10-year of research on the use of in situ and satellite soil moisture observations for hydrological applications, enjoy!!!

1. Luca Brocca
T. Moramarco, S. Barbetta, A. Tarpanelli, S. Camici, C. Massari, G. Zucco,
C. Corradini, P. Maccioni, L. Ciabatta
Research Institute for Geo-Hydrological Protection (IRPI-CNR), Perugia, Italy
50th anniversary symposium:
State of the art measurements of catchment-scale hydrological processes
http://hydrology.irpi.cnr.it
10 th Sept 2015

2. Soil moisture is a key
variable of the
climate system.
Soil moisture
generally refers to the
amount of water
stored in the
unsaturated soil zone,
although its exact
definition can vary
depending on the
context, i.e. whether
it is defined in
relative, absolute or
indirect terms, and
depending on the
reference storage.
What is soil moisture?

3. Casentino basin
central Italy
30% increase of soil
moisture produces a
8-fold increase of
peak discharge!
FLOOD
DROUGHT
WEATHER
PREDICTION
CLIMATE
SYSTEM
LANDSLIDES
CROP
PRODUCTION
Why soil moisture?

4. ANTECEDENT
WETNESS
CONDITIONS
Brocca et al., 2009 JHE;
Massari et al., 2014
HESS; Tramblay et al.,
2012; …
SOIL MOISTURE
SPATIAL-TEMPORAL
VARIABILITY
Brocca et al., 2007 JoH;
2009 GEOD; 2010;
2014 WRR; Zucco et
al., 2014; …
FLOOD FREQUENCY
ANALYSIS
Camici et al., 2011
WRR
SOIL MOISTURE &
LANDSLIDE
PREDICTION
Brocca et al., 2012 RS;
Ponziani et al., 2012
LASL
RAINFALL-RUNOFF
MODELLING
Brocca et al., 2011
HYP; 2013 HESS; Tayfur
et al., 2015 WARM
SOIL MOISTURE
MODELLING
Brocca et al., 2008
HYP; 2014 HYP; Lacava
et al., 2012
SOIL MOISTURE &
DROUGHT
MONITORING
Maccioni et al., 2014
JHE; Rahmani et al.,
2015 JAG
REMOTE SENSING
VALIDATION
Brocca et al., 2011 RSE;
Dorigo et al., 2015 RSE;
Wagner et al., 2013
IEEE TGRS; …
SOIL MOISTURE
DATA ASSIMILATION
Brocca et al., 2010
HESS; 2012 IEEE TGRS;
Massari et al., 2015 RS
GEOPHYSICAL
METHODS
Calamita et al., 2012
JoH; 2015 JoH
SOIL MOISTURE
FOR SOIL EROSION
Todisco et al., 2015
HESS
COSMIC-RAY
NEUTRONS
Franz et al., 2015 GRL
SOIL MOISTURE &
CLIMATE CHANGE
Camici et al., 2014 JHE;
Ciabatta et al., 2015
JoH
NUMERICAL
WEATHER
PREDICTION
Capecchi & Brocca et
al., 2014 METZET
FROM SURFACE TO
ROOT-ZONE
MODELLING
Brocca et al., 2010 RSE;
Manfreda et al., 2014
HESS
SM2RAIN
Brocca et al., 2013
GRL; 2014 JGR;
Massari et al., 2014
AWR; Ciabatta et al.,
2015 JHM; 2015 JAG
10-year of research on soil moisture

5. Soil moisture
monitoring
with in situ
and remote
sensing
Understanding
the spatial-
temporal
variability of soil
moisture at
different spatial
scales
Assimilation of
in situ and
remote sensing
soil moisture
measurements
into rainfall-
runoff modelling
Detecting
rainfall from the
bottom up:
using soil
moisture
observations for
measuring
rainfall
(SM2RAIN)
Storyline
2014
GRL paper
2010
HESS paper
2007
JoH paper
2005 2015

6. Soil moisture monitoring
IN SITU
(TDR, FDR, Gravimetric,
Geophysical methods,
COSMOS, GPS)
REMOTE SENSING
(AMSR-E, AMSR2, SAR, Scatterometers,
ASCAT, SMOS, SMAP...)
HYDROLOGICAL
MODELLING

7. VS
A= ~10-1 m2
satellite
pixels ~25 km
~25 km
A = ~109 m2
in-situ
measurements
~50 cm
~50 cm
HOW IS IT POSSIBLE TO
VALIDATE SATELLITE SOIL
MOISTURE ESTIMATES WITH
IN-SITU MEASUREMENTS?
The scale issue (for RS validation)!
25 August 2015

8. ~25 kmsatellite
pixels
Typical catchment
size for hydrological
studies.
HYDROLOGIST
too coarse for
hydrological
applications !
The scale issue (for hydrology)!

9. Filling the scale gap
COSMOS rover: cosmic-ray neutrons
12 km
12km
22 surveys in 5 months:
~300 measures/5 hours
Also GPS (see Kristine Larson), Geophysics methods (EMI, Resistivity)

10. What is the relation
between point and area-
averaged soil moisture
measurements?
PLOT SCALE
400-9000 m2
CENTRALITALY
Brocca et al., 2009 (GEOD)
SMALL CATCHMENT
SCALE ~50 km2
20
25
30
35
40
45
50
20 30 40 50
Mean soil moisture (%)"Representative"sitesoilmoisture(%)
Castel Rigone
Casale Belfiore
Val di Rosa
Brocca et al., 2010 (WRR)
CATCHMENT SCALE
~250 km2
Brocca et al., 2012 (JoH)
USA
Cosh et al., 2006 (JoH)
AFRICA
de Rosnay et al., 2009 (JoH)
ASIA
Zhao et al., 2010 (HYP)
Soil moisture temporal stability
Brocca et al., 2010 (WRR)

11. Soil moisture information content
Simply
matching
mean and
variance
Different land
models show
substantial
differences
“Large differences are typical between soil
moisture estimates from different climate
models […] in modelling studies [], the
temporal anomalies of soil moisture are
usually of greater interest as most of the
informative content of soil moisture data is
not in their absolute values, but in their
temporal dynamics”

12. Absolute soil moisture vs anomalies
ABSOLUTE SOIL MOISTURE
TEMPORAL MEAN: time-invariant component
TEMPORAL ANOMALIES: time-varying component

13. ABSOLUTE SM ANOMALIES RELATIVE SM
Absolute soil moisture vs anomalies
For large scale and spatial
heterogeneous soil moisture
network (France, Spain,
Switzerland, Australia) the
time invariant component
(green bar) is the major
contributor to the total
spatial variance.
Australia France
Italy Spain
Switzerland USA
Spain
Total variability
Time invariant comp. (temp. mean)
Time variant comp. (anomalies)
Covariance
Network size between 200 and 150000 km²
Absolute and anomaly soil moisture
data behave very differently.
How to use this understanding for
remote sensing validation and in
hydrological applications (e.g., data
assimilation)?

14. In situ vs remote sensing
Median correlation ~0.6-0.7
~1500 measurement stations / 40 networks

15. In situ & RS for RR modelling
Satellite vs modelled soil moisture
In situ soil moisture as initial condition of RR modelling
Tramblay et al., 2012 (HESS)
Brocca et al., 2009 (JHE)
In situ soil moisture
measurement at an
experimental plot are
used to set the initial
conditions of an
event-based rainfall-
runoff model with
successfully results.
Satellite and modelled
soil moisture data are in
good agreement for a
period of 25 years!
137
km²
60
km²
13
km²
R²

16. A Simplified Continuous RR model
Advantages
1) No need of continuous rainfall and
evapotranspiration datasets.
Good in poorly gauged areas!
2) Parsimony and simplicity.
Good for operational purposes!
Applications to:
- 35 catchments in Italy for
National Department of
Civil Protection
- in Greece for FLIRE (Life+)
project
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38
0
10
20
P[mm/h] rainfall
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38
0
50
100
150
t [h]
Q[m3
/s]
Qobs
QSM
in situ
QSMASCAT
QSMERA-LAND
QMISDc

17. In rainfall-runoff modelling …
In the last decades a number studies performed data assimilation experiments
and tested different techniques and approaches for soil moisture assimilation
within rainfall-runoff modelling …
In situ soil moisture Satellite soil moisture
Loumagne et al., 2001 (HSJ) Pauwels et al., 2001, 2002 (JoH, HYP)
Aubert et al., 2003 (JoH) Francois et al., 2003 (HSJ)
Anctil et al., 2008 (JoH) Crow et al., 2005 (GRL)
Brocca et al. 2009 (JHE) Brocca et al., 2010, 2012, 2013 (HESS, IEEE TGRS, IGARSS)
Lee et al. 2011 (AWR) Draper et al., 2011 (HESS)
Matgen et al. 2011 (AWR) Chen et al., 2011, 2014 (AWR, JHM)
Massari et al., 2014 (HESS) Matgen et al., 2012 (AWR)
Alvarez-Garreton et al., 2014, 2015 (JoH, HESS)
Wanders et al., 2014 (HESS)
Lievens et al., 2015 (RSE)
Corato et al., 2015 (RSE)
Soil moisture data assimilation
From 2010
However, few studies demonstrated
the value of assimilating real soil
moisture data for improving runoff
prediction and there are still many
controversial issues to be solved…

18. Data Assimilation ingredients
Bias Handling
1) Variance matching
2) Least square rescaling
3) Cdf matching
4) Triple collocation
Filtering
1) Soil water index
(Swi)
2) Others
3) No filtering
Rainfall runoff
model
1) Lumped
2) Distributed
3) Single layer
4) Multiple layers
Assimilation technique
1) Variational
2) Sequential
Observations
1) In situ
2) Satellite data
3) Land surface model data
Observation error
1) Temporal variability
of the obs. error
2) Spatial correlation
between the
observations
3) Masking
“Cooking” techniques
The problem is often not the
ingredients but the cooking
technique …
Model error
1) Model error covariance estimation (i.e. EnkF: ensemble size)
2) What to perturb. (parameters, inputs, states etc …)
3) How to perturb (amount of perturbation)
A complex recipe?

19. Bias
handling
Filtering Temporal
variability
Spatial
variability
What to
perturb
Bias
correction
Ensemble
size
Ensemble
verification
Given 1 RR model (e.g., HBV), 1 observation
dataset (e.g., SMOS), and 1 assimilation
technique (e.g., EnKF), we can obtain 2300
different results!!!
The task can be even more
difficult if we consider different
catchments, climatic, soil, land use
conditions, ….
… for a complex topic?
Only changing the cooking techniques

20. Tiber River Basin
Basin Area (km2)
Tevere at Ponte Felcino 2080
Nestore at Marsciano 725
Chiani at Morrano 457
Topino at Bevagna 440
Marroggia at Azzano 258
Niccone at Migianella 137
Rainfall-runoff
data from 1989 at
hourly time
resolution
6 sub-
catchments
(140-2080 km²)
A systematic study…

21. …toward data assimilation guidelines

22. RAINFALL SOIL MOISTURE
The soil moisture variations are strongly related to the amount of rainfall falling into the soil.
Therefore, we can use soil moisture observations for estimating rainfall by considering the “soil
as a natural raingauge”.
Doing hydrology backward

23. Is it raining?
radar raingauge
Remote
sensing of
rainfall
TOP-DOWN PERSPECTIVE
BOTTOM-UP PERSPECTIVE: CAN WE USE SOIL MOISTURE DATA TO INFER
THE AMOUNT OF WATER FALLING INTO THE SOIL?
“Top down” vs “bottom up”

24. Ptrue=94 mm
With only two
overpasses the
bottom up approach
provides a better
estimate of the
accumulated rainfall
Pbottom-up=(92-2)= 90 mm
5 0 2 8 The
underestimation is
due to the satellite
overpasses in
period with low
rainfall
Ptop-down=(5+0+2+8)*4= 60 mm
2
92
“Top down” vs “bottom up”

25. precipitation
surface runoff
evapotranspiration
drainage
soil water
capacity
relative saturation
Inverting for p(t):
= soil depth X porosity
Assuming: + +
during rainfall
Soil water
balance
equation
SM2RAIN algorithm

26. SM2RAIN dataset from ASCAT, 0.25°, 2007-2013, freely available
SM2RAIN papers … so far!

27. calibration validation
0.75<R<0.95
In situ soil moisture observations
R
fRMSE fRMSE
R
Application to in situ observations …

28. Correlation map between 5-day rainfall
from GPCC and the rainfall product
obtained from the application of
SM2RAIN algorithm to ASCAT, AMSR-E
and SMOS data plus TMPA 3B42RT
(VALIDATION period 2010-2011)
… and to satellite data: global scale

29.  2007-2009
 ERA-Interim
as benchmark
 5-day
cumulated
The correlation
is 25% higher
than TMPA real
time rainfall
product
0.504  0 .640
Integration of multiple datasets
SM2RAIN (ASCAT+QUIKSCAT)
TMPA (3B42RT)
Median correlation (+/- 50° lat. band) = 0.640
Median correlation (+/- 50° lat. band) = 0.504 TOP-DOWN
BOTTOM-UP

30. Time step: 1-day
Bottom up + Top down
TOP-DOWN
BOTTOM-UP TOP-DOWN
BOTTOM-UP
Central Italy: R=0.86

31. Future directions …
Improving, testing, and integrating NEW monitoring
techniques able to provide soil moisture measurements at
catchment scale: COSMOS, GPS, Electromagnetic induction,
Remote sensing (e.g., SMAP), …
Investigating the assimilation of in situ and satellite soil
moisture observations in rainfall-runoff modelling for
different basins, climates, …
… also in contrast with conventional hydrological approaches
(e.g., assimilation of river discharge)
SM2RAIN: from research to operational applications, thanks
to funding from new research project starting in September:
ESA SMOS+rainfall, ESA CCI, EUMETSAT H-SAF

32. … and open issues
How to reduce the spatial scale gap between in situ
measurements, modelling, and remote sensing?
What is the role of soil moisture spatial variability? Absolute
soil moisture or temporal anomalies? Spatial or temporal
variability? Surface or root-zone measurements?
How much improvement can we expect from using in situ and
satellite soil moisture observations in hydrological
applications?
Is it really useful? What is the role of soil moisture spatial
variability?
Are we able to model/simulate soil moisture spatial
variability?
Models usually provide good simulation for soil moisture
temporal evolution, but not in space

33. This presentation is available for download at:
http://hydrology.irpi.cnr.it/repository/public/presentations/2015/ Wageningen-l.-brocca
FOR FURTHER INFORMATION
URL: http://hydrology.irpi.cnr.it/people/l.brocca
URL IRPI: http://hydrology.irpi.cnr.it