RPC Review (03/02/09) Evaluation of GPM Precipitation Estimates for Land Data Assimilation Applications Mississippi State University Geosystems Research.

1RPC Review (03/02/09)

Evaluation of GPM Precipitation Estimates for Land Data Assimilation Applications

Mississippi State University Geosystems Research Institute

2RPC Review (03/02/09)

GPM Evaluation Team & Collaborators

• MSU Team– Robert Moorhead– Valentine Anantharaj– Georgy Mostovoy– Yangrong Ling– QiQi Lu– Graduate students (2)

• External Collaborators– Paul Houser (GMU CREW)– Joe Turk (Naval Research Laboratories & JPL)

• Partner Agencies– Garry Schaeffer (USDA NRCS)– Steve Hunter (United States Bureau of Reclamation)

3RPC Review (03/02/09)

GPM Evaluations: Purpose and Activities

4RPC Review (03/02/09)

Purpose of RPC GPM Experiments

• Evaluate usefulness of GPM data for decision support in water resources management and other cross-cutting applications.

Test, characterize, and evaluate GPM data in conjunction with other precipitation products in the context of land surface modeling for earth science applications.

5RPC Review (03/02/09)

Experimental Objectives of GPM Evaluation

1. Verify space-based precipitation estimation using ground-based radar and rain gauge data for different cases representing different synoptic condition and surface types

2. Validate precipitation forcing impacts using land surface model simulations.

3. Evaluate potential impacts on water resources applications.

6RPC Review (03/02/09)

Satellite Omission Experiments

NASA RPC Review (4/14/08) 7

Motivation

• Precipitation is the main forcing for hydrological and land surface models

• Precipitation events that result in flooding often evolve over short space and time scales, where properly instrumented surface networks may not be available

• Satellite data often provides the only source of timely precipitation data over many of the world’s remote watersheds

• Hydrological modeling is a key focus of the future NASA/JAXA Global Precipitation Measurement Mission (GPM) in 2013; however

• The members of the GPM constellation will change owing to launch schedules before and during the mission; therefore

• How can we leverage today’s existing environmental satellite constellation to examine the impact of (existing and future) satellites and orbits on hydrological applications?

8RPC Review (03/02/09)

Current(10-Satellite)LEO Satellite Constellation

Revisit Time

Color Codes:SSMIDMSP F-13/14/15AMSR-EAquaAMSU-BNOAA-15/16/17TMITRMMCoriolisWindsatSSMISF-16

Revisit Scale: White= 0 hours Black= 6+ hours (shaded boxes represent 15-minute coverage)

9RPC Review (03/02/09)

Observing Times for an Ideal Precipitation-Based Low-Earth Orbiting Satellite Observing Constellation

Orbits are equally spaced with a 1.5 to 3-hour revisit timeAscending Descending

0

12

18 6

Pattern progresses from day to day

10RPC Review (03/02/09)

What We Have Today: DMSP and NOAA Satellites

NOAA Satellites as of Late 2006Ascending Descending

0

6

12

18NOAA-15

NOAA-16

NOAA-18

NOAA-17

DMSP Satellites as of Late 2006Ascending Descending

0

6

12

18F-14

F-13

F-16

F-15

“Building Blocks” for HRPPs: NOAA/DMSP SatellitesLate Summer 2008

Midnight

Noon

6 PM 6 AM

TRMM 28-day repeat at equator

F13

F14

F15

F16

F17NOAA-15NOAA-16

NOAA-17

NOAA-18METOP-A

Aqua

Coriolis

FY-3A (launched 29 May 2008)

AscendingDescending

F14 only direct-broadcast since 24 Aug 2008F18 launch date: 8 November 2008

NASA RPC Review (4/14/08) 12

Case 0: All satellites included (baseline)

Case 1: Omit all crosstrack soundersCase 2: Omit morning crosstrack soundersCase 3: Omit afternoon crosstrack sounders

Case 4: Omit TRMM TMI and PR and AquaCase 5: Omit TRMM PR onlyCase 6: Omit TRMM TMI onlyCase 7: Omit TRMM TMI and PR

Case 8: Omit all morning satellitesCase 9: Omit all afternoon satellites

Examining Impact of Satellite Type for the GPM EraSatellite Omission Experiments

NASA RPC Review (4/14/08) 13

Examining Impact of Satellite Type for the GPM EraSatellite Omission Experiments: Baseline

24-hour accumulations ending 2007/06/29 12Z 12-hour accumulations ending 2007/06/29 12Z

6-hour accumulations ending 2007/06/29 12Z 3-hour accumulations ending 2007/06/29 12Z

ALL Satellites CaseNOAA 15/16/17/18 (crosstrack)METOP-A (crosstrack)DMSP F-13/14/16/17 (conical)Aqua (conical)Coriolis (conical, over water)TRMM TMI (conical), PR

Green box illustrates largest performance impact is the omission of the morning overpass crosstrack sounders (“No AM XT” and “No AM” configurations)

Not much difference amongst sat-omission runs for the NRL-Blend “adjustment-based” HRPP technique

Bias

Equitable ThreatScore

Probabilityof

Detection

FalseAlarmRate

NOGAPS NWP model

summer winter

Seasonal PerformanceEast of 100W Longitude (> 1 mm/day)

Seasonal PerformanceWest of 100W Longitude (> 1 mm/day)

“All Satellites” CaseNOAA 15/16/17/18 (crosstrack)METOP-A (crosstrack)DMSP F-13/14/16/17 (conical)Aqua (conical)Coriolis (conical, over water)TRMM TMI (conical), PR

Generally less skill in winter months over western US “high altitude” region

Larger bias and spread for both satellites and model, especially in summer months

Higher false alarm rate for satellite-estimated, especially in winter months

Bias

Equitable ThreatScore

Probabilityof

Detection

FalseAlarmRate

NOGAPS NWP model

summer winter

PrincipleUse land surface models (LSM) and other types of hydrological observations (other than raingauge) to examine the impact of these GPM proxy data upon streamflow, discharge, soil moisture and other runoff measurements

Using Land Surface Models for GPM Ground Validation

GPM Ground ValidationExperimental Setup

• Incorporate the NASA Land Information System (LIS) with the NOAH LSM to simulate land surface and hydrological states

• Examine performance impact of different GPM constellations (e.g, Gottschalk et. al, 2005)

The Arkansas River is the longest tributary in the Mississippi-Missouri system. From its source in Colorado, the river travels through Kansas and northeastern Oklahoma. There it is joined by the Canadian, Cimarron, Neosho-Grand, and Verdigris Rivers, crosses the state of Arkansas where it empties into the Mississippi River.

Arkansas-Red River Basins Arkansas

CanadianRed

Cimarron

The analysis domain below covers the south-central United States where there are several well-instrumented watersheds. The impact of precipitation in a LSM is dependent upon many physical factors, soil type, vegetation, etc. Soil moisture analysis at a given time is likely to be the cumulative result of precipitation that has fallen for weeks or months prior.

To accommodate this, the results are shown after 5 months of simulation time, valid at 18 UTC on 31 October 2007. Soil moisture simulations are performed with 0.1o latitude x 0.1o longitude resolution and the North American Land Data Assimilation System (NLDAS) forcing fields (except for precipitation) are used to run the Noah LSM.

Impact of GPM Precipitation Estimates Upon Land Surface Models

No Crosstrack Sounders No AM Crosstrack Sounders No PM Crosstrack Sounders

No TMI+PR+Aqua No AM Satellites No PM Satellites

Satellite Denial Experiments: Effect on LSM

Satellite Denial Experiments: Effect on LSMExample: Omit All Crosstrack Sounders

Soil moisture difference relative to the “all-satellites” configuration

Upper Layer(0-10 cm)

Deep Layer(0.4-1 m)

Valid at 18 UTC on 31 October 2007

Both the gauge-based verification analysis and the LSM satellite denial experiments indicate that the greatest impact to the “all satellites” GPM configuration appears when the crosstrack sounders and the morning crossing (LTAN near 1800) satellites are omitted.

The removal of the morning satellites likely has less to do with the specific local time-of-day observation than it does with the fact that the bulk of the current (2008) satellites such as DMSP, Coriolis and several NOAA have early morning crossing times.

While this example demonstrates only one time step, these LSM simulations are being extended to cover multi-year DJF and JJA seasonal analyses.

Conclusions and Future Efforts

21

Summary of Progress

• Completed Tasks– Synthesis of GPM-proxy data (based on NRL-blend algorithm)– Verification of precipitation products

• Manuscript in draft, to be submitted to an AMS journal– Satellite Omission Experiments

• Book chapter accepted (Springer Verlag)• Manuscript in final draft, to be submitted to IEEE-JSTARS

• Final Steps (in progress)– Evaluation against USBR application metrics (soil moisture and

evaporation)– Final evaluation (document and publish)

NASA RPC Review (3/2/09)

Note: “verification” refers specifically to the satellite precipitation verification approach as used by the International Precipitation Working Group which is different from the systems engineering verification process defined by ASP.

22RPC Review (03/02/09)

Contact Information

Valentine Anantharaj<[email protected]>

Tel: (662)325-5135