GRACE and geophysical applications Annette Eicker Institute of Geodesy and Geoinformation University of Bonn Petra: PT coeff humid? (alpha?) Das erste Mal Assimilierung in WGHM? TexPoint fonts used in EMF. Read the TexPoint manual before you delete this box.: AAAAAAAAA
Outline The GRACE mission - observation principle What are the challenges? Applications (examples)
The GRACE mission
Gravity Recovery and Climate Experiment GRACE: launch: March 2002 altitude: ~450 km distance: ~250 km orbit period: 94 min polar orbit JPL
GRACE Observations distance: ca. 250 000 000 000 μm accuracy: 12 digits Comparison: average human hair: 0.05 mm => 50 μm (keep in mind: speed: 27.500 km/h) K-band microwave ranging instrument accuracy: < 1 μm GPS receiver: accuracy: 2-3 cm
GRACE results: gravity field Long-term mean gravity field (differences of gravity compared to ellipsoid) gravity anomalies [mGal]
GRACE results: gravity field Long-term mean gravity field (differences of gravity compared to ellipsoid) Temporal variations e.g. monthly models
GRACE results: gravity field First time observation of temporal gravity field variations on the global scale! New measurement type for hydrology, oceanography, glaciology, geophysics… download GRACE solutions: http://icgem.gfz-potsdam.de/ICGEM/ Temporal variations e.g. monthly models
But there are also some challenges….
Outline The GRACE mission - observation principle What are the challenges? Applications (examples)
Challenges Why can GRACE data be a little difficult? 1) GRACE observes the gravity field from far away => Downward continuation 1)
Upward/downward continuation Satellite altitude [mgal] Gravitational potential in spherical harmonics 6378 km 6378 + 450 km 450 km (10.000 km) (160 km) (110 km) Dampening factors Ground level [mgal]
Upward/downward continuation Satellite altitude c observation noise dampening of small features (high frequencies) GRACE sees only smoothed version of gravity field Amplification of (high frequency) noise [mgal] noise Ground level [mgal]
Challenges Why can GRACE data be a little difficult? 1) GRACE observes the gravity field from far away => Downward continuation The gravity field changes continuously, but it takes time to collect the data => Aliasing 2)
Ground tracks
GRACE It takes time to collect satellite data, but the gravity field changes continuously 1 day 15 days 30 days
GRACE It takes time to collect satellite data, but the gravity field changes continuously tides: tidal forces (sun, moon, planets) Earth tides ocean tides atmospheric variations and the reaction of the ocean pole tides Short-periodic gravity changes: atmospheric variations
GRACE It takes time to collect satellite data, but the gravity field changes continuously tides: tidal forces (sun, moon, planets) Earth tides ocean tides atmospheric variations and the reaction of the ocean pole tides Short-periodic gravity changes: ocean tides
GRACE It takes time to collect satellite data, but the gravity field changes continuously Short-periodic gravity changes: Models: JPL DE405 IERS2003 EOT11a AOD1B OMCT tides: tidal forces (sun, moon, planets) Earth tides ocean tides atmospheric variations and the reaction of the ocean pole tides Short-term gravity variations have to be reduced BUT: every model has errors
Aliasing residual signal (after reduction of models) monthly models unmodelled variations Undersampling of the short-term variations => „Aliasing“ This results in …
Monthly solution 2007 - 04 Why north-south stripes? very high measurement accuracy along-track sampling along the orbit => gravity field (unmodelled short-periodic effects) might have changed completely between neighboring arcs
Challenges Why can GRACE data be a little difficult? 1) GRACE observes the gravity field from far away => Downward continuation 2) The gravity field changes continuously, but it takes time to collect the data => Aliasing We have to do something about the noise => Filtering, Leakage 3)
Filtering Gaussian filter 2007 - 04 water height [cm]
Filtering Gaussian filter Filter: 200 km 2007 - 04 water height [cm]
Filtering Gaussian filter Filter: 250 km 2007 - 04 water height [cm]
Filtering Gaussian filter Filter: 300 km 2007 - 04 water height [cm]
Filtering Gaussian filter Filter: 400 km 2007 - 04 water height [cm]
Filtering Gaussian filter Filter: 500 km 2007 - 04 water height [cm] stronger filtering => less noise But: filtering implies spatial averaging also of the signal => „Leakage effect“
(unfiltered, without noise) Leakage modelled signal (unfiltered, without noise) signal (500 km Gauss filter) stronger filtering => less noise But: filtering implies spatial averaging also of the signal => „Leakage effect“
(unfiltered, without noise) Leakage river basin average: 7.4 cm/year 4.0 cm/year modelled signal (unfiltered, without noise) signal (500 km Gauss filter) stronger filtering => less noise But: filtering implies spatial averaging also of the signal => „Leakage effect“
Leakage Leakage Leakage-Out Leakage-In Filter Signal original Region of interest Leakage-Out Leakage-In stronger filtering => less noise But: filtering implies spatial averaging also of the signal => „Leakage effect“
Leakage Leakage Leakage-Out Leakage-In Filter Signal original Region of interest Leakage-Out Leakage-In Generally damping of signal in region of interest => underestimation of amplitude => Estimation of re-scaling factor to obtain full signal. (Can be difficult!)
Challenges Why can GRACE data be a little difficult? 1) GRACE observes the gravity field from far away => Downward continuation 2) The gravity field changes continuously, but it takes time to collect the data => Aliasing 3) We have to do something about the noise => Filtering, Leakage GRACE observes the integral mass signal => Loading, Signal separation 4)
Loading GRACE measures gravitational potential => conversion to mass Mantle Crust Mass But: GRACE has no depth perception
Loading mass gain mass loss GRACE measures gravitational potential => conversion to mass Mass But: GRACE has no depth perception Crust Signal separation of variations at surface and in mantle using loading theory Mantle gravity field coefficients equivalent water height elastic response of the Earth (load love numbers)
Signal Separation Hydrology Ice Atmosphere Ocean Separation of integral mass signal: Reduction of unwanted signals using models => Model errors included in mass estimate statistical / mathematical approaches (e.g. PCA, ICA) GIA Mantle and Crust
Outline The GRACE mission - observation principle What are the challenges? Applications (examples)
GRACE results (ITG-Grace03) Already reduced: tides (ocean, Earth, …), atmosphere & ocean variations
water height [cm/year] Trend and amplitude Annual amplitude Trend water height [cm/year] water height [cm]
Annual amplitude water height [cm]
Hydrology 1gt = 1km³ water! Amazon [giga tons]
Hydrology 1gt = 1km³ water! Amazon [giga tons] equator Orinoco
Hydrology GRACE time series provide valuable information to hydrologists WHY? Improvement of global hydrological models canopy snow soil groundwater local lakes local wetlands river global wetlands WaterGAP: models water storages and flows on 0.5° x 0.5 ° grid global lakes (Döll et al 2003) Problems: model physics, insufficient data coverage (e.g. percipitation)
Hydrology 1gt = 1km³ water! Amazon GRACE [giga tons]
Hydrology 1gt = 1km³ water! Amazon GRACE [giga tons] Underestimation of amplitude in the model WaterGAP Possible solution: model calibration
Hydrology Underestimation of amplitude in the model! GRACE WaterGAP WaterGAP calibrated (Werth et al. 2009) Underestimation of amplitude in the model! Possible solution: model calibration
Trend 47
India
India GRACE Groundwater withdrawal seems to be detectable by GRACE Rodell et al. (2009), Nature
India Why is this so important? Groundwater withdrawal seems to be detectable by GRACE e.g. altimetry e.g. SMOS canopy surface waters soil snow groundwater For the first time it is possible to observe groundwater changes from space
Trend Greenland Antarctica Alaska 51
Trend 52
Greenland Mass loss in Greenland as observed by GRACE Mass loss: ~ 240 Gt/year (GFZ-RL05 time series) water height [m] Jakobshavn glacier Acceleration?! water height trend [m/year] Leakage is very important (and difficult) in Greenland!! (ITG-Grace regional solution)
How is the melted ice distributed in the oceans? Sea level change
Sea level What happens when ice is melting in Greenland? Simulated sea level change after 40 years FESOM (Brunnabend et al 2012) Water is being attracted Ice Greenland Ocean Step: Ice generates gravity
Sea level What happens when ice is melting in Greenland? Greenland At same time: - continents rise (reduced loading) sea floor deforms (increased loading) this again changes gravity … Sea level is sinking at the Greenland coast Greenland Ocean Simulated sea level change after 40 years FESOM (Brunnabend et al 2012) Step: Ice generates gravity
Sea level What happens when ice is melting in Greenland? At same time: - continents rise (reduced loading) sea floor deforms (increased loading) this again changes gravity … „Fingerprint“ of Greenland ice melting Sea level is sinking at the Greenland coast Greenland Ocean Simulated sea level change after 40 years FESOM (Brunnabend et al 2012) Step: Ice generates gravity
Contributions to sea level change Ice Sea level trend as observed by GRACE (Riva et al. 2010) 58 Hydrology (Jensen et al. 2013)
Sea level What happens when ice is melting in Greenland? At same time: - continents rise (reduced loading) sea floor deforms (increased loading) this again changes gravity … „Fingerprint“ of Greenland ice melting Sea level is sinking at the Greenland coast Greenland Ocean Simulated sea level change after 40 years Cannot be observed by GRACE FESOM (Brunnabend et al 2012) Step: Ice generates gravity Increase in global temperature heats the ocean (=> volume change) Step: Ice generates gravity
Altimetry 60 Determination of geometric sea level variations
Sea level (NASA) Altimetry observes both mass variations and volume change (steric sea level change) Separation only possible when combining GRACE and altimetry 61 (NASA)
Altimetry Sea level 62 Böning et al. (2012)
Sea level Altimetry Sea level drops 6 mm in 2010 ! What happened here? 63 Sea level drops 6 mm in 2010 ! What happened here? Böning et al. (2012)
Altimetry Sea level 64 Böning et al. (2012)
Sea level Altimetry Water redistribution observed by GRACE 65 Water redistribution observed by GRACE Böning et al. (2012)
Trend per year 66
Glacial isostatic adjustment (GIA) 1 Mantle Crust Ice 3 Mantle Crust Ice 2 Mantle Crust Ice 4 Mantle Crust Viscoelastic response of the Earth
Glacial isostatic adjustment (GIA) GRACE Model (adjusted to GRACE) (Sasgen 2011)
Trend 69
Earthquakes Sumatra-Andaman earthquake (December 2004) Han et al. 2006, Science
Earthquakes Model Sumatra-Andaman earthquake (December 2004) Tohoku earthquake (Fukushima, April 2004) Model GRACE Han et al. 2006, Science Matsuo and Heki, 2011
Summary and outlook The GRACE mission - observation principle GRACE offers many interesting applications, e.g. in hydrology, oceanography, glaciology, geophysics, … but there are still enough challenges left, => we are far from completly undertstanding the data The GRACE mission - observation principle Applications (examples) Outlook: end of GRACE mission lifetime to be expected in the next 1-3 years (batteries are a problem!) GRACE follow-on mission will be launched August 2017 => longer time series, important for climate research GRACE-FO: laser interferometer as technoloy demonstrator => distance between satellites 5-50x more accurate but that does not mean that the gravity field will be equally more accurate