Research Topic Hydrosphere

Research Topic: Hydrosphere Quantifying water cycle dynamics for hydrological prediction

C. Montzka, J. Bendix, H. Bogena, P. Dietrich, A. Güntner, I. Hajnsek, H.-J.

Hendricks Franssen, S. Itzerott, T. Jagdhuber, H. Kunstmann, A. Lehmann,

L. Samaniego, M. Schwank, S. Suchandt, H. Vereecken, J. Wickert

Goals of the Research Topic Hydrosphere

• Retrieval and sensor data fusion: How

can we optimally retrieve and

synergistically merge different remote

sensing technologies in order to improve

hydrological variables/parameters?

• Validation: How precise are hydrological

variables observed using remote sensing

technologies at different spatial and

temporal scales?

• Data assimilation: How can remotely

sensed hydrological products be further

used in order to better understand and

predict the hydrological cycle? SEITE 5

⇒ Quantitative understanding of dynamic processes in the Hydrosphere

⇒ Strong focus on soil moisture

Why is soil moisture important for hydrology?

•Plays an essential role in the prediction of …

• regional weather (lower atmospheric boundary)

SEITE 6

Soil moisture remote sensing data not yet routinely included in operational forecast models (e.g., ECMWF)

Tem

pera

ture




• floods (antecedent soil moisture)

SEITE 7

Cedar Rapids, Iowa, USA, 13-06-2008


SEITE 8




• impact of droughts (reduced plant productivity)


SEITE 9





• groundwater recharge (sustainable WRM)


SEITE 10





• groundwater recharge (sustainable WRM)

⇒Exhaustive characterization of soil moisture in very high spatial and temporal resolution

⇒Better hydrological, meteorological and climatic predictions

Where are we now and what are the problems?

SEITE 11

In-situ soil moisture measurements:

150 Sensor units 18 Router units 900 Soil moisture sensors 300 Temperature sensors


In-situ soil moisture measurements:

• Only few measurements (∼dm3), high spatio-temporal variability

• Hardly long time series

• No standard procedure for intermediate scale (∼ hm2)

SEITE 12


Available remote sensing offers large scale data, but with limitations:

Model (~200m)

SMOS (~35km) ASCAT (~25km)

High temporal resolution, but low spatial resolution

Not adequate for integration into regional scale models

SEITE 13



⇒ Large differences between prediction and measurements

⇒ No relevance for utilization in hydrological models by data assimilation

46 d

ays

te

mpo

ral r

esol

utio

n (A

LOS)

Top soil

20 cm depth

SEITE 14



⇒ Small differences between prediction and measurements

⇒ Improved estimation of soil moisture profile in the root zone by integration into hydrological models

1 da

y

tem

pora

l res

olut

ion

Top soil

20 cm depth

SEITE 15



• Model needed for translating measurement in soil moisture (uncertainty, bias)

• Disturbances: vegetation, surface roughness

• Lack of ground truth verification

• Infrequent visits (e.g., ALOS) OR

• Low spatial resolution (e.g., SMOS, ASCAT)

• Limited to upper 5 cm of soil

SEITE 16

Remote sensing for improving hydrological predictons:

⇒ Spatio-temporal scales of Tandem-L observations move us forward to enhanced hydrological predictions

SEITE 17

Synthetic study, (potential of data)

Small well controlled studies with real data

Large scale studies incl. verification of improved predictions

Operational use for improving large scale predictions

WORK PACKAGES

SEITE 18

WP H1: Derivation of surface soil moisture under the vegetation cover using multi-parametric SAR Jagdhuber & Hajnsek (DLR-HR)

20

H-pol V-pol

Soil

Vegetation

Azimuth

Nadir

Range

Surface Volume Dihedral Scattering

Soil

Removal of Vegetation Component and Inversion for Soil Moisture

Surface Soil moisture

Ground: Surface & Dihedral

- = + Vegetation Total Scattering

Polarimetric Decompositions

19. April Land use

= Mask for forested and urban areas

Validation with in situ data for all crops

RMSE=5.22vol.% 0

9

18

27

36

45

[vol.%]

WP H2: Multi-scale ground measurements data for the validation of remotely sensed soil moisture products - Bogena & Montzka (FZJ)

Spat

ial s

cale

WP H3: GNSS based soil moisture measurements – Wickert & Güntner (GFZ)

SEITE 22

Global GNSS network

GNSS antenna

Correlation of the GNSS multipath signal (L-band) (reflected from ground) with soil moisture

Sutherland, South Africa, 2013

WP H4: Ocean Surface Current Measurement with SAR along-track Interferometry - Suchandt & Lehmann (DLR-IMF/GEOMAR)

SEITE 23

Tidal currents at Orkney Islands measured with TerraSAR-X (2009, 2010)

3 m/s 3 m/s

23.11.09 26.12.09 12.11.09 26.04.10

Surface current data processing

system

SEITE 24

Understanding & modeling the ocean response to climate variability and

extremes is essential for climate projection studies. Models must be validated by

observations.

Space borne SAR Along-track interferometry provides large-area information

on ocean surface motion with high spatial and velocity resolution.

Main objective: Combination of remote sensing data, models and

in-situ measurements to obtain climatologic relevant information

Milestones:

Adaption & validation of TerraSAR-X data processing system

Cross-data analysis for a pilot study area: in-situ data (ADCP),

numerical circulation model (BSIOM), surface current velocities (TanDEM-X)

TanDEM-L mission requirements for ocean current measurements

Tidal currents measured

by TerraSAR-X

3 m/s 3 m/s

WP H4: Ocean Surface Current Measurement with SAR along-track Interferometry - Suchandt & Lehmann (DLR-IMF/GEOMAR)

WP H5: Airborne active and passive microwave data fusion for soil moisture retrieval - Montzka & Bogena (FZJ)

Das, Entekhabi and Njoku (2011) F-SAR backscatter overlayed with PLMR brightness temperature

Backscatter (3 km)

Soil moisture (36 km)

Soil moisture (9 km)

The SMAP active/passive fusion approach

SEITE 25

WP H6: Demand-driven observation of soil moisture - Dietrich & Wollschläger (UFZ)

Aim: provide demand-driven, detailed information on near surface to deeper vadose zone soil water content and subsurface architecture Approach: improve process understanding and predictive capability by demand generated through:

- interaction with modeling and data assimilation - interaction with ground based measurement - consideration of underlying characteristics of observed processes

for calibration and gathering of additional data suitable for processing of remote sensing data

predictions and decisions mobile platforms

permanent installations

SEITE 26

Milestones: • Parameters influencing signal and definition of test sites and methods • Parameter estimation and analysis for 3D model of plants • Transfer of influence parameter knowledge to airborne data processing • Validated method for soil moisture estimation from HSS and radar data

Optical Hyperspectral

SAR

Platform airborne/spaceborne airborne/spaceborne Radiation reflected sunlight own radiation Spectrum visible/infrared microwave Frequency hyper-frequency Several frequencies (C,

L) Polarimetry N.A. polarimetric phase Interferometry N.A. interferometric phase Acquisition time day time day/night Weather blocked by clouds see through clouds Influenced by vegetation vegetation

polarimetric decomposition and inversion

Detection of

)()(

21191800

21191800

RRRRNSMI

+−

=

NSMI (Haubrock et. al 2007)

Detection of promising band ratios

Bron

ster

t et.

al 2

012

WP H7: The use of hyperspectral optical and L-band radar data for retrieving surface soil moisture – Itzerott (GFZ)

WP H8: Data assimilation of multi-scale soil moisture - Hendricks Franssen (FZJ)

Assimilating SMOS, SMAP, Tandem-L, CRP, TDR: scale dependence.

Multi-scale data assimilation with two different algorithms will be tested in synthetic and real world application:

Prior downscaling

Utilization of retrieval operator for

downscaling during assimilation

Main Result: Value of Tandem-L data for improving predictions with land surface models.

One possible multi-scale DA approach (Montzka et al., 2012)

SEITE 28

WP H9: Assimilation of soil moisture and snow water into the mesoscale hydrologic model - Samaniego (UFZ)

Samaniego et al. 2011

Soil moisture index at 2011-04-30

Wet Dry

Research Questions

• Assimilation SM & SWC → minimize parameter uncertainty of mHM?

• Subgrid-scale variability of state variable → better effective parameters at larger scales (e.g. 1-5 km)?

• Statistical data exploration (dimensionality, stochastic dependency)

• Data assimilation, parameter inference, and model improvement

• Evaluation at TERENO sites

Phases

SEITE 29

WP H10: High resolution soil moisture parameterization of land surface models – Bendix & Thies (Uni Marburg)

Investigation of growth and competition processes of pasture grass (setaria sphacelata) and southern bracken in southern Ecuador

Adaptation of the Community Land Model

Soil moisture information with high spatio-temporal resolution by Tandem-L:

Improved model accuracy

Enhanced understanding of growth and competition processes

Development of DA scheme for level 3 Tandem-L soil moisture product

1. Development of a downscaling method for SMOS and SMAP soil moisture product to 50m Tandem-L resolution

2. Model validation (EC data)

SEITE 30

Location of the study site and vegetation units from satellite classification

Göttlicher et al. 2011

Curatola et al. 2012

WP H10: High resolution soil moisture parameterization of land surface models – Bendix & Thies (Uni Marburg)

SEITE 31

WP H11: Copula Based Merging of Modeled and Satellite-Derived Soil Moisture Fields and Ecosystem Fluxes – Kunstmann & Lorenz (KIT)

Problem: Inconsistencies in the spatial and temporal resolution as well as spatial and temporal availability of current (large-scale) soil moisture products.

Main objective: Data merging of satellite- and airbourne-based, in-situ, and modeled soil moisture in order to provide reliable and consistent estimates of gridded soil moisture.

Methods: Application of Copula-techniques in order to describe the dependency between different sources of soil moisture and other climatic variables (e.g. temperature, precipitation, …).

Data assimilation based on the statistical dependency between the different input-datasets.

Empirical (left) and theoretical (right) Copula density between observed and modeled soil moisture

First experiments over the Mississippi basin using SMOS, NLDAS-data and observations from the ISM-Network

Observed (black) and modeled (gray) soil moisure over five observation stations of the ISM-Network

WP H11: Copula Based Merging of Modeled and Satellite-Derived Soil Moisture Fields and Ecosystem Fluxes – Kunstmann & Lorenz (KIT)

L-band test data sets

DLR F-SAR on Do228

SEITE 34

TERENO – Terrestrial Environmental Observatories

Expected impact

• A strong improvement in the characterization of soil moisture contents and associated hydrological and energy fluxes:

• A better understanding of what is exactly measured and therefore a better conversion of measured signals of individual products (e.g., radar, microwave) to soil moisture.

• The spatial and temporal interpolation of soil moisture products to achieve a higher resolution (fusion, downscaling).

• Additional interpolation and extrapolation in space and time (e.g., root zone soil moisture), making use of land surface models and data assimilation techniques.

• Insights in the relative value of different data products.

• Main impact: improved hydrological, meteorological and climatic predictions and sufficient understanding for operational use.

SEITE 35

Direction towards multi-sensor approaches Link with passive sensors / multi-sensor platform efforts to link active and

passive sensors (SMAP-mission)

Link with hyperspektral sensors -> hybrid soil moisture inversion

Link with GNSS reflectrometry-based soil moisture patterns

Link with ground-based measurements (pattern comparison with geophysical techniques)

Direction towards data assimilation into models Assimilation

Direct input

Link through common test sites

Links and joint efforts with project partners: Within Hydrosphere and beyond

Scientific questions

• Does the assimilation of soil moisture help to minimize parameter uncertainty of

hydrological models?

• To what extent does the assimilation of subgrid-scale (e.g. 50-500 m) variability of

state variables contribute to the derivation of effective parameters at larger scales

(e.g. 1-5 km) in order to characterize hydrological processes?

• Is Tandem-L able to improve the regionalization of land surface models?

• What are further techniques to be used for a global soil moisture monitoring and

validation system in conjunction with remote sensing products?

• Can an improved land surface model enhance our understanding about soil-

vegetation – atmosphere water fluxes and coupled plant water/nutrient dynamics?

SEITE 37

Scientific questions…

You may find first answers to these questions in the Hydrosphere session on

Wednesday 9th October, 9:00 – 11:45

and the Poster session on

Tuesday 8th October, 15:00

SEITE 38

Documents

Research Topic Hydrosphere