Multi-scale assimilation of SMAP data: comparison between land surface and land surface-subsurface model

2020 ◽  
Author(s):  
Haojin Zhao ◽  
Roland Baatz ◽  
Carsten Montzka ◽  
Harry Vereecken ◽  
Harrie-Jan Hendricks Franssen
Keyword(s):  
Parameter Estimation ◽  
Soil Moisture ◽  
Data Assimilation ◽  
Simulation Model ◽  
Land Surface ◽  
Cosmic Ray ◽  
Remotely Sensed ◽  
Surface Model ◽  
Systematic Bias

<p>Soil moisture plays an important role in the coupled water and energy cycles of the terrestrial system. However, the characterization of soil moisture at the large spatial scale is far from trivial. To cope with this challenge, the combination of data from different sources (in situ measurements by cosmic ray neutron sensors, remotely sensed soil moisture and simulated soil moisture by models) is pursued. This is done by multiscale data assimilation, to take the different resolutions of the data into account. A large number of studies on the assimilation of remotely sensed soil moisture in land surface models has been published, which show in general only a limited improvement in the characterization of root zone soil moisture, and no improvement in the characterization of evapotranspiration. In this study it was investigated whether an improved modelling of soil moisture content, using a simulation model where the interactions between the land surface, surface water and groundwater are better represented, can contribute to extracting more information from SMAP data. In this study over North-Rhine-Westphalia, the assimilation of remotely sensed soil moisture from SMAP in the coupled land surface-subsurface model TSMP was tested. Results were compared with the assimilation in the stand-alone land surface model CLM. It was also tested whether soil hydraulic parameter estimation in combination with state updating could give additional skill compared to assimilation in CLM stand-alone and without parameter updating. Results showed that modelled soil moisture by TSMP did not show a systematic bias compared to SMAP, whereas CLM was systematically wetter than TSMP. Therefore, no prior bias correction was needed in the data assimilation. The results illustrate how the difference in simulation model and parameter estimation result in significantly different estimated soil moisture contents and evapotranspiration.  </p>

Cosmic Ray Neutron Sensing: Integration with land surface modelling using data assimilation for improved field-scale soil moisture estimates

2020 ◽  
Author(s):  
Amol Patil ◽  
Benjamin Fersch ◽  
Harrie-Jan Hendricks-Franssen ◽  
Harald Kunstmann
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Spatial Resolution ◽  
Land Surface ◽  
Root Zone ◽  
Cosmic Ray ◽  
Soil Parameters ◽  
Surface Model ◽  
Field Scale ◽  
Spatio Temporal

<p>Soil moisture is a key variable in atmospheric modelling to resolve the partitioning of net radiation into sensible and latent heat fluxes. Therefore, high resolution spatio-temporal soil moisture estimation is getting growing attention in this decade. The recent developments to observe soil moisture at field scale (170 to 250 m spatial resolution) using Cosmic Ray Neutron Sensing (CRNS) technique has created new opportunities to better resolve land surface atmospheric interactions; however, many challenges remain such as spatial resolution mismatch and estimation uncertainties. Our study couples the Noah-MP land surface model to the Data Assimilation Research Testbed (DART) for assimilating CRN intensities to update model soil moisture. For evaluation, the spatially distributed Noah-MP was set up to simulate the land surface variables at 1 km horizontal resolution for the Rott and Ammer catchments in southern Germany. The study site comprises the TERENO-preAlpine observatory with five CRNS stations and additional CRNS measurements for summer 2019 operated by our Cosmic Sense research group. We adjusted the soil parametrization in Noah-MP to allow the usage of EU soil data along with Mualem-van Genuchten soil hydraulic parameters. We use independent observations from extensive soil moisture sensor network (SoilNet) within the vicinity of CRNS sensors for validation. Our detailed synthetic and real data experiments are evaluated for the analysis of the spatio-temporal changes in updated root zone soil moisture and for implications on the energy balance component of Noah-MP. Furthermore, we present possibilities to estimate root zone soil parameters within the data assimilation framework to enhance standalone model performance.</p>

scholarly journals Improved Hydrological Simulation Using SMAP Data: Relative Impacts of Model Calibration and Data Assimilation

2018 ◽  
Vol 19 (4) ◽  
pp. 727-741 ◽  
Author(s):  
Randal D. Koster ◽  
Qing Liu ◽  
Sarith P. P. Mahanama ◽  
Rolf H. Reichle
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Land Surface ◽  
Model Calibration ◽  
Remotely Sensed ◽  
Surface Model ◽  
Remotely Sensed Data ◽  
Satellite Mission ◽  
Hydrological Simulation ◽  
Near Surface

Abstract The assimilation of remotely sensed soil moisture information into a land surface model has been shown in past studies to contribute accuracy to the simulated hydrological variables. Remotely sensed data, however, can also be used to improve the model itself through the calibration of the model’s parameters, and this can also increase the accuracy of model products. Here, data provided by the Soil Moisture Active Passive (SMAP) satellite mission are applied to the land surface component of the NASA GEOS Earth system model using both data assimilation and model calibration in order to quantify the relative degrees to which each strategy improves the estimation of near-surface soil moisture and streamflow. The two approaches show significant complementarity in their ability to extract useful information from the SMAP data record. Data assimilation reduces the ubRMSE (the RMSE after removing the long-term bias) of soil moisture estimates and improves the timing of streamflow variations, whereas model calibration reduces the model biases in both soil moisture and streamflow. While both approaches lead to an improved timing of simulated soil moisture, these contributions are largely independent; joint use of both approaches provides the highest soil moisture simulation accuracy.

Improving estimates of UK soil moisture using the JULES land surface model with COSMOS-UK measurements in the LaVEnDAR data assimilation framework

2020 ◽  
Author(s):  
Elizabeth Cooper ◽  
Ewan Pinnington ◽  
Richard Ellis ◽  
Eleanor Blyth ◽  
Simon Dadson ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Soil Texture ◽  
Land Surface ◽  
Land Surface Model ◽  
Cosmic Ray ◽  
Surface Model ◽  
Soil Physics ◽  
Soil Moisture Sensors ◽  
The Uk

<p>Soil moisture predictions are increasingly important in hydrological, ecological and agricultural applications. In recent years the availability of wide-area assessments of current and future soil-moisture states has grown, yet few studies have combined model-based assessments with observations beyond the point scale. Here we use the JULES land surface model together with COSMOS-UK data to evaluate the extent to which data assimilation can improve predictions of soil moisture across the United Kingdom.</p><p>COSMOS-UK is a network of soil moisture sensors run by UKCEH. The network provides soil moisture measurements at around 50 sites throughout the UK using innovative Cosmic Ray Neutron Sensors (CRNS). Half hourly measurements of the meteorological variables that the Joint UK Land Environment Simulator (JULES) requires as driving data are also recorded at COSMOS-UK sites, allowing us to run JULES at observation locations. This provides a unique opportunity to compare soil moisture outputs from JULES with CRNS observations; these measurements have a footprint of up to 12 ha (approx 30 acres) and are therefore better scale matched with JULES outputs than those from point sensors.</p><p>We have used the Land Variational Ensemble Data Assimilation Framework (LaVEnDAR) to combine soil moisture estimates from JULES with daily CRNS observations from one year at a number of COSMOS-UK sites. We show that this results in improved soil moisture predictions from JULES over several years. This has been achieved by optimising parameters in the pedo-transfer function used to derive JULES soil physics parameters from soil texture information. Using data assimilation with LaVEnDAR in this way allows us to explore the relationships between soil moisture estimates, soil physics parameters and soil texture, as well as improving the agreement between JULES model outputs and observations.</p>

scholarly journals Improving soil moisture prediction of a high-resolution land surface model by parameterising pedotransfer functions through assimilation of SMAP satellite data

2021 ◽  
Vol 25 (3) ◽  
pp. 1617-1641
Author(s):  
Ewan Pinnington ◽  
Javier Amezcua ◽  
Elizabeth Cooper ◽  
Simon Dadson ◽  
Rich Ellis ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Land Surface ◽  
Mean Squared Error ◽  
Land Surface Model ◽  
Cosmic Ray ◽  
Pedotransfer Functions ◽  
Surface Model ◽  
Squared Error ◽  
Surface Models

Abstract. Pedotransfer functions are used to relate gridded databases of soil texture information to the soil hydraulic and thermal parameters of land surface models. The parameters within these pedotransfer functions are uncertain and calibrated through analyses of point soil samples. How these calibrations relate to the soil parameters at the spatial scale of modern land surface models is unclear because gridded databases of soil texture represent an area average. We present a novel approach for calibrating such pedotransfer functions to improve land surface model soil moisture prediction by using observations from the Soil Moisture Active Passive (SMAP) satellite mission within a data assimilation framework. Unlike traditional calibration procedures, data assimilation always takes into account the relative uncertainties given to both model and observed estimates to find a maximum likelihood estimate. After performing the calibration procedure, we find improved estimates of soil moisture and heat flux for the Joint UK Land Environment Simulator (JULES) land surface model (run at a 1 km resolution) when compared to estimates from a cosmic-ray soil moisture monitoring network (COSMOS-UK) and three flux tower sites. The spatial resolution of the COSMOS probes is much more representative of the 1 km model grid than traditional point-based soil moisture sensors. For 11 cosmic-ray neutron soil moisture probes located across the modelled domain, we find an average 22 % reduction in root mean squared error, a 16 % reduction in unbiased root mean squared error and a 16 % increase in correlation after using data assimilation techniques to retrieve new pedotransfer function parameters.

scholarly journals Improving Soil Moisture Prediction of a High–Resolution Land Surface Model by Parameterising Pedotransfer Functions through Assimilation of SMAP Satellite Data

2020 ◽  
Author(s):  
Ewan Pinnington ◽  
Javier Amezcua ◽  
Elizabeth Cooper ◽  
Simon Dadson ◽  
Rich Ellis ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Land Surface ◽  
Mean Squared Error ◽  
Land Surface Model ◽  
Cosmic Ray ◽  
Pedotransfer Functions ◽  
Surface Model ◽  
Squared Error ◽  
Surface Models

Abstract. Pedotransfer functions are used to relate gridded databases of soil texture information to the soil hydraulic and thermal parameters of land surface models. The parameters within these pedotransfer functions are uncertain and calibrated through analyses of point soil samples. How these calibrations relate to the soil parameters at the spatial scale of modern land surface models is unclear, because gridded databases of soil texture represent an area average. We present a novel approach for calibrating such pedotransfer functions to improve land surface model soil moisture prediction by using observations from the Soil Moisture Active Passive (SMAP) satellite mission within a data assimilation framework. Unlike traditional calibration procedures data assimilation always takes into account the relative uncertainties given to both model and observed estimates to find a maximum likelihood estimate. After performing the calibration procedure we find improved estimates of soil moisture for the JULES land surface model (run at a 1 km resolution) when compared to estimates from a cosmic-ray soil moisture monitoring network (COSMOS-UK). The spatial resolution of these COSMOS probes is much more representative of the 1 km model grid than traditional point based soil moisture sensors. For 11 cosmic-ray neutron soil moisture probes located across the modelled domain we find an average 22 % reduction in root-mean squared error, a 16 % reduction in unbiased root-mean squared error and a 16 % increase in correlation after using data assimilation techniques to retrieve new pedotransfer function parameters.

Building a coupled data assimilation system for the atmosphere, land-surface and subsurface on the catchment scale

2021 ◽  
Author(s):  
Bernd Schalge ◽  
Barbara Haese ◽  
Bastian Waldowski ◽  
Natascha Brandhorst ◽  
Emilio Sanchez ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Land Surface ◽  
Coupled Model ◽  
Cosmic Ray ◽  
Atmospheric Model ◽  
Surface Model ◽  
Catchment Scale ◽  
Near Surface ◽  
Fully Coupled

<p>We present a data assimilation (DA) system for the atmosphere-land-surface-subsurface system on the catchment scale. The Neckar catchment in SW-Germany served as the specific case where the DA in combination with the coupled atmosphere-land surface-subsurface model TSMP was used<!-- This needs further clarification. -->. TSMP couples the atmospheric model COSMO, the land-surface model CLM and the hydrological model ParFlow to the DA framework PDAF. We will discuss how the ensemble system is set up in order to work properly and what issues we faced during our initial testing. For the atmosphere we found that it is important to have a good ensemble of lateral forcings as changing internal parameters for various parametrizations does not introduce sufficient variability on its own due to the rather small size of our domain. For the sub-surface the choice of parameters becomes most important and as such parameter estimation will be a valuable tool for improving DA results significantly. Finally, we are showing some first DA results with our system concerning soil moisture with two different assimilation methods <!-- This should be more precise. What do you exactly mean? Two data assimilation methods? Two different simulation scenarios? -->with a fully coupled model setup. In the first assimilation scenario in-situ soil moisture data measured by cosmic ray probes are assimilated, while in the second assimilation scenario remotely sensed near surface soil moisture is assimilated. The first results are encouraging and we discuss additional planned simulation scenarios with the fully coupled atmosphere-land surface-subsurface modelling system as well as plans to test strongly coupled DA, where measurements are used to update states across compartments, possibly resulting in additional accuracy gain compared to traditional uncoupled DA.</p><p><span> </span></p>

scholarly journals The COsmic-ray Soil Moisture Interaction Code (COSMIC) for use in data assimilation

2013 ◽  
Vol 17 (8) ◽  
pp. 3205-3217 ◽  
Author(s):  
J. Shuttleworth ◽  
R. Rosolem ◽  
M. Zreda ◽  
T. Franz
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Fast Neutron ◽  
Land Surface ◽  
Soil Depth ◽  
Cosmic Ray ◽  
High Energy ◽  
Fast Neutrons ◽  
Surface Model ◽  
Neutron Intensity

Abstract. Soil moisture status in land surface models (LSMs) can be updated by assimilating cosmic-ray neutron intensity measured in air above the surface. This requires a fast and accurate model to calculate the neutron intensity from the profiles of soil moisture modeled by the LSM. The existing Monte Carlo N-Particle eXtended (MCNPX) model is sufficiently accurate but too slow to be practical in the context of data assimilation. Consequently an alternative and efficient model is needed which can be calibrated accurately to reproduce the calculations made by MCNPX and used to substitute for MCNPX during data assimilation. This paper describes the construction and calibration of such a model, COsmic-ray Soil Moisture Interaction Code (COSMIC), which is simple, physically based and analytic, and which, because it runs at least 50 000 times faster than MCNPX, is appropriate in data assimilation applications. The model includes simple descriptions of (a) degradation of the incoming high-energy neutron flux with soil depth, (b) creation of fast neutrons at each depth in the soil, and (c) scattering of the resulting fast neutrons before they reach the soil surface, all of which processes may have parameterized dependency on the chemistry and moisture content of the soil. The site-to-site variability in the parameters used in COSMIC is explored for 42 sample sites in the COsmic-ray Soil Moisture Observing System (COSMOS), and the comparative performance of COSMIC relative to MCNPX when applied to represent interactions between cosmic-ray neutrons and moist soil is explored. At an example site in Arizona, fast-neutron counts calculated by COSMIC from the average soil moisture profile given by an independent network of point measurements in the COSMOS probe footprint are similar to the fast-neutron intensity measured by the COSMOS probe. It was demonstrated that, when used within a data assimilation framework to assimilate COSMOS probe counts into the Noah land surface model at the Santa Rita Experimental Range field site, the calibrated COSMIC model provided an effective mechanism for translating model-calculated soil moisture profiles into aboveground fast-neutron count when applied with two radically different approaches used to remove the bias between data and model.

scholarly journals Evaluation of a cosmic-ray neutron sensor network for improved land surface model prediction

2017 ◽  
Vol 21 (5) ◽  
pp. 2509-2530 ◽  
Author(s):  
Roland Baatz ◽  
Harrie-Jan Hendricks Franssen ◽  
Xujun Han ◽  
Tim Hoar ◽  
Heye Reemt Bogena ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Data Assimilation ◽  
Land Surface ◽  
Land Surface Model ◽  
Cosmic Ray ◽  
Surface Model ◽  
Hydraulic Parameters ◽  
Soil Map ◽  
Soil Hydraulic Parameters ◽  
Soil Hydraulic

Abstract. In situ soil moisture sensors provide highly accurate but very local soil moisture measurements, while remotely sensed soil moisture is strongly affected by vegetation and surface roughness. In contrast, cosmic-ray neutron sensors (CRNSs) allow highly accurate soil moisture estimation on the field scale which could be valuable to improve land surface model predictions. In this study, the potential of a network of CRNSs installed in the 2354 km2 Rur catchment (Germany) for estimating soil hydraulic parameters and improving soil moisture states was tested. Data measured by the CRNSs were assimilated with the local ensemble transform Kalman filter in the Community Land Model version 4.5. Data of four, eight and nine CRNSs were assimilated for the years 2011 and 2012 (with and without soil hydraulic parameter estimation), followed by a verification year 2013 without data assimilation. This was done using (i) a regional high-resolution soil map, (ii) the FAO soil map and (iii) an erroneous, biased soil map as input information for the simulations. For the regional soil map, soil moisture characterization was only improved in the assimilation period but not in the verification period. For the FAO soil map and the biased soil map, soil moisture predictions improved strongly to a root mean square error of 0.03 cm3 cm−3 for the assimilation period and 0.05 cm3 cm−3 for the evaluation period. Improvements were limited by the measurement error of CRNSs (0.03 cm3 cm−3). The positive results obtained with data assimilation of nine CRNSs were confirmed by the jackknife experiments with four and eight CRNSs used for assimilation. The results demonstrate that assimilated data of a CRNS network can improve the characterization of soil moisture content on the catchment scale by updating spatially distributed soil hydraulic parameters of a land surface model.

scholarly journals DasPy 1.0 – the Open Source Multivariate Land Data Assimilation Framework in combination with the Community Land Model 4.5

2015 ◽  
Vol 8 (8) ◽  
pp. 7395-7444 ◽  
Author(s):  
X. Han ◽  
X. Li ◽  
G. He ◽  
P. Kumbhar ◽  
C. Montzka ◽  
...  
Keyword(s):  
Parameter Estimation ◽  
Soil Moisture ◽  
Data Assimilation ◽  
Open Source ◽  
Land Surface Temperature ◽  
Land Surface ◽  
Cosmic Ray ◽  
Community Land Model ◽  
Land Data Assimilation ◽  
L Band

Abstract. Data assimilation has become a popular method to integrate observations from multiple sources with land surface models to improve predictions of the water and energy cycles of the soil-vegetation-atmosphere continuum. Multivariate data assimilation refers to the simultaneous assimilation of observation data from multiple model state variables into a simulation model. In recent years, several land data assimilation systems have been developed in different research agencies. Because of the software availability or adaptability, these systems are not easy to apply for the purpose of multivariate land data assimilation research. We developed an open source multivariate land data assimilation framework (DasPy) which is implemented using the Python script language mixed with the C++ and Fortran programming languages. LETKF (Local Ensemble Transform Kalman Filter) is implemented as the main data assimilation algorithm, and uncertainties in the data assimilation can be introduced by perturbed atmospheric forcing data, and represented by perturbed soil and vegetation parameters and model initial conditions. The Community Land Model (CLM) was integrated as the model operator. The implementation allows also parameter estimation (soil properties and/or leaf area index) on the basis of the joint state and parameter estimation approach. The Community Microwave Emission Modelling platform (CMEM), COsmic-ray Soil Moisture Interaction Code (COSMIC) and the Two-Source Formulation (TSF) were integrated as observation operators for the assimilation of L-band passive microwave, cosmic-ray soil moisture probe and land surface temperature measurements, respectively. DasPy has been evaluated in several assimilation studies of neutron count intensity (soil moisture), L-band brightness temperature and land surface temperature. DasPy is parallelized using the hybrid Message Passing Interface and Open Multi-Processing techniques. All the input and output data flows are organized efficiently using the commonly used NetCDF file format. Online 1-D and 2-D visualization of data assimilation results is also implemented to facilitate the post simulation analysis. In summary, DasPy is a ready to use open source parallel multivariate land data assimilation framework.

scholarly journals Assimilation of Remotely Sensed Soil Moisture and Snow Depth Retrievals for Drought Estimation

2014 ◽  
Vol 15 (6) ◽  
pp. 2446-2469 ◽  
Author(s):  
Sujay V. Kumar ◽  
Christa D. Peters-Lidard ◽  
David Mocko ◽  
Rolf Reichle ◽  
Yuqiong Liu ◽  
...  
Keyword(s):  
United States ◽  
Soil Moisture ◽  
Data Assimilation ◽  
Land Surface ◽  
Snow Depth ◽  
Root Zone ◽  
Remotely Sensed ◽  
Open Loop ◽  
Surface Model ◽  
The Impact

Abstract The accurate knowledge of soil moisture and snow conditions is important for the skillful characterization of agricultural and hydrologic droughts, which are defined as deficits of soil moisture and streamflow, respectively. This article examines the influence of remotely sensed soil moisture and snow depth retrievals toward improving estimates of drought through data assimilation. Soil moisture and snow depth retrievals from a variety of sensors (primarily passive microwave based) are assimilated separately into the Noah land surface model for the period of 1979–2011 over the continental United States, in the North American Land Data Assimilation System (NLDAS) configuration. Overall, the assimilation of soil moisture and snow datasets was found to provide marginal improvements over the open-loop configuration. Though the improvements in soil moisture fields through soil moisture data assimilation were barely at the statistically significant levels, these small improvements were found to translate into subsequent small improvements in simulated streamflow. The assimilation of snow depth datasets were found to generally improve the snow fields, but these improvements did not always translate to corresponding improvements in streamflow, including some notable degradations observed in the western United States. A quantitative examination of the percentage drought area from root-zone soil moisture and streamflow percentiles was conducted against the U.S. Drought Monitor data. The results suggest that soil moisture assimilation provides improvements at short time scales, both in the magnitude and representation of the spatial patterns of drought estimates, whereas the impact of snow data assimilation was marginal and often disadvantageous.

Sign in / Sign up

Export Citation Format

Share Document