Comparison of soil moisture in GLDAS model simulations and satellite observations over the Murray Darling Basin.
Liu, Y. Y., M.F. McCabe, J.P. Evans, A.I.J.M. Van Dijk, R.A.M. de Jeu and H. Su
In Anderssen, R.S., R.D. Braddock and L.T.H. Newham (eds) 18th World IMACS Congress and MODSIM09 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand and International Association for Mathematics and Computers in Simulation, July 2009, pp. 2798-2804. ISBN: 978-0-9758400-7-8.
Soil moisture is a key hydrometeorological variable that can be derived from both modeling
simulations and satellite observations. This study compares Global Land Data Assimilation System (GLDAS)
output over the Murray Darling Basin against retrievals from a newly developed remote sensing product
using the AMSR-E sensor onboard NASA’s Aqua satellite. GLDAS is comprised of a number of land surface
models, two of which include the Community Land Model (CLM) and NOAH land surface scheme, which
provide a temporally and spatially consistent characterization of the hydrological cycle. GLDAS derived
estimates are 3-hourly products with 0.25-degree spatial resolution, while satellite based observations offer
twice-daily instantaneous retrievals at similar spatial scales. The models represent different soil moisture
averaging depths (roughly 2, 5, and 10 cm in CLM and 10 cm in NOAH) and retrievals from AMSR-E C-
band approximate the soil moisture in the top 1.5 cm layer. The spatial distribution and coherence of soil
moisture are investigated seasonally and under both wetting and drying conditions. From the spatial aspect,
AMSR-E observations and GLDAS simulations show similar seasonal patterns, while simulated soil
moisture is slightly higher during summer and autumn over the north-eastern Murray Darling Basin (MDB).
This may be explained by the positive biases of GLDAS forcing precipitation data. From the temporal
perspective, the best match between AMSR-E soil moisture and model simulations is found over the regions
with strong precipitation in warm months, e.g. north-eastern MDB. Over the regions with high precipitation
during cool months, AMSR-E soil moisture is systematically higher than model simulations. For the regions
with extremely low annual rainfall, the peak values in soil moisture between AMSR-E and model simulations
match very well, while low values of soil moisture display the greatest differences.
Generally, the agreements between AMSR-E observations and GLDAS simulations vary under different
wetting and drying conditions. Both of them can represent the ‘true’ soil moisture to some extent. How to
best blend soil moisture products derived from these two different techniques, in addition to data assimilation
approaches, will be explored in future research.
Figure 5:(a) Correlation coefficient (R) between monthly average AMSR-E soil moisture and modelling
simulated soil moisture (The average of soil moisture from 2, 5, and 10-cm in CLM and 10-cm in NOAHtaken before calculating correlation coefficient) for the study period (December 2003 through November
2005) and; (b) the same as (a), but for RMSE in the unit of m 3 /m3.
This page is maintaind by Jason Evans |
Last updated 31st January 2013