1. Soil Moisture: Synergistic approach for the merge of thermal and ASCAT information
2nd User Training Workshop of Land Surface Analysis Satellite Application Facility (Land SAF)
Instituto de Meteorologia, Lisbon
8-10 March 2006

2. Soil Moisture Definition Scaling Cross-section of a soil Air Water
Solid Particles
Thin, remotely sensed soil layer
Root zone: layer of interest for most applications
Soil profile

3. Remote Sensing of Soil Moisture
Different applications have different requirements
e.g. runoff modeling vs. NWP
Product characteristics
Thematic content
Spatial resolution
Temporal sampling
Accuracy
Data availability
There is no universal method (= sensor + algorithm) that satisfies all user requirements
Sensor
Algorithm
Application

4. Hydrology SAF - Surface Water Budget
Hydrologists are primarily interested in runoff and water budgets
 … Soil moisture
 … Change in 
P … Precipitation
R … Runoff
ET… Evapotranspiration
Snow
In Red: SAF Products

5. Land SAF - Surface Energy Budget
Meteorologists must get the surface fluxes (radiation, water) right
Surface Energy Budget
C … Specific heat capacity
Ts .. Change in Ts
Q* … Net radiation
H … Sensible heat flux
 … Latent heat of vaporization
ET… Evapotranspiration
G … Ground heat flux
Surface Radiation Budget
Q*… Net radiation
 … Short-wave albedo (AL)
S… Short-wave radiation flux (DSSF)
L… Long-wave radiation flux (DSLF)
 … Long-wave emissivity (EM)
 … Stefan-Boltzmann constant
Ts… Land surface temperature (LST)
In Red: SAF Products

6. Soil Moisture – Linkage between SAFs
Soil moisture does not enter directly the energy balance equation
But strongly influences several terms
Evapotranspiration
Specific heat capacity
Soil dry = 800 J/kgK
Soil wet = 1480 J/kgK
Water = 4180 J/kgK
Emissivity
Albedo
Different Models of Actual ET
Actual ET
Potential ET
Decreasing importance
Field Capacity
Soil Moisture
Wilting Point

7. Soil Moisture - Thermal Approaches
Exploit the impact of  upon Ts and change of Ts (thermal inertia)
Other influences
Vegetation
Wind speed
Soil properties
Air humidity
etc.
Methods range from
Simple indices
Assimilation with SVAT-models
Soil moisture characterisation by surface temperature/vegetation index diagrams.
From Sandholt et al. (2002).

8. Land SAF Approach in Development Phase
Adjust soil moisture so that modelled surface temperature rise matches observed temperature rise
Used a modified version of the TESSEL code
Based on high-quality field data reasonable results for surface layer, but not for deeper layers
from Portmann et al. (2003)

9. Operational Readiness of Thermal Approaches
Many attempts have faced major difficulties
but more and more encouraging results are reported
Soil moisture from METEOSAT albedo and surface temperature versus EUROFLUX in-situ data according to Verstraeten
et al. (2006). Their method is based upon an index:

10. Microwave Soil Moisture Products
Global, coarse-resolution (25 km) soil moisture products
Operational, meteorological, polar orbiting satellites
METOP ASCAT
Scatterometer  = 5,7 cm
NPOESS CMIS
Radiometer  = 4,6 & 2,8 cm
SMOS
Radiometer  = 21 cm
Future METOP and NPOESS constellation

11. Global ASCAT Surface Soil Moisture Product
Characteristics
25 km spatial resolution
82 % daily global coverage
Available in NRT (135 min)
Limitations for hydrologic users
Resolution >> model grid
Surface layer (~0-2 cm)
Orbit geometry
Assimilation system required
Hydrology SAF
European disaggregated product
ECMWF profile soil moisture
Low
High
ERS Scatterometer 25 km surface soil moisture 09/10/05

12. European Disaggregated Surface Product
Characteristics
European coverage
Geo-referenced using a DEM
Disaggregated to 1 km based on land cover information
No soil moisture information over dense forests, cities, rocks, glaciers, water bodies
Could be available in NRT within 40 min using EARS*
Driven by experience from a hydrologic study over Austria by Parajka et al. (2005)
Masking of non-sensitive areas
Rocks
Forest
Water
* EUMETSAT Advanced Retransmission Service

13. Thermal versus Microwave Soil Moisture Data
Comparison of four satellite data sets versus in-situ data collected by Univ. of Salamanca, Spain
23 TDR probes installed at 5 cm depth
ERS scatterometer
Two AMSR-E products
NASA - NSIDC
Vrije Univ. Amsterdam
Thermal index
Actual/potential ET
Company EARS in Delft

14. Results of Comparison I

15. Results from Comparison II
a No data between Jan and Aug. 2003
b After normalisation
c Obtained by averaging R and RMSE values from Table 3

16. Insights from Comparison
Correlation between in-situ and EO data is in general modest because
TDR data represent sub-surface layer (2-8 cm) while EO data see only the top-most surface (0-2 cm)
Time difference between observations
Scaling: point-like versus areal data
Absolute soil moisture values are not comparable
All EO products captured soil moisture variations
most "consistent" data were AMSR-E soil moisture data from VUA, not the in-situ data!
Retrieval algorithm appears to be more important than the sensor

17. Can we get absolute soil moisture values?
We have only poor knowledge of soil properties at regional to global scales
soil type, layers, depth, texture
hydrologic properties (conductivity, wilting level, etc.)
thermal properties (specific heat capacity, etc.)
No!
Plant available water capacity 0-1 m (USDA)
Sandy soil profile from one
Salamanca TDR site

18. Synergy of thermal and ASCAT information
Benefits from merging
Better spatial and temporal resolution
Better modelling of  and Ts over soil profile
Moisture and temperature diffusion in soil profile are related
Deriving soil properties from  and Ts time series?

19. Conclusions
Synergistic information provided by MSG and METOP should be exploited to estimate soil moisture
in 0-1 m soil layer
1-4 km spatial resolution with daily sampling
using a data-driven, parsimonious modelling approach
that enables the "closure" of the surface energy balance
Complementary approach to ECMWF
Assimilation of ASCAT in TESSEL at 25 km
Potential first approaches, e.g.
Exponential filtering of ATI and ASCAT time series (Vertraeten et al., 2006)
Solving of flow equations using Ts and 