Evolutionary assimilation of streamflow in distributed hydrologic modeling using in-situ soil moisture data

This study has applied evolutionary algorithm to address the data assimilation problem in a distributed hydrological model. The evolutionary data assimilation (EDA) method uses multi-objective evolutionary strategy to continuously evolve ensemble of model states and parameter sets where it adaptively determines the model error and the penalty function for different assimilation time steps. The assimilation was determined by applying the penalty function to merge background information (i.e., model forecast) with perturbed observation data. The assimilation was based on updated estimates of the model state and its parameterizations, and was complemented by a continuous evolution of competitive solutions.The EDA was illustrated in an integrated assimilation approach to estimate model state using soil moisture, which in turn was incorporated into the soil and water assessment tool (SWAT) to assimilate streamflow. Soil moisture was independently assimilated to allow estimation of its model error, where the estimated model state was integrated into SWAT to determine background streamflow information before they are merged with perturbed observation data. Application of the EDA in Spencer Creek watershed in southern Ontario, Canada generates a time series of soil moisture and streamflow. Evaluation of soil moisture and streamflow assimilation results demonstrates the capability of the EDA to simultaneously estimate model state and parameterizations for real-time forecasting operations. The results show improvement in both streamflow and soil moisture estimates when compared to open-loop simulation, and a close matching between the background and the assimilation illustrates the forecasting performance of the EDA approach.     

Upscaling geochemical reaction rates using pore-scale network modeling

Geochemical reaction rate laws are often measured using crushed minerals in well-mixed laboratory systems that are designed to eliminate mass transport limitations. Such rate laws are often used directly in reactive transport models to predict the reaction and transport of chemical species in consolidated porous media found in subsurface environments. Due to the inherent heterogeneities of porous media, such use of lab-measured rate laws may introduce errors, leading to a need to develop methods for upscaling reaction rates. In this work, we present a methodology for using pore-scale network modeling to investigate scaling effects in geochemical reaction rates. The reactive transport processes are simulated at the pore scale, accounting for heterogeneities of both physical and mineral properties. Mass balance principles are then used to calculate reaction rates at the continuum scale. To examine the scaling behavior of reaction kinetics, these continuum-scale rates from the network model are compared to the rates calculated by directly using laboratory-measured reaction rate laws and ignoring pore-scale heterogeneities. In this work, this methodology is demonstrated by upscaling anorthite and kaolinite reaction rates under simulation conditions relevant to geological CO₂ sequestration. Simulation results show that under conditions with CO₂ present at high concentrations, pore-scale concentrations of reactive species and reaction rates vary spatially by orders of magnitude, and the scaling effect is significant. With a much smaller CO₂ concentration, the scaling effect is relatively small. These results indicate that the increased acidity associated with geological sequestration can generate conditions for which proper scaling tools are yet to be developed. This work demonstrates the use of pore-scale network modeling as a valuable research tool for examining upscaling of geochemical kinetics. The pore-scale model allows the effects of pore-scale heterogeneities to be integrated into system behavior at multiple scales, thereby identifying important factors that contribute to the scaling effect.     

Testing of observation operators designed to estimate profile soil moisture from surface measurements

The accurate estimation of profile soil moisture is usually difficult due to the associated costs, strong spatiotemporal variability, and nonlinear relationship between surface and profile moisture. Here, we used data sets from the Soil and Climate Analysis Network to test how reliably observation operators developed based on the cumulative distribution function matching method can predict profile soil moisture from surface measurements. We first analysed how temporal resolution (hourly, daily, and weekly) and data length (half year, 1 year, 2 years, and 4 years) affected the performance of observation operators. The results showed that temporal resolution had a negligible influence on the performance of observation operators. However, a leave‐one‐year‐out cross‐validation showed that data length affected the performance of observation operators; a 2‐year interval was identified as the most suitable duration due to its low uncertainty in prediction accuracy. The robustness of the observation operators was then tested in three primary climates (humid continental, humid subtropical, and semiarid) of the continental United States, with the exponential filter employed as an independent method. The results indicated that observation operators reliably predicted profile soil moisture for most of the tested stations and performed almost equally well as the exponential filter method. The presented results verified the feasibility of using the cumulative distribution function matching method to predict profile soil moisture from surface measurements.     

Determination of bare surface soil moisture from combined temporal evolution of land surface temperature and net surface shortwave radiation

Land surface soil moisture (SSM) is an important variable for hydrological, ecological, and meteorological applications. A multi‐linear model has recently been proposed to determine the SSM content from the combined diurnal evolution of both land surface temperature (LST) and net surface shortwave radiation (NSSR) with the parameters TN (the LST mid‐morning rising rate divided by the NSSR rising rate during the same period) and t_d (the time of daily maximum temperature). However, in addition to the problem that all the coefficients of the multi‐linear model depend on the atmospheric conditions, the model also suffers from the problems of the nonlinearity of TN as a function of the SSM content and the uncertainty of determining the t_d from the diurnal evolution of the LST. To address these problems, a modified multi‐linear model was developed using the logarithm of TN and normalizing t_d by the mid‐morning temperature difference instead of using the TN and t_d. Except for the constant term, the coefficients of all other variables in the modified multi‐linear model proved to be independent of the atmospheric conditions. Using the relevant simulation data, results from the modified multi‐linear model show that the SSM content can be determined with a root mean square error (RMSE) of 0.030m³/m³, provided that the constant term is known or estimated day to day. The validation of the model was conducted using the field measurements at the Langfang site in 2008 in China. A higher correlation is achieved (coefficient of determination: R² = 0.624, RMSE = 0.107m³/m³) between the measured SSM content and the SSM content estimated using the modified multi‐linear model with the coefficients determined from the simulation data. Another experiment is also conducted to estimate the SSM content using the modified model with the constant term calibrated each day by one‐spot measurements at the site. The estimation result has a relatively larger error (RMSE = 0.125m³/m³). Additionally, the uncertainty of the determination of the coefficients is analysed using the field measurements, and the results indicate that the SSM content obtained using the modified model accurately characterizes the surface soil moisture condition. Copyright © 2012 John Wiley & Sons, Ltd.     

III. Measuring surface soil moisture using passive microwave remote sensing

Soil moisture is one of the few directly observable hydrological variables that has an important role in water and energy budgets necessary for climate studies. At the present time there is no practical approach to measuring and monitoring soil moisture at the frequency and scale necessary for these large scale analyses. Current and developing satellite systems have not addressed this important question. A solution utilizing passive microwave remote sensing is presented here and an optimum system, soil moisture estimation algorithms and a microwave simulation model are described.     

Evaluation of multiple surface soil moisture for Korean regional flux monitoring network sites: Advanced Microwave Scanning Radiometer E,land surface model,and ground measurements

In the past few decades, there have been great developments in remotely sensed soil moisture, with validation efforts using land surface models (LSMs) and ground‐based measurements, because soil moisture information is essential to understanding complex land surface–atmosphere interactions. However, the validation of remotely sensed soil moisture has been very limited because of the scarcity of the ground measurements in Korea. This study validated Advanced Microwave Scanning Radiometer E (AMSR‐E) soil moisture data with the Common Land Model (CLM), one of the most widely used LSMs, and ground‐based measurements at two Korean regional flux monitoring network sites. There was reasonable agreement regarding the different soil moisture products for monitoring temporal trends except National Snow and Ice Data Centre (NSIDC) AMSR‐E soil moisture, albeit there were essential comparison limitations by different spatial scales and soil depths. The AMSR‐E soil moisture data published by the National Aeronautics and Space Administration and Vrije Universiteit Amsterdam (VUA) showed potential to replicate temporal variability patterns (root‐mean‐square errors = 0·10–0·14 m³ m^?3 and wet BIAS = 0·09 ? 0·04 m³ m^?3) with the CLM and ground‐based measurements. However, the NSIDC AMSR‐E soil moisture was problematic because of the extremely low temporal variability and the VUA AMSR‐E soil moisture was relatively inaccurate in Gwangneung site characterized by complex geophysical conditions. Additional evaluations should be required to facilitate the use of recent and forthcoming remotely sensed soil moisture data from Soil Moisture and Ocean Salinity and Soil Moisture Active and Passive missions at representative future validation sites. Copyright © 2011 John Wiley & Sons, Ltd.     

The value of remotely sensed surface soil moisture for model calibration using SWAT

Remotely sensed (RS) data can add value to a hydrological model calibration. Among this, RS soil moisture (SM) data have mostly been assimilated into conceptual hydrological models using various transformed variable or indices. In this study, raw RS surface SM is used as a calibration variable in the Soil and Water Assessment Tool model. This means the SM values were not transformed into another variable (e.g., soil water index and root zone SM index). Using a nested catchment, calibration based only on RS SM and optimizing model parameters sensitive to SM using particle swarm optimization improved variations in streamflow predictions at some of the gauging stations compared to the uncalibrated model. This highlighted part of the catchments where the SM signal directly influenced the flow distribution. Additionally, highlighted high and low flow signals were mostly influenced. The seasonal breakdown indicates that the SM signal is more useful for calibrating in wetter seasons and in areas with higher variations in elevation. The results identified that calibration only on RS SM improved the general rainfall–runoff response simulation by introducing delays but cannot correct the overall routing effect. Furthermore, catchment characteristics (e.g., land use, elevation, soil types, and precipitation) regulating SM variation in different seasons highlighted by the model calibration are identified. This provides further opportunities to improve model parameterization.     

Heated distributed temperature sensing for field scale soil moisture monitoring

Characterizing both spatial and temporal soil moisture (θ) dynamics at site scales is difficult with existing technologies. To address this shortcoming, we developed a distributed soil moisture sensing system that employs a distributed temperature sensing system to monitor thermal response at 2 m intervals along the length of a buried cable which is subjected to heat pulses. The cable temperature response to heating, which is strongly dependent on soil moisture, was empirically related to colocated, dielectric-based θ measurements at three locations. Spatially distributed, and temporally continuous estimates of θ were obtained in dry conditions (θ≤ 0.31) using this technology (root mean square error [RMSE] = 0.016), but insensitivity of the instrument response curve adversely affected accuracy under wet conditions (RMSE = 0.050).     

A land surface soil moisture data assimilation framework in consideration of the model subgrid-scale heterogeneity and soil water thawing and freezing

The Ensemble Kalman Filter (EnKF) is well known and widely used in land data assimilation for its high precision and simple operation. The land surface models used as the forecast operator in a land data assimilation system are usually designed to consider the model subgrid-heterogeneity and soil water thawing and freezing. To neglect their effects could lead to some errors in soil moisture assimilation. The dual EnKF method is employed in soil moisture data assimilation to build a soil moisture data as- similation framework based on the NCAR Community Land Model version 2.0 (CLM 2.0) in considera- tion of the effects of the model subgrid-heterogeneity and soil water thawing and freezing: Liquid volumetric soil moisture content in a given fraction is assimilated through the state filter process, while solid volumetric soil moisture content in the same fraction and solid/liquid volumetric soil moisture in the other fractions are optimized by the parameter filter. Preliminary experiments show that this dual EnKF-based assimilation framework can assimilate soil moisture more effectively and precisely than the usual EnKF-based assimilation framework without considering the model subgrid-scale heteroge- neity and soil water thawing and freezing. With the improvement of soil moisture simulation, the soil temperature-simulated precision can be also improved to some extent.     

Scaling characteristics of spatial patterns of soil moisture from distributed modelling

Characterizing the spatial dynamics of soil moisture fields is a key issue in hydrology, offering an avenue to improve our understanding of complex land surface–atmosphere interactions. In this paper, the statistical structure of soil moisture patterns is examined using modelled soil moisture obtained from the North American Land Data Assimilation System (NLDAS) at 0.125° resolution. The study focuses on the vertically averaged soil moisture in the top 10 cm and 100 cm layers. The two variables display a weak dependence for lower values of surface soil moisture, with the strength of the relationship increasing with the water content of the top layer. In both cases, the variance of the soil moisture follows a power law decay as a function of the averaging area. The superficial layer shows a lower degree of spatial organization and higher temporal variability, which is reflected in rapid changes in time of the slope of the scaling functions of the soil moisture variance. Conversely, the soil moisture in the top 100 cm has lower variability in time and larger spatial correlation. The scaling of these patterns was found to be controlled by the changes in the soil water content. Results have implications for the downscaling of soil moisture to prevent model bias.     

Generation of continuous surface soil moisture dataset using combined optical and thermal infrared images

Surface soil moisture (SSM) is a critical variable for understanding water and energy flux between the atmosphere and the Earth's surface. An easy to apply algorithm for deriving SSM time series that primarily uses temporal parameters derived from simulated and in situ datasets has recently been reported. This algorithm must be assessed for different biophysical and atmospheric conditions by using actual geostationary satellite images. In this study, two currently available coarse‐scale SSM datasets (microwave and reanalysis product) and aggregated in situ SSM measurements were implemented to calibrate the time‐invariable coefficients of the SSM retrieval algorithm for conditions in which conventional observations are rare. These coefficients were subsequently used to obtain SSM time series directly from Meteosat Second Generation (MSG) images over the study area of a well‐organized soil moisture network named REMEDHUS in Spain. The results show a high degree of consistency between the estimated and actual SSM time series values when using the three SSM dataset‐calibrated time‐invariable coefficients to retrieve SSM, with coefficients of determination (R²) varying from 0.304 to 0.534 and root mean square errors ranging from 0.020 m³/m³ to 0.029 m³/m³. Further evaluation with different land use types results in acceptable debiased root mean square errors between 0.021 m³/m³ and 0.048 m³/m³ when comparing the estimated MSG pixel‐scale SSM with in situ measurements. These results indicate that the investigated method is practical for deriving time‐invariable coefficients when using publicly accessed coarse‐scale SSM datasets, which is beneficial for generating continuous SSM dataset at the MSG pixel scale.     

Effects of soil moisture aggregation on surface evaporative fluxes

The effects of small-scale heterogeneity in land surface characteristics on the large-scale fluxes of water and energy in the land-atmosphere system has become a central focus of many climatology research experiments. The acquisition of high resolution land surface data through remote sensing and intensive land-climatology field experiments (like HAPEX, FIFE, and BOREAS) has provided data to investigate the interactions between microscale land-atmosphere interactions and macroscale models. To determine the effect of small scale heterogeneities, the spatially averaged evaporative fraction is analytically derived for spatially variable soil moisture and soil-atmospheric controls on evaporation at low soil moisture. This average evaporative fraction is compared with the evaporative fraction determined using the spatially averaged soil moisture, as if from a lumped, or aggregated, land surface model. Results show that the lumped-model based evaporation will over estimate evaporation during periods of low atmospheric demands (early morning/late afternoon, Winter periods, etc.) and under estimate evaporation during periods of high demand (midday Summer periods.) The accuracy of using ‘effective’ parameters in lumped macroscale models depends on the variability of soil moisture and the sensitivity of the soil-vegetation system to low soil moisture.     

Intercomparison of the JULES and CABLE land surface models through assimilation of remotely sensed soil moisture in southeast Australia

Numerous land surface models exist for predicting water and energy fluxes in the terrestrial environment. These land surface models have different conceptualizations (i.e., process or physics based), together with structural differences in representing spatial variability, alternate empirical methods, mathematical formulations and computational approach. These inherent differences in modeling approach, and associated variations in outputs make it difficult to compare and contrast land surface models in a straight-forward manner. While model intercomparison studies have been undertaken in the past, leading to significant progress on the improvement of land surface models, additional framework towards identification of model weakness is needed. Given that land surface models are increasingly being integrated with satellite based estimates to improve their prediction skill, it is practical to undertake model intercomparison on the basis of soil moisture data assimilation. Consequently, this study compares two land surface models: the Joint UK Land Environment Simulator (JULES) and the Community Atmosphere Biosphere Land Exchange (CABLE) for soil moisture estimation and associated assessment of model uncertainty. A retrieved soil moisture data set from the Soil Moisture and Ocean Salinity (SMOS) mission was assimilated into both models, with their updated estimates validated against in-situ soil moisture in the Yanco area, Australia. The findings show that the updated estimates from both models generally provided a more accurate estimate of soil moisture than the open loop estimate based on calibration alone. Moreover, the JULES output was found to provide a slightly better estimate of soil moisture than the CABLE output at both near-surface and deeper soil layers. An assessment of the updated membership in decision space also showed that the JULES model had a relatively stable, less sensitive, and more highly convergent internal dynamics than the CABLE model.     

Error distribution modelling of satellite soil moisture measurements for hydrological applications

Satellite‐based soil moisture data accuracies are of important concerns by hydrologists because they could significantly influence hydrological modelling uncertainty. Without proper quantification of their uncertainties, it is difficult to optimize the hydrological modelling system and make robust decisions. Currently, the satellite soil moisture data uncertainty has been limited to summary statistics with the validations mainly from the in situ measurements. This study attempts to build the first error distribution model with additional higher‐order uncertainty modelling for satellite soil moisture observations. The methodology is demonstrated by a case study using the Soil Moisture and Ocean Salinity satellite soil moisture observations. The validation is based on soil moisture estimates from hydrological modelling, which is more relevant to the intended data use than the in situ measurements. Four probability distributions have been explored to find suitable error distribution curves using the statistical tests and bootstrapping resampling technique. General extreme value is identified as the most suitable one among all the curves. The error distribution model is still in its infant stage, which ignores spatial and temporal correlations, and nonstationarity. Further improvements should be carried out by the hydrological community by expanding the methodology to a wide range of satellite soil moisture data using different hydrological models. Copyright © 2016 John Wiley & Sons, Ltd.     

Numerical investigation of the impact of uncertainties in satellite rainfall estimation and land surface model parameters on simulation of soil moisture

This study aims at evaluating the uncertainty in the prediction of soil moisture (1D, vertical column) from an offline land surface model (LSM) forced by hydro-meteorological and radiation data. We focus on two types of uncertainty: an input error due to satellite rainfall retrieval uncertainty, and, LSM soil-parametric error. The study is facilitated by in situ and remotely sensed data-driven (precipitation, radiation, soil moisture) simulation experiments comprising a LSM and stochastic models for error characterization. The parametric uncertainty is represented by the generalized likelihood uncertainty estimation (GLUE) technique, which models the parameter non-uniqueness against direct observations. Half-hourly infra-red (IR) sensor retrievals were used as satellite rainfall estimates. The IR rain retrieval uncertainty is characterized on the basis of a satellite rainfall error model (SREM). The combined uncertainty (i.e., SREM + GLUE) is compared with the partial assessment of uncertainty. It is found that precipitation (IR) error alone may explain moderate to low proportion of the soil moisture simulation uncertainty, depending on the level of model accuracy—50–60% for high model accuracy, and 20–30% for low model accuracy. Comparisons on the basis of two different sites also yielded an increase (50–100%) in soil moisture prediction uncertainty for the more vegetated site. This study exemplified the need for detailed investigations of the rainfall retrieval-modeling parameter error interaction within a comprehensive space-time stochastic framework for achieving optimal integration of satellite rain retrievals in land data assimilation systems.     

Comparison of soil moisture products from microwave remote sensing,land model,and reanalysis using global ground observations

High-quality soil moisture (SM) datasets are in great demand for climate, hydrology, and other fields, but detailed evaluation of SM products from various sources is scarce. Thus, using 670 SM stations worldwide, we evaluated and compared SM products from microwave remote sensing [Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E) (C- and X-bands) and European Space Agency's Climate Change Initiative (ESA CCI)], land surface model [Global Land Data Assimilation System (GLDAS)], and reanalysis data [ECMWF Re-Analysis-Interim (ERA-Interim) and National Centers for Environmental Prediction (NCEP)] under different time scales and various climates and land covers. We find that: (a) ESA CCI and GLDAS have the closest values to the in situ SM on the annual scale, whereas others overestimate the SM; ERA-Interim (averaged R = 0.58) and ESA CCI (averaged R = 0.54) correlate best with the in situ data, while GLDAS performs worst. (b) Overall, the deviations of each product vary in seasons. ESA CCI and ERA-Interim products are closer to the in situ SM at seasonal scales, and AMSR-E and NCEP perform worst in December–February and June–August, respectively. (c) Except for NCEP and ERA-Interim, others can well reflect the intermonthly variation of the in situ SM. (d) Under various climates and land covers, AMSR-E products are less effective in cold climates, whereas GLDAS and NCEP products perform poorly in arid or temperate and dry climates. Moreover, the Bias and R of each SM product differ obviously under different forest types, especially the AMSR-E products. In summary, SM from ESA CCI is the best, followed by ERA-Interim product, and precipitation is an important auxiliary data for selecting high-quality SM stations and improving the accuracy of SM from GLDAS. These results can provide a reference for improving the accuracy of the above SM products.     

Estimation of soil moisture status using near infrared reflectance

The use of a prototype near infrared reflectance meter for estimating the water content of soil is described. The instrument, developed from one used for estimating the water content of forage is based on the measurement of reflectance of infrared light emitted at wavelengths of 1450 nm, a strong water absorption band, and 1300 nm a weak water absorption band. Calibration curves of reflectance and reflectance ratio versus moisture content for pure sands and sand/clay mixes are presented. Problems associated with the measurement of moisture content using this technique on swelling soils are highlighted. The use of a modified form of this instrument for estimating soil moisture status in the field is discussed.     

Simulating sensitivity of soil moisture and evapotranspiration under heterogeneous soils and land uses

Soil moisture (SM) plays an important role in land surface and atmospheric interactions. It modifies energy balance at the surface and the rate of water cycling between the land and atmosphere. In this paper we provide a sensitivity assessment of SM and ET for heterogeneous soil physical properties and for three land uses including irrigated maize, rainfed maize, and grass at a climatological time-scale by using a water balance model. Not surprisingly, the study finds increased soil water content in the root zone throughout the year under irrigated farming. Soil water depletes to its lowest level under rainfed maize cultivation. We find a ‘land use’ effect as high as 36 percent of annual total evapotranspiration, under irrigated maize compared to rainfed maize and grass, respectively. Sensitivity analyses consisting of comparative simulations using the model show that soil characteristics, like water holding capacity, influence SM in the root zone and affect seasonal total ET estimates at the climatological time-scale. This ‘soils’ effect is smaller than the ‘land use’ effect associated with irrigation but, it is a source of consistent bias for both SM and ET estimates. The ‘climate’ effect basically masks the ‘soils’ effect under wet conditions. These results lead us to conclude that appropriate representation of land use, soils, and climate are necessary to accurately represent the water and energy balance in real landscapes.