Model-data fusion for land-atmosphere coupling: Remote sensing evapotranspiration and soil moisture dynamics in Murrumbidgee catchment

Umair Rabbani

    Research output: ThesisDoctoral Thesis

    126 Downloads (Pure)

    Abstract

    Increasing stress on the world’s fresh water resources demands
    improvements to be made in increasing water productivity and encouraging
    efficient use of this limited resource. Global scale land surface models help
    us understand and predict the behaviour of terrestrial, atmospheric, climatic
    and hydrological processes that govern the ability to continuously update
    environmental policies. Irrigated agriculture is the largest consumer of the
    world’s fresh water resources and is a key component of the terrestrial
    water cycle. Global food security, which greatly depends upon irrigated
    agriculture, is facing serious threats due to limited fresh water availability.
    One of the most important factors for sustainability of irrigation in the
    future, under water stress circumstances, is the effort towards developing a
    better understanding of the land-atmosphere coupling processes that
    govern the hydrological principles and lead to effective use of water with
    better model predictions. Quantification of evapotranspiration is one of the
    vital components for water budgeting, efficient irrigation scheduling,
    cropping practices and water regulation in an irrigation system. Many
    remote-sensing algorithms have been developed over the years to model
    spatial actual evapotranspiration for larger areas to help improve water
    balances. More recently, efforts to derive surface soil moisture information
    to improve agricultural monitoring quality through remote sensing have also
    increased with the advancements in microwave sensing and retrieval
    models.

    Continuous efforts over the past few years to model soil moisture
    and relate it to point- and remote-sensed observations have led to somewhat improved availability and quality of surface soil moisture
    datasets. As a result of increased availability of microwave soil moisture
    datasets, it is now recommendable to test the potential of estimating rootzone
    soil moisture from remote sensing derived surface soil moisture to
    understand its spatial distribution. This study aims to explore the coupling
    of microwave derived soil moisture with surface energy balance
    components, and investigate the potential of estimating root-zone soil
    moisture by using a practicable, simplified assimilation technique in the
    Murrumbidgee catchment.

    The study can be categorised into three stages. In the first stage, a
    Large Aperture Scintillometer (LAS) was installed over a horticultural farm
    near Leeton, NSW, to provide ground calibration for remote sensing energy
    balance modelling. LAS scintillation data was used to calculate sensible heat
    flux for the entire half-hourly time-series data. The latent heat flux was then
    estimated by solving an energy balance of fluxes measured by net
    radiometer and soil heat flux, and H calculated using LAS. The LAS
    performed very well for the purpose of heat flux estimation for energy
    balance closure, provided the data was filtered for bad or missing values
    generated by various meteorological conditions or sensor errors. A network
    of meteorological stations enabled the testing of sensitivity of micrometeorological variations occurring within the path-length of the LAS on
    sensible heat flux calculations. One-way between groups ANOVA analysis
    and Tukey’s HSD analysis suggested that moving the AWS further along the
    stretched path length will generate statistically significant differences in
    sensible heat flux, and eventually influence the whole energy balance
    closure i.e. meteorological conditions differed enough towards the centre of the LAS beam to produce a mean difference of ~14 W/m2 in the sensible
    heat flux. The difference in the instantaneous estimate of H reached up to
    100 W/m2 in some instances. A 50–100 W/m2 difference in sensible heat
    flux can result in a 1.25–2.5 mm.d-1 error in daily evapotranspiration flux
    and affect the entire water balance for large regions. Further, energy
    balance modelling over the Murrumbidgee catchment was performed using
    Terra/MODIS data for year 2010/11. The results revealed that SEBS
    overestimated soil heat flux for higher values while it underestimated net
    radiation for higher values.

    Later, root-zone soil moisture dynamics were modelled using an
    exponential filter (Wagner et al., 1999) using an AMSR-E surface soil
    moisture dataset over six ground calibration sites. Time-step-based
    statistical analysis between SEBS-derived actual evapotranspiration was
    analysed with in-situ observations of surface soil moisture. Weak negative
    correlation was observed between moisture and actual evapotranspiration,
    which was not seen while relating surface fluxes to AMSR-E-derived
    moisture at these sites. Finally, cross-correlation analysis was carried out
    between measured surface and measured root-zone moisture time-series
    with a time lag of 1 day to match Aqua/AMSR-E temporal resolution. A
    strong positive correlation was found between the observed surface
    moisture (SMSL(t)) and observed root-zone moisture of next (SMRZ(t+1)). An
    exponential filter was applied on the AMSR-E soil moisture time-series to
    calculate sub-surface moisture. The model performed poorly in estimating
    root-zone moisture from AMSR-E data. A maximum correlation of 0.4681 with a low Nash-Sutcliffe coefficient value of -1.23 was observed. The applied exponential filter model-DA showed potential for root-zone moisture extraction, but the accuracy observation serves as a pre-requisite to base our understanding of spatial distribution of moisture and its coupling with actual evapotranspiration on it. Improvements in remotely sensed soil moisture observations will act as the cornerstone in enhancing understandings in land-atmosphere coupling by facilitating an operational assimilation scheme for estimating spatial root-zone soil moisture that is representative of the actual moisture state.
    Original languageEnglish
    QualificationDoctor of Philosophy
    Awarding Institution
    • Charles Sturt University
    Supervisors/Advisors
    • Hafeez, Muhammad, Principal Supervisor
    • Chemin, Yann, Co-Supervisor
    Award date18 Nov 2014
    Place of PublicationAustralia
    Publisher
    Publication statusPublished - 2014

    Fingerprint

    Dive into the research topics of 'Model-data fusion for land-atmosphere coupling: Remote sensing evapotranspiration and soil moisture dynamics in Murrumbidgee catchment'. Together they form a unique fingerprint.

    Cite this