Recent applications of GRACE gravity data for continental hydrology Andreas Güntner Helmholtz...

Post on 04-Jan-2016

213 views 0 download

Tags:

transcript

Recent applications of GRACE gravity data for continental hydrology

Andreas Güntner

Helmholtz Centre PotsdamGFZ German Research Centre for Geosciences

Andreas Güntner | GRACE for continental hydrology 2

Water storage variations from time-variable gravity data

Temporal variations of the gravity field of the Earth

Water mass variations on the continents after removal of other mass components S: Water storage change

P: PrecipitationE: EvaporationQ: Runoff

ΔS = P - Q - E

Only integrative and large-scale measurement of ΔS for hydrology

Andreas Güntner | GRACE for continental hydrology 3

11/2011: About 200 ISI paper on GRACE and continental hydrology

Water storage variations 169

- Total water storage 104

- Groundwater 25

- Inland glaciers 13

- Surface water storage 19

- Snow 6

Water balance, other variables 33

Evaluation of hydrological models 31

- Model calibration / data assimilation 7

GRACE processing, filtering 54

Main focus of GRACE hydrology papers

Andreas Güntner | GRACE for continental hydrology 4

11/2011: About 200 ISI paper on GRACE and continental hydrology

Studies on water storage variations for particular river basins

Andreas Güntner | GRACE for continental hydrology 5

ET = P - Q - ΔS

Water cycle components from GRACE data - Resolving for evapotranspiration

Ground and/or satellite-based data GRACE

Andreas Güntner | GRACE for continental hydrology 6

ET = P - Q - ΔSWater cycle components from GRACE data - Resolving for evapotranspiration

Moiwo et al. (2011), Hydr.Sci.J.

Hai River Basin, North China (320 000 km²)

ETWH: Model-based ET using remote sensing data

ETGP: GRACE-based ET

Andreas Güntner | GRACE for continental hydrology 7

Water cycle components from GRACE data - Resolving for continental runoff

ΔS = P – Q - ET

Atmospheric water balance

ΔW = C + ET - P

Terrestrial water balance

S Land water storage changeP PrecipitationET EvaporationQ RunoffW Atmospheric water storage changeC Water vapour convergence

Q = -ΔW + C - ΔS

Combined atmospheric-terrestrial water balance

Andreas Güntner | GRACE for continental hydrology 8

Water cycle components from GRACE data - Resolving for continental runoff

Syed et al. (2007), GRL

+ includes ungauged river basins

+ includes groundwater discharge into oceans

Total continental discharge of the Pan-Arctic drainage area

Andreas Güntner | GRACE for continental hydrology 9

Water storage variations from time-variable gravity data

ΔTWSGRACE = ΔSgroundwater + ΔScanopy + ΔSsnow + ΔSsoil + ΔSlakes + ΔSwetlands + ΔSriver

GRACE-based total water

storage variations ΔTWSGRACE

are a composite of various

continental water storage

compartments

Andreas Güntner | GRACE for continental hydrology 10

GRACE hydrology studies with focus on lake water balances

9 studies among 200 ISI papers on GRACE and continental hydrology (11/2011)

Andreas Güntner | GRACE for continental hydrology 11

GRACE hydrology studies with focus on surface water dynamics(river flow, floodplains, inundation areas)

15 studies among 200 ISI papers on GRACE and continental hydrology (11/2011)

Andreas Güntner | GRACE for continental hydrology 12

GRACE hydrology studies with focus on inland glaciers

13 studies among 200 ISI papers on GRACE and continental hydrology (11/2011)

Andreas Güntner | GRACE for continental hydrology 13

GRACE hydrology studies with focus on groundwater storage variations

25 studies among 200 ISI papers on GRACE and continental hydrology (11/2011)

Andreas Güntner | GRACE for continental hydrology 14

Water storage variations from time-variable gravity data

ΔSgroundwater = ΔTWSGRACE + ΔScanopy + ΔSsnow + ΔSsoil + ΔSlakes + ΔSwetlands + ΔSriver

Resolving GRACE-based total

water storage variations

ΔTWSGRACE for single storage

compartments

• Other compartments can usually be estimated based on hydrological / land surface model data only

• Other compartments may not be fully accounted for in models

• Uncertainties / errors accumulate in the variable of interest

Andreas Güntner | GRACE for continental hydrology 15

WaterGAP Global Hydrology model (WGHM)

ISBA-TRIP

Global Land Data Assimilation System (GLDAS)

ΔTWS = ΔScanopy + ΔSsnow + ΔSsoil + ΔSgroundwater + ΔSrivers + ΔSlakes/reservoirs + ΔSwetlands

ΔTWS = Δ Scanopy +ΔSsnow + ΔSsoil + ΔSgroundwater + ΔSrivers

ΔTWS = ΔScanopy + ΔSsnow + ΔSsoil

Soil depth = root zone

Soil depth = root zone + deep soil layer

Soil depth GLDAS-CLM = 3.43 mGLDAS-MOSAIC = 3.50 mGLDAS-NOAH = 2.00 mGLDAS-VIC = 1.90 m

Water storage compartments from hydrological modelsfor GRACE TWS signal separation

Andreas Güntner | GRACE for continental hydrology 16

Relevance of deep unsaturated zone water storage for GRACE TWS signal separation

Snow

Soil 0-30cm

Soil 30-150cm

Saprolith 1.5 – 11m

Groundwater > 11m

Creutzfeldt et al., 2010, WRR; Creutzfeldt et al., GJI, 2010

Hydrological gravity effect

Superconducting gravimeter residuals

Local gravity effect of water storage compartments

Station Wettzell / Germany

Un

satu

rate

d zo

ne

Andreas Güntner | GRACE for continental hydrology 17

Example: Water storage variations in Central Asian Mountains

Total study area:500 000 km²

Andreas Güntner | GRACE for continental hydrology 18

Example: Water storage variations in Central Asian Mountains

Source: GGHYDRO (Cogley, 2003)

Total study area:500 000 km²

Can we estimate glacier mass changes from GRACE?

Andreas Güntner | GRACE for continental hydrology 19

1) Selection of GRACE product (processing type and centre, filtering)

2) Compensation for filter effects (smoothing, leakage)

Estimating correction function (e.g. rescaling factor)

Hydrological models

3) Reduction of unwanted hydrological signal components

4) Analysis of residuals

Isolation of single water storage compartmentsfrom GRACE TWS data

Andreas Güntner | GRACE for continental hydrology 20

Water storage variations in Central Asian Mountains

Andreas Güntner | GRACE for continental hydrology 21

1) Selection of GRACE product (processing type and centre, filtering)

2) Compensation for filter effects (smoothing, leakage)

Estimating correction function (e.g. rescaling factor)

Hydrological models

3) Reduction of unwanted hydrological signal components

4) Analysis of residuals and error assessment

Isolation of single water storage compartmentsfrom GRACE TWS data

Ensemble of GRACE products

Andreas Güntner | GRACE for continental hydrology 22

Water storage variations in Central Asian Mountains

Andreas Güntner | GRACE for continental hydrology 23

Water storage variations in Central Asian Mountains

Andreas Güntner | GRACE for continental hydrology 24

Compensation for filter effects

Multiplicative scaling factor derived from least-square adjustment

• Mainly sensitive to seasonal dynamics

• Leakage effects (e.g. phase shifts) are not compensated

• Rescaling functions depend on the hydrological model used

• Rescaling functions may not apply for the variable of interest

Andreas Güntner | GRACE for continental hydrology 25

Compensation for filter effects: example Central Asia

Multiplicative scaling factor derived from least-square adjustment

CLM MOSAIC NOAH VIC ISBA-TRIP WGHM

G300 1.47 1.32 1.18 1.29 0.99 0.98

G500 1.70 1.51 1.20 1.65 0.95 1.08

DDK1 1.63 1.54 1.23 1.68 1.02 1.17

DDK2 1.55 1.48 1.20 1.75 0.91 1.06

DDK3 1.83 1.63 1.21 1.84 0.85 1.06

G300/G500: Gauss filter 300/500 km. DDK: Decorrelation filter by Kusche (2007)

Andreas Güntner | GRACE for continental hydrology 26

Compensation for filter effects: example Central Asia

Multiplicative scaling factor derived from least-square adjustment

CLM MOSAIC NOAH VIC ISBA-TRIP WGHMWGHM

No surface storage

G300 1.47 1.32 1.18 1.29 0.99 0.98 1.11

G500 1.70 1.51 1.20 1.65 0.95 1.08 1.45

DDK1 1.63 1.54 1.23 1.68 1.02 1.17 1.58

DDK2 1.55 1.48 1.20 1.75 0.91 1.06 1.23

DDK3 1.83 1.63 1.21 1.84 0.85 1.06 1.45

G300/G500: Gauss filter 300/500 km. DDK: Decorrelation filter by Kusche (2007)

Andreas Güntner | GRACE for continental hydrology 27

1) Selection of GRACE product (processing type and centre, filtering)

2) Compensation for filter effects (smoothing, leakage)

Estimating correction function (e.g. rescaling factor)

Hydrological models

3) Reduction of unwanted hydrological signal components

4) Analysis of residuals and error assessment

Isolation of single water storage compartmentsfrom GRACE TWS data

Ensemble of GRACE products

Carefully consider particular region and model differences

´ (e.g., Werth et al. 2009,Longuevergne et al. 2010)

Andreas Güntner | GRACE for continental hydrology 28

Reducing GRACE mass variations in Central Asian Mountainsby water storage from hydrological models

- 7 GRACE products

- 5 different filters

- 6 different rescaling values for each filter

- 6 different LSMs / hydrological models for signal separation

→ bootstrapping approach

Andreas Güntner | GRACE for continental hydrology 29

GRACE mass variations in Central Asian Mountains after reducing for model-based TWS

792 realisation of different plausible GRACE products, rescaling factors and hydrological reduction models

Trend-0.2 ± 5.7 mm/a

Andreas Güntner | GRACE for continental hydrology 30

GRACE mass variations in Central Asian Mountains after reducing for model-based TWS

792 realisation of different plausible GRACE products, rescaling factors and hydrological reduction models

Trend+13.9 mm/a

Trend-12.8 mm/a

Andreas Güntner | GRACE for continental hydrology 31

Conclusions and perspectives

• Caveats in using single GRACE products, filter and correction methods or hydrological model data sets→ use ensemble approach

• Multi-sensor applications of GRACE (in conjunction with, e.g., altimetry, satellite-based snow, soil moisture and ET products) for assessing dynamics of continental hydrology and signal decomposition

• Extended use of GRACE to inform structure and parameterization of land surface / hydrological models