Assimilation of Satellite Soil Moisture Blended Products from SMOPS in NCEP Global Forecast System

Assimilation of Satellite Soil Moisture Blended Products from SMOPS in NCEP Global Forecast System

Weizhong Zheng1,2, Xiwu Zhan3, Jicheng Liu4, Michael Ek1, Jesse Meng1,2, Jiarui Dong1,2 and Helin Wei1,2

1NOAA/NCEP/EMC, College Park, MD 207402IMSG, Kensington, MD 20895

3NOAA/NESDIS/STAR, College Park, MD 207404GMU, Fairfax, VA 22030

JCSDA Science Workshop, June 5-7, 2013

1

OUTLINE:

ObjectiveObjective

Embed EnKF in GFSEmbed EnKF in GFS

Testing with Testing with SMOS SMSMOS SM

Testing with Testing with SMOPS Blended SMSMOPS Blended SM * Fits to SfcObs* Fits to SfcObs

* Fits to RaObs* Fits to RaObs

* Scores* Scores

* CONUS Precipitation* CONUS Precipitation

Summary and future plan Summary and future plan 2

““Online” satellite soil moisture data Online” satellite soil moisture data assimilation in NCEP GFSassimilation in NCEP GFS

Examine how satellite SM data impact Examine how satellite SM data impact NCEP GFS forecastsNCEP GFS forecasts

OBJECTIVESOBJECTIVES

3

NOAA NESDIS Soil Moisture Operational Product System (SMOPS)

6-hour operational soil moisture product from the Office of Satellite and Product Operations (OSPO) of NOAA/NESDIS. Carries near real-time global gridded soil moisture products from several individual satellite sensors as well as a blended product.

Soil moisture layer from individual sensors: (AMSR-E), ASCAT, SMOS, and WindSat.

Blended product: CDF matching is used to merge all available values from above individual sensors to a common climatology, resulting in better spatial and temporal coverage.

4

SMOPS Blended Product from WindSat, SMOS and ASCAT

WindSat SMOS ASCAT

Blended

5

6

7

SMOS GFS_CTL

GFS_EnKF EnKF-CTL

Testing with SMOS SM (May 1 – June 17 2012): 18Z, 1-17 June 2012

8

GFS_CTL

EnKF-CTL GFS_EnKF

SMOS

Testing with SMOS SM (May 1 – June 17 2012): 18Z, 1-17 June 2012

9

Assimilation of Satellite Soil Moisture Blended Products from SMOPS in NCEP Global Forecast System

The simplified ensemble Kalman Filter (EnKF) was embedded in the GFS latest version to assimilate soil moisture observation.

Case: Starting from 00Z April 1–May 4, 2012. (GFS/GSI)

Experiments: CTL: Control run (without SM DA) EnKF: Sensitivity run (Daily EnKF run) And precipitation perturbation.

PRT: Perturbation size for each layer soil moisture.10

Comparison of soil moisture from SMOPS Blended 18Z, 1-30 April 2012

GFS_EnKF EnKF-CTL

GFS_CTL SMOPS

11

Comparison of soil moisture from SMOPS Blended 18Z, 1-30 April 2012

SMOPS GFS_CTL

GFS_EnKF EnKF-CTL

12

GFS Top Layer SM Validation GFS Top Layer SM Validation With USDA-SCAN Measurements With USDA-SCAN Measurements

1-30 of April, 20121-30 of April, 2012

East CONUS (26 sites) West CONUS (25 sites) Whole CONUS

RMSE BiasCorr-Coef

RMSE BiasCorr-Coef

RMSE BiasCorr-Coef

CTLCTL 0.135 0.046 0.565 0.124 0.033 0.448 0.129 0.040 0.508

EnKFEnKF 0.130 -0.031 0.613 0.114 -0.021 0.549 0.123 -0.031 0.587

SMOPSSMOPS 0.133 -0.055 0.601 0.098 -0.036 0.402 0.117 -0.048 0.524

13

Soil moisture at 4 layers: Day 3: April 2 – May 4 2012

SM1

SM3

SM2

SM4

CTL

Diff

14

Q2m

(left) and T2m

(right): Day 3: April 2 – May 4 2012

CTL

Diff

15

Tmin2m

(left) and Tmax2m

(right): Day 3: April 2 – May 4 2012

Diff

CTL

16

Sfc RH (%) over CONUS West and East: April 2 – May 4 2012

West East17

Sfc RH (%) RMSE over CONUS West and East: April 2 – May 4 2012

West East18

Sfc T (K) over CONUS West and East: April 2 – May 4 2012

West East19

Sfc T (K) RMSE over CONUS West and East: April 2 – May 4 2012

West East20

RH (%) Bias and RMSE over CONUS: Day 5: April 2 – May 4 2012

Bias RMSE 21

T(K) Bias and RMSE over CONUS: Day 5: April 2 – May 4 2012

RMSE Bias22

+ 0.004

23

24

25

700 hPa

1000 hPa

+ 0.003

+ 0.003

26

- 0.36

- 0.43

Bias

RMSE

27

Bias RMSE

28

+ 0.002

+ 0.007

NHX

SHX

29

NHX SHX

30

Precipitation Skill Scores over CONUS

31

32

33

Summary:

The satellite soil moisture blended products from SMOPS were assimilated in NCEP GFS and the result shows as follows:

1) Improved GFS deeper layer soil moisture estimates comparing with in situ

measurements;

2) Improved GFS forecast scores and reduced its bias and root-mean-square

errors.

3) Showed some positive impact on precipitation on CONUS but not for

heavy precipitation.

Future: assimilate SMAP (remote sensing), COSMOS & other in situ

measurements. Furthermore, consider regions where soil alternates (frozen

and thawed) during the transition seasons (by CREST ongoing project).

34

Thank you!

Questions?

35