1. LIMA Water Cycle Capacity Building workshop
November 30 – December 4, 2009
The South America Land Data Assimilation System
(SALDAS)
Luis Gustavo G. De Goncalves
ESSIC University of Maryland
NASA Goddard Space Flight Center

2. Data Assimilation: Sleepy Driver Analogy
(Land surface model)
You are going in the right general direction…
…but when you open your eyes, you correct towards the center of the road
A land data assimilation system (LDAS) is a numerical modeling scheme which integrates observations from various sources within such models, using data assimilation and other techniques, in order to produce optimal maps of land surface states (e.g., soil moisture, surface temperature) and fluxes (e.g., evapotranspiration, runoff) for weather and climate forecast model initialization, water resources monitoring, and other applications. Data assimilation is a numerical method by which two independent estimates of a variable can be combined to determine one best estimate, which is less uncertain than either of the two inputs. In this case, the variable is a land surface state such as soil moisture. One estimate comes from the land surface model and the other from a satellite observation. The error characteristics of each are used to weight their contributions to the final estimate. The North American and Global Land Data Assimilation System projects (NLDAS and GLDAS) demonstrated that these techniques can be implemented at regional to global scales (Mitchell et al., 2004; Rodell et al., 2004; see and
(Observations)
Courtesy Matt Rodell

3. Ensemble Kalman Filter
Slide credit: Paul Houser
Update: xi(+) = xi(-) + K [ Z - H xi(-)]
K = P Ht [ H P Ht + R]-1
X = soil moisture state (i-th ensemble member)
(-) before update (+) after update
Z = remote sensing data
H = measurement operator
K = weight matrix
P = state error covariance (model uncertainty)
R = measurement error cov
(meas. uncertainty)
Weight matrix K depends on model uncertainty.
Extended KF:
Linearized dynamic matrix equation for P.
Horizontal error correlations too expensive.
EnKF:
Use sample covariance from ensemble.
Nonlinear propagation of uncertainty.
Can account for horizontal error correlations.
Courtesy Matt Rodell

4. South American Land Data Assimilation System (SALDAS)
Goal: combine local observations and parameters with NASA advanced hydrological modeling expertise and capabilities to improve Global and SA NWP, climate and water management through collaboration with various centers (government, universities and research institutes)
CPTEC/INPE
NASA/GSFC
Local resources (observations)
Human Resources
Central to South America
Missions
Land Information System
SALDAS
REGIONAL
Weather and Climate
Water Management
Global
Hydromet Databanks
Capacity
Building

5. South American Land Data Assimilation System (SALDAS)
Background
SALDAS project seeks to provide accurate, near-real-time and retrospective land surface states over South America
Quality of land surface model (LSM) output is closely tied to the quality of the meteorological forcing data used to drive the model
Model and observation-based data used to create high-quality forcing data used by Noah, SSiB, SiB2, CLM2, MOSAIC, and VIC (to be tested) LSMs
Retrospective ( , CPTEC)
Real-time (2002-Present, CPTEC)

6. GRACE Data Assimilation
The Gravity Recovery and Climate Experiment (GRACE) satellite mission produces monthly maps of Earth’s gravity field with enough precision to infer changes in total terrestrial water storage (groundwater + soil moisture + surface water + snow) over large areas (> 160,000 km2)
Catchment land surface model
Ensemble Kalman smoother data assimilation
Catchment LSM (Koster et al., 2000): z
three snow layers
surface excess
root zone excess
“catchment deficit”
Courtesy Matt Rodell

7. GRACE Data Assimilation
Results have higher resolution than GRACE alone, better accuracy than model alone.
GRACE Assimilating Catchment LSM TWS anomaly, mm
January 2003 – June 2006
GRACE TWS anomaly
January 2003 – June 2006
From scales useful for water cycle and climate studies…
To scales needed for water resources and agricultural applications
Zaitchik, Rodell, and Reichle, J. Hydromet., 2008.

8. GRACE Data Assimilation
LDAS models produce continuous time series; near-real time capable.
Monthly GRACE data anchor model results in reality
Missouri
Upper Mississippi
Mississippi River sub-basins
column water (mm)
column water (mm)
GRACE Water Storage
Modeled Water Storage
Model-GRACE Assimilation
Daily estimates are critical for operational applications
column water (mm)
column water (mm)
Lower Miss-Red-Arkansas
Ohio-Tennessee

9. GRACE Data Assimilation
LDAS models produce continuous time series; near-real time capable.
LPB Implementation plan 2005

10. NASA Land Information System (LIS)
South American Land Data Assimilation System (SALDAS)
NASA Land Information System (LIS)
Schematic representation of the land surface modeling into the NASA Land Information System (LIS), a multi-LSM distributed framework where SALDAS in built upon. (S.V. Kumar, et al, 2006)

11. Forcing Specification
South American Land Data Assimilation System (SALDAS)
Forcing Specification
Spatial Domain
3-Hourly files
1/8th and 1/10th Degree over Equator
Quality controlled, adjusted for terrain height
Modeled and observation-based fields

12. Atmospheric Forcing Fields
South American Land Data Assimilation System (SALDAS)
Atmospheric Forcing Fields
Model-Based Estimates of Standard Climate Station Data
Temperature ( 2 m assuming grass)
Specific Humidity ( 2 m assuming grass)
U East-West Wind Component (10 m assuming grass)
V North-South Wind Component (10 m assuming grass)
Surface Pressure ( 0 m assuming grass)
Observation-Based Data
Downward Shortwave Radiation
Precipitation

13. Terrain Height Adjustment
South American Land Data Assimilation System (SALDAS)
Terrain Height Adjustment
ETA temperature, pressure, humidity and longwave radiation adjusted for differences in ETA versus LDAS terrain height
Temperature and pressure corrected using standard lapse rate
Specific humidity and longwave radiation corrected by holding relative humidity constant
Corrections of up to 3.7K, 60hPa, 2.24W/m2, 0.06 Kg/kg

14. South American Land Data Assimilation System (SALDAS)
Forcing File Creation
CPTEC ETA South American Regional Reanalysis SARR
CPTEC ETA Operational Data Assimilation System ODAS
Observations not always available, so CPTEC/SARR and CPTEC/ODAS data used as base
Current stage:
SARR, 6 hourly, 40km, (planned )
Next step:
ODAS, 3 and 6 hourly, 20km, 2002-present (planned 20Km)
Spatially interpolated to 1/10th degree
Temporally interpolated to 3-hourly data
To be quality controlled using ALMA ranges

15. Precipitation

16. Automated Surface Stations
Courtesy: INMET/INPE

17. Observation-based forcing
South American Land Data Assimilation System (SALDAS)
Observation-based forcing
Model-based data subject to model error, so observations used when possible
Radiation
GOES-CPTEC downward shortwave
GOES-CPTEC PAR (not implemented yet)
GOES-CPTEC skin temperature (not implemented yet)
Precipitation
CPC daily gauge data
Combined TRMM-raingauge
Sub daily automated surface stations network
Rain gauges from various SA agencies (~5x > GTS)

18. South American Land Data Assimilation System (SALDAS)
GOES data processed at CPTEC/DSA (Divisao de Satelites Ambientais: Environmental Satellites Division) to create 1/25 degree, hourly, instantaneous surface downward shortwave radiation, PAR and skin temperature fields
Interpolated to 1/10th degree
GL 1.2 GOES downward shortwave radiation (W/m2)
SARR downward shortwave radiation (W/m2)
Merged SALDAS downward shortwave radiation (W/m2)

19. South American Land Data Assimilation System (SALDAS)
Radiation Forcing Processing Statistics

20. Observed Radiation Mean difference SARR - GL1.2 (W/m2)
15Z - January 2004
Mean GL1.2 (W/m2)
15Z - January 2004

21. Observed Radiation Mean difference Mean GL1.2 (W/m2)
SARR - GL1.2 (W/m2)
15Z - June 2004
Mean GL1.2 (W/m2)
15Z - June 2004

22. Temporal Disaggregation Process
Precipitation
Make use of TRMM and raingauges analysis data to form best available product—a temporally disaggregated 3-hourly value
Temporal Disaggregation Process
Description: Use of rain gauge to correct satellite bias precipitation
Spatial Resolution: 0.25 degree
Temporal Resolution: total daily precipitation
Domain: South America
Methodology: Use TRMM sub-daily precipitation pulses to dissagregate the total daily amounts.
Note:
3B42RT data used to derive temporal disaggregation weights
Sum of hourly data values equals original total daily TRMM/gauge
Additive/Ratio Bias correction (Vila et al, 2009) JHM

23. South American Land Data Assimilation System (SALDAS)
Implement data assimilation techniques (EKF, EnKF) for low data density regions
Participants: CPTEC, NASA, COLA, GMU, University of Arizona, Utah University, IPH, Universidade de Vicosa, Universidade de Santa Maria
Streamflow data assimilation (use of ANA monitoring capabilities for determining average soil moisture over low data density basins)
Soil moisture data assimilation (combine AMSR-E Aqua satellite estimates with ground based observations over areas with low vegetation cover) PCD’s
Skin temperature data assimilation (remote sensed skin temperature, DSA derived, combined with surface observations) PCD’s
SENAMHI (Peru)?

24. South American Land Data Assimilation System (SALDAS)
An operational implementation of a land surface data assimilation into the CPTEC-INPE ETA/PSAS/LTEKF atmospheric data assimilation cycle
… Timestep 1
Timestep 2
Timestep … 3
Observations
Error Stats
SALDAS
SALDAS
SALDAS
4DDA
EKF/EnKF
LSM
Observations
Error Stats
CPTEC
RPSAS
Eta
Eta
LSM
LSM

25. Modeling Activities Summary
Applications
Uncoupled
Land Data Assimilation
LDAS
Current
ADAS
Coupled
Earth System
Model with
Atmospheric,
Land and Ocean Data Assimilation
Future
ODAS
LDAS

26. South American Land Data Assimilation System (SALDAS)
La Plata Basin: high spatial and temporal resolution runs help to improve the understanding of the hydrological and meteorological processes over the region. Under the LIS framework, SALDAS can be set to run at up to 1Km spatial resolution with hourly output.
La Plata basin integrated total runoff (Kg/m2) at 1Km resolution (January 2000)
La Plata basin volumetric soil moisture at 1Km resolution (January 2000)

27. South American Land Data Assimilation System (SALDAS)
High resolution satellite-based 1Km crop mapping over the LPB
Courtesy: Mutlu Ozdogan

28. South American Land Data Assimilation System (SALDAS)
Multi-model ensemble
Four LSM’s (NOAH, CLM, MOSAIC and VIC) were run from using GDAS and NASA’s MERRA as atmospheric forcing.
We are interested on seeing how different models under the same conditions simulate the same period of time.
The focus is on the models spread and mean for simulated water balance variables: surface and subsurface runoff and evaporation and how they compare with input precipitation.
We concentrate on three major basins in South America: Amazonas, Tocantins and La Plata, each with its unique characteristics.

29. South American Land Data Assimilation System (SALDAS)
Multi-model ensemble
Three major South American hydrological basins analyzed: Amazonas (blue), Tocantins (green) and La Plata (purple). High resolution simulations (1Km) are being performed at the Parana sub-basin for the La Plata Basin as shown in the figure details.

30. South American Land Data Assimilation System (SALDAS)
Multi-model ensemble

31. South American Land Data Assimilation System (SALDAS)
Summary
Model and observation based data merged to create robust, accurate 1/10th degree 3-hourly forcing data set
CPTEC-SARR/ODAS/ETA data serves as base
CPTEC-DSA/GOES, TRMM/raingauge data used to augment data set
Common set of forcing integral to SALDAS LSM intercomparisons
Five years archived, with continuing production
Validation effort proceeding

32. South American Land Data Assimilation System (SALDAS)
Summary
High resolution simulations up to 1Km over selected regions (e.g. La Plata basin) with soil moisture and soil temperature DA
Improvement of ancillary data over regions outside Brazil based on in situ and remote sensing observations (SENAMHI?)

33. LIMA Water Cycle Capacity Building workshop
November 30 – December 4, 2009
ftp://ftp1.cptec.inpe.br/jgerd/SALDAS/Assimila.html
Gracias
Luis Gustavo G. De Goncalves
ESSIC University of Maryland
NASA Goddard Space Flight Center

34. Latin America Water Resources and Capacity Building
Integrating NASA Earth Sciences Research results into Decision Support Systems for Agriculture and Water Management in South America
Ernesto H. Berbery3, Eric F. Wood4, Maria Assuncao F. Silva Dias5, Osvaldo Luiz L. De Moraes6 and Jurandir Zullo7
&
Latin America Water Resources and Capacity Building
David L. Toll2, Ted Engman2
Luis Gustavo G. de Goncalves1,2
1Earth System Science Interdisciplinary Center, University of Maryland
2Hydrological Sciences Branch, NASA Goddard Space Flight Center
3Department of Atmospheric and Oceanic Science, University of Maryland
4Princeton University
5Centro de Previsao do Tempo e Estudos Climaticos/Instituto Nacional de Pesquisas Espaciais
6Universidade Federal de Santa Maria
7University of Campinas

35. Integrating NASA Earth Sciences Research results into Decision Support Systems for Agriculture and Water Management in South America
OUTLINE
NASA Applied Sciences starting FY10
Use of NASA remote sensing and modeling products combined with surface observations at various scales to improve decisions support systems in agriculture, drought and water resources management for South America (SA)
Build upon a partnership between NASA and various U.S. and international agencies and universities to contribute to the dissemination of NASA Earth Sciences research results within that continent
Provide valuable information based on NASA Earth Sciences products to South American national agencies and other end-users
Paul Simon Water Act for the Poor
legislation that requires special attention to be given to semi-arid regions such as Northeast Brazil, western Argentina and Atacama

36. Integrating NASA Earth Sciences Research results into Decision Support Systems for Agriculture and Water Management in South America
THE AZCR
The “Agricultural Zoning for Climatic Risks” (AZCR) is a program developed in Brazil since 1995, to provide guidance information for agricultural practices.
Development of methods to estimate climatic risks of regional individual crops, intercropping and cattle-farming integration systems
Cattle farming systems are adding up quality to the grain production and yield of pastures, mitigating the environmental impacts and reducing the pressure on the Amazonian forest

37. EOS multi-sensor (AIRS, CERES and MODIS) based evapotranspiration
Integrating NASA Earth Sciences Research results into Decision Support Systems for Agriculture and Water Management in South America
“Incorporating high quality observations of surface conditions from NASA satellites and other observing systems to quantify continental scale daily ET at 5~20Km spatial resolution and water storages and fluxes at basin scale will enable enhancement of decision support tools for agriculture, human use and disaster mitigation over South America.”
EOS multi-sensor (AIRS, CERES and MODIS) based evapotranspiration
The Gravity Recovery and Climate Experiment (GRACE) estimates of monthly variations in terrestrial water storage (TWS)
The Tropical Rainfall Mission Measurement (TRMM) derived precipitation
AMSR-E and TRMM Microwave Imager (TMI) soil moisture
NASA’s Land Information System (LIS) and the South American Land Data Assimilation System (SALDAS)
Land Surface Modeling Framework
CPTEC/INPE atmospheric models
INMET surface observations network

38. Integrating NASA Earth Sciences Research results into Decision Support Systems for Agriculture and Water Management in South America
PRINCETON UNIVERSITY Remote Sensed ET Estimation

39. The NASA Water Resources Program may assist:
NASA Water Resources Program Project Support for Activities in Latin America
“Initiate capacity building {and end to end projects} programs to develop tools for using remote sensing data in support of water management, and to show the value of Earth observations generally in water resource management. The program will be initiated in Latin America” {GEO Tasks WA & DI-07-01}
The NASA Water Resources Program may assist:
Assist with workshop support including the training of students and travel for US visits.
Support graduate students & post-docs
End to End Projects with Decision Support Systems

40. NASA Water Resources Program Project Support for Activities in Latin America
South American Land Data Assimilation System (SALDAS)
Collaborative work for the past 4 years (mostly unfunded) with the Brazilian Center for Weather Forecast and Climate Studies (CPTEC - an equivalent no NOAA/NCEP in SA) a division from the Brazilian National Institute for Space Research (INPE - an equivalent to NASA in SA). CPTEC/INPE is a lead and reference institution in Latin America for operational and research modeling. This collaborative work includes promoting interaction between Latin America students and researchers and US institutions as well as Capacity Building
Recently Funded
NASA LBA Ecology – LSM Intercomparison Project leverage on 8 flux sites in the Amazon region. SALDAS forcing used for wall-to-wall intercomparison.
NASA Applied Sciences Program proposal to improve decisions support systems in agriculture, drought and water resources management for South America
Funded Activities
NASA THP - La Plata Basin combine NASA products with local observations to improve understanding of the hydrological and meteorological processes over the region
IAI - La Plata Basin Land Use/Land Change due to natural and anthropogenic causes
Regions with significant impacts on latent heat flux when comparing initial conditions for operational NWP models models with well-balanced SALDAS fields
Amazon Observational Network of eddy flux tower sites (red dots) on a map of vegetation types in Amazônia and Brazil.
NDVI trends: surrogate for primary production from NOAA-AVHRR images. Red: decrease Blue: increase. [Courtesy of Jobbagy.]
La Plata basin integrated total runoff (Kg/m2) at 1Km resolution (January 2000)

41. Visiting Researchers and Students to NASA/GSFC/HSB
NASA Water Resources Program Project Support for Activities in Latin America
Visiting Researchers and Students to NASA/GSFC/HSB
2006 Enrique Rosero (Ecuador)
Amazon LSM parameter calibration
2008 Joao Mattos (Brazil)
Operational Land-Surface DA
2007 Rafael Rosolem (Brazil)
Amazon Hydrology & Calibration
2009 Claudia Ramos (Brazil)
AMSR-E Soil Moisture DA
2009 Mario Quadro (Brazil)
La Plata Precipitation Recycling
Debora Roberti (Brazil)
Wet Lab for regional soil moisture and ET
In support to agriculture

42. Capacity Building Meetings and Workshops
NASA Water Resources Program Project Support for Activities in Latin America
Capacity Building Meetings and Workshops
PI: Hugo Berbery (UMD)
Public: graduate students and young scientists
Two weeks hands-on training in hydrology, meteorology, ecological modeling and data assimilation
More than 100 registrations from North and South America, Africa, Europe and few from Asia…
To be held at the ITAIPU hydropower plant at the border of Brazil, Argentina and Paraguay
Inter-American Institute
Frederico San Martini, Water Policy Advisor USAID
Raymond Motha, Chief Meteorology, USDA