multi-sensor snow data assimilation · groundwater, soil moisture, snow, and surface water) as an...

Matt Rodell NASA GSFC

Multi-Sensor Snow Data Assimilation

Matt Rodell1, Zhong-Liang Yang2, Ben Zaitchik3, Ed Kim1, and Rolf Reichle1

1NASA Goddard Space Flight Center 2The University of Texas at Austin

3Johns Hopkins University


Title: Multi-Sensor Snow Data Assimilation Problem Statement: MODIS, AMSR-E, and GRACE all provide observations that are relevant to snow water equivalent mapping, each with significant advantages and disadvantages. Hypothesis: Global fields of SWE can be produced with greater accuracy than previously seen by simultaneously assimilating MODIS snow cover, AMSR-E SWE, and GRACE terrestrial water storage observations within a sophisticated land surface model. Team: •  NASA/GSFC: Matt Rodell (PI), Ed Kim, Rolf Reichle •  U. Texas: Liang Yang (Co-PI) •  Johns Hopkins: Ben Zaitchik Timeline: Just getting started

Project Summary


Soil Moisture SWE, Ice, Rainfall Snow Cover

Vegetation Radiation

Remote Sensing of Snow Aqua: MODIS, AMSR-E

GRACE

GRACE provides changes in total water storage, which are dominated by SWE at high latitudes and altitudes, but at very coarse resolution

MODIS provides high resolution snow cover data but not SWE; AMSR-E provides SWE data but with significant errors where snow is deep or wet

Terra: MODIS


Data Integration within a Land Data Assimilation System (LDAS)

INTERCOMPARISON and OPTIMAL MERGING of

global data fields

Satellite data products used to PARAMETERIZE and FORCE land surface models within LIS

ASSIMILATION of satellite based land surface state fields (snow, soil moisture,

etc.)

Ground-based observations used to EVALUATE model output

SW RADIATION

MODIS SNOW COVER

PRECIPITATION

SNOW WATER EQUIVALENT


MODIS Snow Cover Assimilation

Observations

Mosaic LSM

Control Run SWE (mm)

MODIS Snow Cover (%)

Assimilated SWE (mm)

IMS Snow Cover Observed SWE

(mm)

SWE Change (mm)

21Z 17 January 2003 N

oah LSM 0

20

40

60

80

Sno

w W

ater

[m

m]

0

5

10

15

Sno

w W

ater

[m

m]

010

203040

5060

1/1/2003 1/31/2003 3/2/2003 4/1/2003

Sno

w W

ater

[m

m]

Midwestern United States

0

5

10

15

11/15/02 1/14/03 3/15/03

Sno

w w

ater

[mm

]

obscontrolassim

Midwestern United States

•  MODIS snow cover fields used to update the Mosaic and Noah LSMs within GLDAS/LIS

•  Models fill spatial and temporal gaps in data, provide continuity and quality control

•  Assimilated output agrees more closely with IMS snow cover fields (top middle) and ground observations (top right, bottom)

Based on Rodell and Houser, J. Hydromet., 2004.

Mosaic LSM

Noah LSM

•  Assimilated output contains more information (~hourly SWE) than MODIS (~daily snow cover)


Mongolia (n=32)Mongolia (n=32)

West Coast (n=59)West Coast (n=59)

Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07

High Plains (n=103)High Plains (n=103)

Southwest (n=28)!Open!Push!Pull• In situ

Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07

snow

wat

er e

quiv

alen

t, m

m

Advanced Rule-Based MODIS Snow Cover Assimilation

Foreward-looking “pull” algorithm •  Assesses MODIS snow cover observation 24-72 hours ahead •  Adjusts temperature to steer the simulation towards the observation •  Generates additional snowfall if necessary •  Improves accuracy while minimizing water imbalance

Zaitchik and Rodell, J. Hydromet., 2009



GRACE water storage, mm January-December 2003 loop

Model assimilated water storage, mm January-December 2003 loop

Monthly anomalies (deviations from the 2 0 0 3 m e a n ) o f t e r r e s t r i a l w a t e r s t o r a g e ( s u m o f groundwater, soil moisture, snow, and surface water) as an equivalent layer of water. Updated from Zaitchik, Rodell, and R e i c h l e , J . Hydromet., 2008.

GRACE Data Assimilation

From scales useful for water cycle and climate studies!

To scales needed for water resources and agricultural

applications

A B

A B


0

2040

60

80

100120

140

160

Jan-03 Jul-03 Jan-04 Jul-04 Jan-05 Jul-05 Jan-06

dept

h (m

m)

0

20

40

60

80

100

120

140

160


dept

h (m

m)

A

B

dept

h (m

m)

dept

h (m

m)

0

2040

60

80

100120

140

160


dept

h (m

m)

0

20

40

60

80

100

120

140

160


dept

h (m

m)

A

B

dept

h (m

m)

dept

h (m

m)

Model separates SWE, soil moisture, and groundwater; GRACE ensures accuracy.

Mississippi River basin

GRACE Data Assimilation

Catchment LSM TWS

GRACE-Assimilation TWS

From a global, integrated observation To application-specific water storage components

0

20

40

60

80

100

120

140


dept

h (m

m) Snow Water Equivalent

Soil MoistureGroundwaterObserved GroundwaterGRACE Total Water

Zaitchik, Rodell, and Reichle, J. Hydrometeorology, 2008.

r = 0.59 RMS = 23.5

r = 0.69 RMS = 18.7

NCAR Community Land Model (CLM4)

Co-Chairs: David Lawrence (NCAR), Zong-Liang Yang (Univ of Texas at Austin)

10

Layer 1 Ground

Layer 2, 3, … (up to 5 layer)

1 1 , 1 , 1 , , , + + + + snl snl ice snl liq snl z T ! !

Top layer

0 0 , 0 , 0 , , , z T ice liq ! !

LW ! , SW ! LW " , SW "

Conductive heat flux across layer

Sensible heat flux and latent heat flux

Multi-layer Snow Model in Community Land Model

St

LzT

ztTC ice

fice +!!

+!!

!!

=!! "

#$ )(

gg SWLWSHLHG +=++

Snowpack heat diffusion

Surface energy balance

ttttSWEtSWE EQPXX !!+= !1,, System function

11

Retrieving Snow Mass Using GRACE & CLM

Niu, Seo, Yang, et al., 2007, GRL


Data Distribution from GES DISC

http://disc.gsfc.nasa.gov/hydrology

•  Data available in GRIB and NetCDF formats •  Supports on-the-fly subsetting (right) •  Full documentation •  Quick look maps soon to be available •  Supports a growing number of national and international hydrometeorological investigations and water resources applications


Title: Multi-Sensor Snow Data Assimilation Problem Statement: MODIS, AMSR-E, and GRACE all provide observations that are relevant to snow water equivalent mapping, each with significant advantages and disadvantages. Hypothesis: Global fields of SWE can be produced with greater accuracy than previously seen by simultaneously assimilating MODIS snow cover, AMSR-E SWE, and GRACE terrestrial water storage observations within a sophisticated land surface model. Team: •  NASA/GSFC: Matt Rodell (PI), Ed Kim, Rolf Reichle •  U. Texas: Liang Yang (Co-PI) •  Johns Hopkins: Ben Zaitchik Timeline: Just getting started

Project Summary

multi-sensor snow data assimilation · groundwater, soil moisture, snow, and surface water) as an...

Documents