Ghent University Academic Bibliography

Advanced

Soil moisture retrieval from radar remote sensing using alternative approaches to surface roughness measurements

Hans Lievens UGent (2011)
abstract
Soil moisture is a key variable within the climate system, affecting as well water as heat fluxes. Knowledge of the spatial and temporal variability of soil moisture is therefore of high merit for watershed applications such as drought and flood prediction and crop irrigation scheduling. As acquiring ground measurements of soil moisture is labor intensive, it is often limited in space (to a few locations) and time (to a small number of campaigns). Remote sensing offers a useful alternative, as it allows for observing both across time and space. Synthetic aperture radar (SAR) has shown its large potential for retrieving soil moisture maps at regional scales. A SAR sensor onboard an aircraft or spacecraft emits microwaves that are scattered upon reaching the Earth's surface, with the strength and direction of the scattering depending on the surface characteristics, particularly the vegetation, soil moisture and surface roughness. Hence, the part of the signal returning back to the sensor, i.e., the radar backscatter, provides information about the surface characteristics, such as the soil moisture content. However, since the scattered signal is determined by several surface characteristics, the retrieval of soil moisture is an ill-posed problem when using single configuration imagery, operated at a single frequency, look angle and polarization. Unless accurate surface roughness parameter values are available, retrieving soil moisture from radar backscatter usually provides inaccurate estimates. Therefore, surface roughness parameters are often measured in the field, e.g., by collecting surface profile samples. However, the characterization of soil roughness is not fully understood, and a large range of roughness parameter values can be obtained for the same surface when different measurement methodologies are used. In this dissertation, a literature review is made that summarizes the problems encountered when parameterizing soil roughness from field-measured profiles. Further, a theoretical study is performed on synthetical surface profiles, which investigates how errors on roughness parameters are introduced by standard measurement techniques, and how they will propagate through a commonly used backscatter model, i.e., the Integral Equation Model (IEM), into a corresponding soil moisture retrieval error for some of the currently most used SAR configurations. Key aspects influencing roughness parameterization are: the length of the surface profile, the number of profile measurements, the horizontal and vertical accuracy of profile measurements and the removal of topography along profile transects. A number of suggestions are made which allow for circumventing or resolving the problems known to the parameterization of soil surface roughness prior to soil moisture retrieval from SAR. Recently, increasing interest has been drawn to the use of calibrated or effective roughness parameters, as they allow for circumventing field measurements of surface roughness. The principle lies in calibrating the roughness parameters of a backscatter model such as the IEM based on in situ soil moisture measurements and radar backscatter observations. However, notwithstanding unchanged roughness conditions, calibrated roughness parameters often differ between subsequent acquisitions over a single agricultural field. This variability may arguably be due to shortcomings of the backscatter model and unjustifies the use of calibration parameters that are constant over time. Therefore, this dissertation proposes a method that allows for adjusting the calibrated roughness parameters for each acquisition. Subsequently, these adjusted roughness parameters can be propagated through the IEM for soil moisture retrieval. The technique is developed and validated using a large amount of C-band (5.3 GHz) and L-band (1.3 GHz) SAR observations over a large number of fields located within different European countries, with a focus on bare soil fields with medium smooth surface roughness, e.g., resulting from seedbed preparation. The soil moisture retrieval accuracy is estimated between 4 vol% and 6.5 vol%, depending on the frequency and polarization. A further validation of the soil moisture retrieval methodology is performed over a study site in Flevoland, The Netherlands, for which a large time series of SAR imagery has been obtained in the framework of the AgriSAR 2009 campaign, organized by the European Space Agency (ESA). Coincidently with three acquisitions in August and September, 2009, in situ soil moisture measurements have been performed over 71 different bare soil fields, subdivided into four roughness classes. The backscatter observations are found to display a large sensitivity to both soil moisture and surface roughness. The largest sensitivity to soil moisture is observed for medium and rough surfaces; smooth surfaces show less agreement. Also, the presence of cereal harvest remains, i.e., stubbles, further lowers the sensitivity. Given the validity range of the IEM, the soil moisture retrieval technique is only applied to the medium roughness class, leading to accuracies of about 4 vol%. Next, the Flevoland data set is used for testing a second soil moisture retrieval technique based on change detection. This technique allows for circumventing roughness field measurements based on the reasoning that surface roughness changes over a larger time scale comparatively to soil moisture, such that short term changes in backscatter are only driven by changes in soil moisture. The soil moisture retrieval relies on a rescaling of backscatter between dry and wet reference soil moisture measurements. A similar accuracy of about 4 vol% is found. In agricultural catchments, fields are generally covered by vegetation during most part of the growing season. Therefore, SAR-based soil moisture retrieval of agricultural fields is often hampered by vegetation effects on the backscattered signal. The semi-empirical Water Cloud Model (WCM) allows for estimating the backscatter of a vegetated surface, accounting for both the contributions of the vegetation and the underlying soil. A method is proposed to fuse the IEM, adjusted with effective roughness parameterization, with the WCM. Furthermore, a number of vegetation indicators are compared with regard to their adequacy in describing wheat vegetation within the WCM. Based on a series of L-band SAR observations, it is shown that effective roughness parameters are a promising tool for soil moisture retrieval under a wheat canopy and that the use of the leaf area index (LAI) may be recommended above other vegetation indicators, as it leads to the lowest root mean square errors of about 5.5 vol%. These results prove the operational potential of L-band SAR data for soil moisture inferred under a wheat canopy throughout the entire crop growth cycle.
Please use this url to cite or link to this publication:
author
promoter
UGent
organization
year
type
dissertation
publication status
published
subject
keyword
surface roughness, Soil moisture, Synthetic Aperture Radar
pages
XXXIII, 203 pages
publisher
Ghent University. Faculty of Bioscience Engineering
place of publication
Ghent, Belgium
defense location
Gent : Faculteit Bio-ingenieurswetenschappen (A0.030)
defense date
2011-11-25 17:00
ISBN
9789059894839
language
English
UGent publication?
yes
classification
D1
additional info
dissertation consists of copyrighted material
copyright statement
I have transferred the copyright for this publication to the publisher
id
1946575
handle
http://hdl.handle.net/1854/LU-1946575
date created
2011-11-22 14:09:28
date last changed
2017-01-16 10:38:34
@phdthesis{1946575,
  abstract     = {Soil moisture is a key variable within the climate system, affecting as well water as heat fluxes. Knowledge of the spatial and temporal variability of soil moisture is therefore of high merit for watershed applications such as drought and flood prediction and crop irrigation scheduling. As acquiring ground measurements of soil moisture is labor intensive, it is often limited in space (to a few locations) and time (to a small number of campaigns). Remote sensing offers a useful alternative, as it allows for observing both across time and space.
Synthetic aperture radar (SAR) has shown its large potential for retrieving soil moisture maps at regional scales. A SAR sensor onboard an aircraft or spacecraft emits microwaves that are scattered upon reaching the Earth's surface, with the strength and direction of the scattering depending on the surface characteristics, particularly the vegetation, soil moisture and surface roughness. Hence, the part of the signal returning back to the sensor, i.e., the radar backscatter, provides information about the surface characteristics, such as the soil moisture content. However, since the scattered signal is determined by several surface characteristics, the retrieval of soil moisture is an ill-posed problem when using single configuration imagery, operated at a single frequency, look angle and polarization. Unless accurate surface roughness parameter values are available, retrieving soil moisture from radar backscatter usually provides inaccurate estimates. Therefore, surface roughness parameters are often measured in the field, e.g., by collecting surface profile samples. However, the characterization of soil roughness is not fully understood, and a large range of roughness parameter values can be obtained for the same surface when different measurement methodologies are used.
In this dissertation, a literature review is made that summarizes the problems encountered when parameterizing soil roughness from field-measured profiles. Further, a theoretical study is performed on synthetical surface profiles, which investigates how errors on roughness parameters are introduced by standard measurement techniques, and how they will propagate through a commonly used backscatter model, i.e., the Integral Equation Model (IEM), into a corresponding soil moisture retrieval error for some of the currently most used SAR configurations. Key aspects influencing roughness parameterization are: the length of the surface profile, the number of profile measurements, the horizontal and vertical accuracy of profile measurements and the removal of topography along profile transects. A number of suggestions are made which allow for circumventing or resolving the problems known to the parameterization of soil surface roughness prior to soil moisture retrieval from SAR. 
Recently, increasing interest has been drawn to the use of calibrated or effective roughness parameters, as they allow for circumventing field measurements of surface roughness. The principle lies in calibrating the roughness parameters of a backscatter model such as the IEM based on in situ soil moisture measurements and radar backscatter observations. However, notwithstanding unchanged roughness conditions, calibrated roughness parameters often differ between subsequent acquisitions over a single agricultural field. This variability may arguably be due to shortcomings of the backscatter model and unjustifies the use of calibration parameters that are constant over time. Therefore, this dissertation proposes a method that allows for adjusting the calibrated roughness parameters for each acquisition. Subsequently, these adjusted roughness parameters can be propagated through the IEM for soil moisture retrieval. The technique is developed and validated using a large amount of C-band (5.3 GHz) and L-band (1.3 GHz) SAR observations over a large number of fields located within different European countries, with a focus on bare soil fields with medium smooth surface roughness, e.g., resulting from seedbed preparation. The soil moisture retrieval accuracy is estimated between 4 vol\% and 6.5 vol\%, depending on the frequency and polarization.
A further validation of the soil moisture retrieval methodology is performed over a study site in Flevoland, The Netherlands, for which a large time series of SAR imagery has been obtained in the framework of the AgriSAR 2009 campaign, organized by the European Space Agency (ESA). Coincidently with three acquisitions in August and September, 2009, in situ soil moisture measurements have been performed over 71 different bare soil fields, subdivided into four roughness classes. The backscatter observations are found to display a large sensitivity to both soil moisture and surface roughness. The largest sensitivity to soil moisture is observed for medium and rough surfaces; smooth surfaces show less agreement. Also, the presence of cereal harvest remains, i.e., stubbles, further lowers the sensitivity. Given the validity range of the IEM, the soil moisture retrieval technique is only applied to the medium roughness class, leading to accuracies of about 4 vol\%. Next, the Flevoland data set is used for testing a second soil moisture retrieval technique based on change detection. This technique allows for circumventing roughness field measurements based on the reasoning that surface roughness changes over a larger time scale comparatively to soil moisture, such that short term changes in backscatter are only driven by changes in soil moisture. The soil moisture retrieval relies on a rescaling of backscatter between dry and wet reference soil moisture measurements. A similar accuracy of about 4 vol\% is found. 
In agricultural catchments, fields are generally covered by vegetation during most part of the growing season. Therefore, SAR-based soil moisture retrieval of agricultural fields is often hampered by vegetation effects on the backscattered signal. The semi-empirical Water Cloud Model (WCM) allows for estimating the backscatter of a vegetated surface, accounting for both the contributions of the vegetation and the underlying soil. A method is proposed to fuse the IEM, adjusted with effective roughness parameterization, with the WCM. Furthermore, a number of vegetation indicators are compared with regard to their adequacy in describing wheat vegetation within the WCM. Based on a series of L-band SAR observations, it is shown that effective roughness parameters are a promising tool for soil moisture retrieval under a wheat canopy and that the use of the leaf area index (LAI) may be recommended above other vegetation indicators, as it leads to the lowest root mean square errors of about 5.5 vol\%. These results prove the operational potential of L-band SAR data for soil moisture inferred under a wheat canopy throughout the entire crop growth cycle.},
  author       = {Lievens, Hans},
  isbn         = {9789059894839},
  keyword      = {surface roughness,Soil moisture,Synthetic Aperture Radar},
  language     = {eng},
  pages        = {XXXIII, 203},
  publisher    = {Ghent University. Faculty of Bioscience Engineering},
  school       = {Ghent University},
  title        = {Soil moisture retrieval from radar remote sensing using alternative approaches to surface roughness measurements},
  year         = {2011},
}

Chicago
Lievens, Hans. 2011. “Soil Moisture Retrieval from Radar Remote Sensing Using Alternative Approaches to Surface Roughness Measurements”. Ghent, Belgium: Ghent University. Faculty of Bioscience Engineering.
APA
Lievens, H. (2011). Soil moisture retrieval from radar remote sensing using alternative approaches to surface roughness measurements. Ghent University. Faculty of Bioscience Engineering, Ghent, Belgium.
Vancouver
1.
Lievens H. Soil moisture retrieval from radar remote sensing using alternative approaches to surface roughness measurements. [Ghent, Belgium]: Ghent University. Faculty of Bioscience Engineering; 2011.
MLA
Lievens, Hans. “Soil Moisture Retrieval from Radar Remote Sensing Using Alternative Approaches to Surface Roughness Measurements.” 2011 : n. pag. Print.