Arctic terrestrial water storage changes from GRACE satellite estimates and a

land surface hydrology model

Fengge Sua Douglas E. Alsdorfb, C.K. Shumb

Dennis P. Lettenmaiera

aUniversity of Washington, Seattle, WAbThe Ohio State University, Columbus, OH

Background

• Since 2002, the The Gravity Recovery And Climate Experiment (GRACE) satellite mission has provided a basis for estimating spatial and temporal variations of global terrestrial water storage.

• Evaluation of the accuracy of the GRACE terrestrial water storage is complicated by the absence of direct observations of terrestrial water storage.

• Land surface hydrology models, forced with observations, provide an opportunity for evaluating GRACE estimates regionally and globally.

Methodology

GRACE provides: anomaly of monthly total water storage (TWS)

GRACE-derived TWS Change (TWSC):

Hydrology model:

We use Eq(3) and Eq(4) calculated from output of hydrology model

to compare with GRACE TWS and GRACE-derived TWSC, Eq(1).

(2)

(3)

(4)

(1)

REPT

W

1 iii TWSTWSTWSC

1 iii WWTWSC

WWTWS ii

GRACE Data

GRACE data are available from three science processing centers: CSR, GFZ, JPL http://grace.jpl.nasa.gov/data/mass/

We also use the data from OSU (Ohio State University).

Here we use data from at smoothing radii of 300 km.

Data length: 2002.8-2008.8 GRACE resolution: 1°×1°

Hydrology Model

VIC large scale land surface hydrology model.

Study area: pan-Arctic

Simulation period: 2002-2007

Model resolution: 100 km × 100 km

Precipitation forcing: GPCP 1dd (http://ftp.ncdc.noaa.gov/pub/data/gpcp/1dd/doc/ )

Temp and Wind: NCEP/NCAR reanalysis

Land mask of pan-Arctic

Mackenzie

Lena

Yenisei

Ob

Monthly anomalies of basin-averaged TWS estimated from GRACE and VIC

GRACE

VIC

Lena

Yenisei

Ob

Mackenzie

The two estimates show good correspondence in seasonal and inter-annual variation, however the VIC model shows greater amplitude.

Monthly anomalies of TWS

R2=0.79

R2=0.93

R2=0.67

R2=0.86

Monthly basin-averaged total water storage change (TWSC) from VIC and GRACE.

VIC

GRACE

Estimates of TWSC from GRACE and VIC track each other fairly well in the seasonal variation.

Lena

Yenisei

Ob

Mackenzie

Monthly TWSC for the four major Arctic river basins

R2=0.75

R2=0.64

R2=0.73

R2=0.58

Monthly TWS and TWSC over the entire pan-Arctic

GRACE

VIC

R2=0.77

R2=0.88

Annual cycle of monthly basin-averaged P, E, R and TWSC.

Negative TWSC in warm season is mostlyinfluenced by E and R; positive TWSC in cold season is mostly influence by P.

VIC TWSC

GRACE TWSC

Evap

Runoff

Precip

Spatial patterns of mean monthly TWSC from GRACE and VIC over the pan-Arctic

GRACEVIC

Spatial fields of VIC simulated SWE

Snow accumulates

starting from October and

SWE reaches to the

maximum on April.

SWE corresponds to the

positive TWSC for

November-April.

Spatial fields of VIC simulated evaporation and runoff

Evaporation Runoff

Conclusion• The difference among the existing GRACE datasets is much smaller than the

difference between the GRACE estimates and the VIC model simulation.

• All GRACE estimates show good correspondence to the VIC model in the spatial

and temporal variations of TWS/TWSC over the Arctic river basins, while the

VIC model shows greater amplitude. Possible reasons:1) uncertainty from the VIC

model input and the model itself, 2) the smoothing effects on the GRACE data

(smoothing attenuates the real signals).

• Spatial patterns of TWSC from both GRACE and VIC over the pan-Arctic show

dominant positive signals during winter and negative signals during summer.

• The spatial pattern of TWSC can be explained by the VIC model simulation: snow

mostly contributes the positive TWSC signals for the months November-April; E

and R deplete the TWS and contribute to negative signals of TWSC for the

months May-September over the pan-Arctic.

Acknowledgments

• Qiuhong Tang (UW)• Lei Wang (OSU)• Jianbin Duan (OSU)