|
Titel |
Surface soil moisture mapping at high spatial resolution over agricultural bare soils by using TERRASAR-X and FORMOSAT-2 data. |
VerfasserIn |
Frederic Baup, Danielle Ducrot, Jean-François Dejoux, Claire Marais-Sicre, Armand Lopes, Olivier Hagolle, Gérard Dedieu, Eric Ceschia, Vincent Rivalland, Laurie De Brondeau |
Konferenz |
EGU General Assembly 2010
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 12 (2010) |
Datensatznummer |
250043106
|
|
|
|
Zusammenfassung |
This work aims to investigate the capabilities of TERRASAR-X SAR (Synthetic Aperture
Radar) and FORMOSAT-2 high resolution and multi-spectral data to estimate the surface soil
moisture content of bare agricultural soils. Emphasis is put on the use of time series
of TerraSAR Strip Map mode images acquired during the autumn 2009, between
the 28 of September and the 11 of November, that matches with the period where
agricultural fields are mainly bare (more than 40% of the area is composed of bare
soils).
The analysis is firstly performed by extracting the bare soil surfaces from FORMOSAT-2
NDVI data. Then, X-band backscattering coefficients acquired at high incidence angle (43Ë )
over bare soils are compared with field measurements collected over local sites located in the
south-west of Toulouse (43Ë 29’36”N, 01Ë 14’14”E) in France. Field data consists of Surface
Soil Moisture (SSM) and surface soil roughness respectively recorded by a Theta Probe
sensor and a 2-meters long profilometer. Measurements are performed along several
meters transects depending of the field size (at least several hectares). Changes in
soil practices (plough, sown-¦) and spatial tillage orientation are also monitored in
time.
Results show the low sensitivity of the radar backscattering coefficient (43Ë ) to the
tillage orientation. Signal variation lower than 0.5dB is observed when considering a relative
angle view ranged between 0 and 90Ë (θsatellite view angle - θfield tillage orientation). A
well marked correlation between radar data and SSM measurements (r2 =0.75) is then
observed whatever the soil practices, which is strongly important since C-band or L-band
data do not allow significant SSM estimations without considering soil roughness
corrections. Surface soil moisture is accurately estimated at local scale (rmse about
5%) thanks to the high spatial resolution images and to the small wavelength data
(about 3 cm) less sensitive to soil roughness changes. Finally, soil moisture maps
are processed, and well indicate the spatial variability of soil moisture over bare
agricultural fields. Spatial analyses are performed by considering soil practices and soil
properties.
These results suggest that high resolution X-band images could be used to derive
multi-temporal SSM maps over agricultural bare soils, by neglecting the tillage orientation
contrary to analyses performed by C- and L-band data.
The following work consists to 1) assimilate the high resolution soil moisture maps in
physical models to improve evapotranspiration estimates at regional scale 2) combine X-band
data with C- and L-band data in order to estimate soil moisture over vegetated fields. |
|
|
|
|
|