Satellite soil moisture data assimilation into the Australian Water - - PowerPoint PPT Presentation

▶
satellite soil moisture data assimilation into
SMART_READER_LITE
LIVE PREVIEW

Satellite soil moisture data assimilation into the Australian Water - - PowerPoint PPT Presentation

Aug 21, 2022 280 likes •445 views

Satellite soil moisture data assimilation into the Australian Water Resources Assessment modelling system Luigi Renzullo 1 , Brent Henderson 2 , Warren Jin 2 , Jean-Michel Perraud 1 , Matthew Stenson 1 , Albert van Dijk 3 CSIRO LAND AND WATER 1

slide-1
SLIDE 1

Satellite soil moisture data assimilation into the Australian Water Resources Assessment modelling system

Luigi Renzullo 1, Brent Henderson 2, Warren Jin 2, Jean-Michel Perraud 1, Matthew Stenson 1, Albert van Dijk 3

1 CSIRO Land and Water 2 CSIRO Computational Informatics 3 Australian National University, Fenner School

CSIRO LAND AND WATER 6th WMO Symposium on Data Assimilation 7 – 11 Oct 2013, University of Maryland, USA

slide-2
SLIDE 2

Water Resources Information in Australia

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

//www.bom.gov.au/water/

Water balance across Australia

(2000-2006)

Net water using Net water producing

  • Commonwealth Water Act 2007
  • Australian Bureau of Meteorology (BoM)
  • Mandate: ”Manage Australia’s water resources information …”;
  • new responsibilities; new BoM Water Division formed.
  • National water accounts & assessments
  • Water Information Research & Development Alliance
  • WIRADA: An R & D initiative between the BoM and CSIRO;
  • partnership of $50M over 5 years (July 2008 – June 2013)
  • Australian Water Resources Assessment (AWRA) system
  • Comprehensive reconstruction of the water balance for the whole country
  • Scale and accuracy acceptable for water resources management

www.bom.gov.au/water/awra/2012/ www.bom.gov.au/water/nwa/2012/ AWRA Report 2012 NWA 2012

slide-3
SLIDE 3

AWRA system

Australian Water Resources Assessment modelling system

  • Developed CSIRO-BoM for reporting on WRA

and NWA

System model components

AWRA Landscape model (AWRA-L)

– Hybrid land surface model / conceptual RR model – Daily time step – 0.05-degree resolution grid across continent – Top-layer (S0), shallow root (Ss) & deep root (Sd) soil layers

AWRA River model (AWRA-R)

– Node-link model (simplified sourceRivers)

AWRA Groundwater model (AWRA-G)

– Models aquifer dynamics SW-GW processes (incl. lateral transfer between cells, SW-GW interactions, recharge from

  • verbank flows, models impact of extraction, ..)

surface water saturated area ground water surface soil deep root zone shallow root zone precipitation precipitation short-wave radiation short-wave radiation minimum temperature minimum temperature maximum temperature maximum temperature fraction deep-rooted vegetation fraction deep-rooted vegetation available energy ET ET maximum transpiration maximum root uptake vegetation adjustment surface water saturated area ground water surface soil deep root zone shallow root zone precipitation precipitation short-wave radiation short-wave radiation minimum temperature minimum temperature maximum temperature maximum temperature fraction deep-rooted vegetation fraction deep-rooted vegetation available energy ET ET maximum transpiration maximum root uptake vegetation adjustment

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

slide-4
SLIDE 4

Overall goal: Develop and deploy a modelling environment to integrate surface measurement and remote sensing data systems for comprehensive water balance

* e.g. Streamflow, water table, bore data, reservoir data; vegetation indices, soil moisture, land surface temperature

AWRA system: data assimilation objectives

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

20 cm 4 m 6 m 8 m

/ /

2 4 6

Pixel counts

(x10 4)

Available water

0.05 0.10 0.15 0.20 Volumetric (m 3 m -3)

Soil layer thickness

Top soil Shallow root Deep root

Specific goals of this study: Evaluate assimilation satellite soil moisture retrievals on soil water representation in AWRA-L

* Assess active and passive remotely-sensed soil moisture retrievals constraint on AWRA-L top-layer and shallow root-zone moisture estimates. * Evaluate modelling against in situ measurements & cosmic-ray data

Method summary: Sequential updating of AWRA-L model states (soil water storages) using the Ensemble Kalman Filter (EnKF) based on perturbed forcing and triple collocation for errors on satellite soil moisture products

slide-5
SLIDE 5

Soil moisture data sets

  • 2. Technische Universität Wien (TUW) soil relative wetness

products

  • Estimates derived from active microwave ASCAT backscatter

signal using the change detection algorithm (Wagner et al., 1999, Rem. Sens. Environ.)

  • Our holding: 1 Jan 2007 – 31 Dec 2011.
  • Surface degree of saturation (0-1)
  • 0.125 x 0.125 - NN resampling to 0.05 x 0.05
  • Top ~1-2cm soil layer

AMSR-E

Descending ~1.30 am local time pass only

ASCAT

Average of descending ~9pm previous day and ascending ~9 am current day

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

  • 1. Vrije Universiteit Amsterdam (VUA) – NASA soil

moisture products

  • Estimates derived from passive microwave AMSR-E

brightness temperatures using the LPRM algorithm (Owe et al., 2001, IEEE Trans. Geosci. Rem. Sens.)

  • Our holding: 1 Jul 2002 – 30 Sep 2011.
  • Volumetric soil moisture (m3m-3)
  • 0.25 x 0.25 - NN resampling to 0.05 x 0.05
  • Top ~1-2cm soil layer
slide-6
SLIDE 6
  • Pattern and magnitude of errors appear consistent with others work, e.g. Dorigo et al., 2010, HESS

Continental error estimates: using triple collocation (CDF* matched SM obs)

VUA – AMSR-E TUW – ASCAT

Relative wetness

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

slide-7
SLIDE 7

IQR

w0

(median)

a

w0

a − w0 f

Continental satellite DA into AWRA-L

  • Continental AWRA-L data assimilation
  • EnKF using perturbed forcing
  • multiplicative perturbation on rainfall
  • Additive perturbation on air temp and

shortwave radiation

  • Details in Renzullo et al., 2013, J Hydrol. (in prep)
  • Simulations over the last 13 years. (~2-day turn around

AWRA-L Relative wetness for 7 July 2009

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

slide-8
SLIDE 8

Soil moisture assimilation

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 May Jul Sep Nov Jan Mar 0.0 0.2 0.4 0.6 0.8 1.0

Top layer relative wetness Top layer relative wetness Top layer relative wetness

Open loop Assimilation Observations

ASCAT AMSR-E ASCAT & AMSR-E

2011 2011 2010 2010 2010 2010

slide-9
SLIDE 9

Jan Mar May Jul Sep Nov Jan 0.0 0.2 0.4 0.6

Yanco

Soil moisture [m 3 / m -3] Date [2011]

OzNet COSMOZ

Jan Mar May Jul Sep Nov Jan 0.0 0.2 0.4 0.6

AMSR-E x

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

Evaluation @ OzNet CosmOz OzFlux

slide-10
SLIDE 10

Evaluation: AWRA-L top-layer SM estimation

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

  • 10

10 20 30 0.2 0.4 0.6 0.8 10 20 30 0.2 0.4 0.6 0.8 10 20 30 0.2 0.4 0.6 0.8

r a − r 0 / r 0 r 0

AMSR-E ASCAT AMSRE + ASCAT (a) (b) (c)

100 x

r 0 r 0

0.4 0.5 0.6 0.7 0.8 0.9 M_1 M_3 M_5 M_7 Y_2 Y_4 Y_6 Y_8 Y_11 Y_13 Y_A3 Y_A9 Y_B3 Y_10 K_1 K_4 K_6 K_8 K_10 K_12 K_14 A_2 A_4

  • pen loop

AMSR-E ASCAT

  • 45 OzNet top-layer (0-8 cm)

in situ measurement sites

  • Percentage relative difference

between open-loop (r0) and analysis (ra) correlation

  • Correlation between model

and in situ moisture for 1 July 2007 – 31 May 2011

Correlation

slide-11
SLIDE 11

Evaluation: AWRA-L shallow root-zone

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

  • 10

10 20 30 0.2 0.4 0.6 0.8

r a − r 0 / r 0 r 0

0 – 30 cm 0 – 90 cm AMSR-E ASCAT AMSRE + ASCAT (a) (d) (b) (e) (c) (f)

10 30 50 70 90 0.2 0.4 0.6 0.8 10 30 50 70 90 0.2 0.4 0.6 0.8

  • 70
  • 50
  • 30
  • 10

10 30 50 70 90 0.2 0.4 0.6 0.8 10 20 30 0.2 0.4 0.6 0.8 10 20 30

100 x

  • 36 OzNet shallow root-zone (0-30 cm and 0-90 cm) in situ measurements
  • Percentage relative difference between open-loop (r0) and analysis (ra) correlation
  • Correlation between model and in situ moisture for 1 July 2007 – 31 May 2011
slide-12
SLIDE 12

Evaluation: AWRA-L shallow root-zone

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

  • Cumulative distribution of the analysis

increments of the AWRA-L soil water storage states (normalised by the forecast states estimates) pooled across the OzNet site and only for those times when satellite SM were available for assimilation

slide-13
SLIDE 13

Evaluation: AWRA-L shallow root-zone

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

2011 2012 2011 2012 2011 2012 0.0 0.2 0.4 0.6 2011 2012 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6

Baldry Daly Robson Tullochgorum Weany Yanco

2011 2012 2011 2012 Soil moisture (m3 m-3) Soil moisture (m3 m-3) Soil moisture (m3 m-3)

CosmOz AWRA-L (95% ensemble range)

  • Evaluation against cosmic-ray probes (CosmOz)

0.3 0.4 0.5 0.6 0.7 0.8 0.9 500 1000 1500 2000 2500 Baldry Daly Robson Tullochgorum Tumbarumba Weany Yanco

Depth (mm) Correlation

slide-14
SLIDE 14

6th WMO Data Assimilation Symposium | Luigi Renzullo | 7- 11 Oct 2013 Maryland, USA

  • A system for continental-scale data assimilation has been developed for Australian water resources

assessments

  • Currently ingest satellite soil moisture retrievals into AWRA-L
  • Modular in design can be extended to assimilate wider range of gridded data products (e.g. other SM,

evapotranspiration, vegetation indices, …)

  • Further work will focus on coupling the landscape model (AWRA-L) and river model (AWRA-R) for streamflow DA.
  • EnKF method is applied pixel-wise for the whole of Australia
  • Perturbed rainfall, radiation and air temperature
  • Future work will examine spatially-varying perturbations across the continent
  • Model error will be revisited
  • AWRA-L soil moisture evaluation against ground data
  • AWRA-L open-loop simulation of top- and shallow-root layer SM highly correlated with in situ (OzNet)

measurements

  • Soil moisture assimilation improved AWRA-L top-layer estimation when open-loop simulations were less correlated

with in situ data than satellite SM

  • AWRA-L shallow-root zone (0-30 cm) estimates improved almost always after assimilation; 0-90 cm variable (but

biggest improvements when AWRA-L open-loop estimates were poor).

  • CosmOz data promising new network to extend satellite and model evaluation to variety of landscapes around

Australia

  • Future: SM assimilation impact on wider water balance estimation
  • E.g. evaluation of AWRA-L runoff estimates – preliminary findings showed degradation, which points to

inconsistency between soil water and runoff components in AWRA-L which needs to be explored.

Conclusions & future directions

slide-15
SLIDE 15

Thank you

CSIRO Land and Water Luigi Renzullo Senior Research Scientist t +61 2 6246 5758 e Luigi.Renzullo@csiro.au

CSIRO LAND AND WATER / WATER FOR A HEALTHY COUNTRY FLAGSHIP