 37 GHz is an important finding toward reducing ambiguity in the sensitivity of 37-GHz H-pol brightness temperature to SWE in retrieval models.

An L-band transmitter-receiver system wireless sensor to monitor snow accumulation and snow wetness was designed, fabricated, and tested under laboratory conditions. The sensor was designed to operate at 39 discrete frequencies (39 channels) in the 1.00-1.76-GHz frequency range (0.02-GHz increments). Full-system testing of the first-generation system was conducted using commercial attenuators up to 20.0 dB to test the prototypes against design specifications. It was determined that performance was nearly optimal in the 1-1.2-GHz range. Next, snow layers of varying snow wetness were physically modeled under controlled laboratory conditions. This was achieved by adding varying amounts of water to a layer of fixed porosity foam inside a rectangular tank placed above the transmitter. The attenuation and relative phase shift of the RF signal propagating through the experimental “snowpack” and through the laboratory “atmosphere” were subsequently analyzed as a function of volumetric water content equivalent to snow wetness. Under the space and geometry limitations of the laboratory setup, the data show that the single-frequency measurements exhibit high sensitivity for wetness values up to 24%, whereas multifrequency retrieval is necessary for higher liquid water contents. Measurements from a field deployment during snowfall in January 2009 are also presented. The results suggest that there is potential for using the RF sensor to measure cumulative snowfall for short-duration events.

Adaptation to global climate and environmental change in the context of the water-food nexus will require both understanding the nature of change of freshwater resources (e.g. where does precipitation fall, and how much there is?), and the engineering of regional strategies to manage water harvesting and water use (natural storage in aquifers and in the critical soil zone, and manmade alternatives including land-use/land-cover manipulation such as rainfed agriculture), leading to maximal resilience. In a world of increasing population, achieving and maintaining food security is a fundamental challenge for human development in the 21st century. Food security and sustainable agriculture go hand-in hand. The basis for sustainable agriculture is hydroecological resilience, which implies the Integrated Management of Land and Water Resources ("a land-use decision is a water decision", Malin Falkenmark 2001). IMLWR requires systematic monitoring of the pathways by which joint space-time organization patterns of landform, precipitation, recharge (groundwater), distribution and storage (runoff) interact, and ultimately impact the so-called "green water" stocks critical for crop production (i.e. soil moisture in the unsaturated zone that is directly available to meet vegetation photosynthetic needs). IMLWR is ideally suited for a remote-sensing based monitoring and analysis framework. Here, an interpretive study is presented using a wide variety of remote sensing data (clouds, rainfall, and vegetation) from multiple satellite platforms to assess the condition of freshwater stocks (rainfall) and hydroecological resilience in NW India, specifically the state of Punjab. Finally, the notion of hydrometeorological audit is proposed as a strategy for anticipating modes of failure in water resource systems, and to inform policy in the context of sustainable land-water management and food production.

The objective of our research is to develop a framework to perform a systematic and comprehensive analysis of Land Use, Land Cover (LULC) change along the historical record of the terrestrial tracks of hurricanes and tropical storms since the beginning of the earth observation satellite era. Here we present a phenological disturbance filter based on MODIS vegetation indices to detect and characterize the impact of hurricane Isabel, which made landfall on the Outer Banks in North Carolina on 18thSeptember 2003. The results show that woody wetland areas have a pronounced and localized decrease in phenological activity in the two following years, likely due to the disturbance created by flooding, erosion and wind damage. At the regional scale, we identify a relationship between vegetation stress measured by the persistence of below average EVI anomalies, and the frequency of hurricane and tropical storm in the coastal plain of North Carolina. This analysis also shows a direct link between hurricanes and tropical storms (TS) and drought relief in this region.

Neural networks (NNs) are especially useful in exploratory data analysis to uncover and, or elucidate empirical relationships among data. Parameter estimation, the so-called "training" of neural networks is a variation of standard maximum likelihood estimation, whereby the optimal set of model parameters (the NN weights) maximizes the fit to the calibration (training) data set. In our previous applications of neural networks in hydrometeorology, we focused on the development of complex architectures of neural networks adapted to the characteristics of the available data (multisensor, multiresolution mix of ground-based and satellite observations). These architectures consist of large structures of simpler networks built to embody clearly defined hypothesis of functional relationships that are consistent with the underlying physical processes (rainfall and flood forecasting, wind, temperature and moisture profiles in the atmosphere, temporal evolution of cloud and storm morphologies). One challenge we have not addressed previously is how to quantify the uncertainty in NN-based forecasts or estimates. We begin to address this question through the use of Bayesian neural networks (BNNs) for long-lead flood forecasting (18-hours).

The representation of subpixel variability in soil moisture estimates from passive microwave data was investigated through sensitivity analysis and by comparison against the spatial structure of soil moisture fields derived from radar data. This work shows that the subpixel variability not represented in brightness temperature fields is directly associated with the spatial organization of soil hydraulic properties and the spatial distribution of vegetation. The significant implication of this result is that the physical connection between soil moisture estimates at the pixel scale and local values within the pixel weakens strongly as the sensor resolution decreases. Subsequently, the application of scaling and fractal interpolation principles to downscale passive microwave data to the spatial resolution of radar data was investigated as a means to recover spatial structure. In particular, ESTAR soil moisture data was successfully downscaled from 200 to 40 m using only one radar frequency (e.g., L-band). This application suggests that the combined use of active and passive single-band microwave remote-sensing of soil moisture is a viable approach to improve the spatial resolution of soil moisture remote-sensing.