precipitation estimation using combined radar/radiometer measurements within … · 2013-01-16 ·...

Precipitation Estimation Using Combined Radar/Radiometer Measurements Within the GPM Framework

Arthur Hou

NASA Goddard Space Flight Center

7th European Conference on Radar in Meteorology and Hydrology

25-29 June 2012, Toulouse, France

GPM Constellation of Satellites

GPM Core Observatory

(NASA/JAXA) 2014

Suomi NPP(NASA/NOAA)

MetOp B/C(EUMETSAT)

2012

JPSS-1 (NOAA)

Megha-Tropiques(CNES/ISRO)

DMSP F19/F20(DOD) GCOM-W1

(JAXA)

NOAA 19(NOAA)

Next-Generation Unified Global Precipitation Products Using GPM Core Observatory as Reference

Current Generation of Satellite Precipitation Products

Current global rainfall products are based on MW or MW+IR observations from uncoordinated satellite missions using a variety of merging techniques

50N

TRMM Realtime 3hr global rain map at 0.25o resolution

Arthur Hou, GPM Overview, ERAD 2012, Toulouse, France 3

� TRMM* radar provided an anchor for rainfall estimates by passive microwave sensors in the tropics and subtropics.

� Further advances require better sensors and remote-sensing algorithms (especially for light rain and falling snow).

50S

*Tropical Rainfall Measuring Mission launched by NASA and NASDA in 1997

� Use coordinated precipitation measurements by a constellation of microwave radiometers to achieve global coverage and sampling through partnerships.

� Use combined observations from active and passive sensors on a reference satellite - i.e. the GPM Core - to improve precipitation estimates from microwave radiometers.

- Translation of rain-affected microwave radiances into precipitation parameters requires knowledge of water contents throughout the

GPM Mission Concept


parameters requires knowledge of water contents throughout the atmospheric column.

- Current generation of rainfall retrievals relies on limited model-simulated hydrometeor database to relate satellite MW radiances to rain rates within a statistical framework.

- GPM will use a global observation-constrained hydrometeor database constructed from GPM Core sensor measurements to improve the accuracy and consistency of precipitation measurements from all constellation radiometers.

GPM Core Observatory


GPM Microwave Imager (GMI): 10-183 GHz

(NASA)

Dual-frequency Precipitation Radar (DPR): Ku-Ka bands

(JAXA/NICT)Non-Sun-Synchronous orbit at 65o inclination & 407 km

5

� Set a new standard for precipitation measurements from space:- Using combined measurements from an advanced radar/radiometer system

specifically designed for this purpose.

- Better measurement accuracy and higher sensitivity to light rain and snowfall relative to TRMM.

� Provide observations at different times of the day to improve GPM constellation sampling to characterize the diurnal variation of precipitation and provide better estimates of rainfall accumulation.

- By filling the gaps between observations at fixed local times by GPM

Role of the GPM Core Observatory


- By filling the gaps between observations at fixed local times by GPM constellation satellites flying in polar orbits.

� Provide a radiometric reference to reconcile differences in center frequency, viewing geometry, resolution among constellation radiometers

- By converting observations of one satellite to virtual observations of another using the GMI as a transfer standard.

� Unify and improve precipitation estimates from all constellation radiometers:

- By providing an a priori common observational hydrometeor database consistent with DPR and GMI measurements.

Core Observatory Integration & Testing at NASA/GSFC


Item KuPR at 407 km KaPR at 407 km TRMM PR at 350 km

Antenna Type Active Phased Array (128) Active Phased Array (128)Active Phased Array

(128)

Frequency 13.597 & 13.603 GHz 35.547 & 35.553 GHz 13.796 & 13.802 GHz

Swath Width 245 km 120 km 215 km

Horizontal Reso 5 km (at nadir) 5 km (at nadir) 4.3 km (at nadir)

Tx Pulse Width 1.6 µµµµs (x2) 1.6/3.2 µµµµs (x2) 1.6 µµµµs (x2)

Range Reso 250 m (1.67 µµµµs) 250 m/500 m (1.67/3.34 µµµµs) 250m

Observation Range 18 km to -5 km (mirror image around nadir)

18 km to -3 km (mirror image around nadir)

15km to -5km (mirror image at nadir)

DPR Instrument Characteristics


* Minimum detectable rainfall rate is defined by Ze=200 R1.6 (TRMM/PR: Ze=372.4 R1.54 )

PRF VPRF (4206 Hz±±±±170 Hz) VPRF (4275 Hz±±±±100 Hz) Fixed PRF (2776Hz)

Sampling Num 104～～～～112 108～～～～112 64

Tx Peak Power > 1013 W > 146 W > 500 WMin Detect Ze(Rainfall Rate)

< 18 dBZ( < 0.5 mm/hr )

< 12 dBZ (500m res)( < 0.2 mm/hr )

< 18 dBZ( < 0.7 mm/hr )

Measure Accuracy within ±±±±1 dB within ±±±±1 dB within ±±±±1 dB

Data Rate < 112 Kbps < 78 Kbps < 93.5 Kbps

Mass < 365 kg < 300 kg < 465 kgPower Consumption < 383 W < 297 W < 250 W

Size 2.4××××2.4××××0.6 m 1.44 ××××1.07××××0.7 m 2.2××××2.2××××0.6 m

Estimation of PSD & Precipitation Rate from the DPR

• A characteristic size parameter (D0) of thePSD can be estimated from the difference(in dB) between Ku- and Ka-band radarreflectivity factors

• Ambiguities include unknown shape parameter (µ) of the gamma PSDdistribution and the snow mass density (ρ)

• Characteristic number concentration of PSD

RAIN


• Characteristic number concentration of PSDis found from D0 and the radar equation

• Step-by-step estimation of attenuationcorrection based on PSD estimates

• Precipitation rate and the equivalent watercontent are derived from the PSD for anassumed velocity distribution

SNOW

Meneghini et al.

GMI Instrument Characteristics

Frequency NEDT Req. (K)Expected*NEDT (K)

Expected Beam Efficiency (%)

Expected Cal. Uncertainty (K)

Resolution (km)

10.65 GHz(V & H) 0.96 0.96 92.1 1.04 19.4 x 32.2

18.7(V & H) 0.84 0.82 93.3 1.08 11.2 x 18.3

23.8(V) 1.05 0.82 94.3 1.26 9.2 x 15.0

36.5(V & H) 0.65 0.56 98.5 1.20 8.6 x 14.4

89.0 0.57 0.40 95.6 1.19 4.4 x 7.3


89.0(V & H) 0.57 0.40 95.6 1.19 4.4 x 7.3165.5

(V & H) 1.5 0.81 91.9 1.20 4.4 x 7.3

183.31±±±±3(V) 1.5 0.87 91.7 1.20 4.4 x 7.3

183.31±±±±7(V) 1.5 0.81 91.7 1.20 4.4 x 7.3

Data Rate: ~30 kbpsPower: 162 WattsMass: 166 kg

* Analysis data as of Sept. 2010

Deployed Size: 1.4 m x 1.5 m x 3.5 m Antenna Size: 1.2 mSwath: 885 kmResolution and swath for GMI on Core

Combined DPR+GMI retrievals

• Using GMI radiance measurements as additional constraintson the DPR profiling algorithm:

Assumptions regarding the particle size distribution, ice microphysics, cloud water and water vapor vertical distribution are refined – using a variational procedure that minimizes departures between simulated and observed brightness temperatures (according to the sensitivity of simulated brightness temperatures to assumptions in DPR retrievals).


• Retrievals are consistent with both DPR reflectivities and GMIradiances within a maximum-likelihood estimation framework.

• Results enable the construction of an a priori database thatrelates hydrometeors to brightness temperatures for therange of Tb values observed over the globe, which can then beused for precipitation retrievals from all constellation radiometers.

• Unified radiometer rainfall retrieval using a common a priori hydrometeor database consistent with combined DPR+GMI measurements.

• Proof-of-concept demonstration using TRMM PR and TMI:

Prototype GPM Radar-Enhanced Radiometer Retrieval

• Outer Swath: TMIrainfall retrieval using an a-priori cloud database derived from PR reflectivity and


from PR reflectivity and TMI radiances within the inner swath.

• Inner Swath: PR rainfall retrieval (at different spatial resolution)

Kummerow et al.

GPM Ground Validation

- Refine algorithm assumptions & parameters (especially for solid precipitation mand surface emissivity information over land) - Characterize uncertainties in satellite retrievals & GV measurements

For pre-launch algorithm development & post-launch product evaluation

(satellite simulator)

(in situ microphysics)

Physical Validation: Dream ScenarioThree complementary approaches:

• Direct statistical validation (surface):

- Leveraging off operational networks to identify and resolve first-order discrepancies between satellite and ground-based precipitation estimates


microphysics)

“Truth” estimated through the convergence of satellite & ground-based estimates

estimates

• Physical process validation (vertical column):

- Cloud system and microphysical studies geared toward testing and refinement of physically-based retrieval algorithms

• Integrated hydrologic validation/applications (4-dimensional):

- Identify space-time scales at which satellite precipitation data are useful to water budget studies and hydrological applications; characterization of model and observation uncertainties

Constrain DSD parameters with disdrometer measurements

•The gamma distribution, used to represent the rain DSD, has three free parameters:

– Concentration (N0 or Nw)

– Characteristic size (D0 or Dm)

– Spread (µ or σm)

•These parameters, as measured by disdrometer, are not statistically independent (orthogonal)


disdrometer, are not statistically independent (orthogonal)

•Dual-frequency radar can only solve for two parameters at each range gate

•Relationships between parameters can be used to constrain the dual-frequency radar solution (better than assuming µ=constant)

Williams et al.

Use GV data to refine scattering tables for improved satellite precipitation retrieval

Use dual-pol radars to understand spatial structure and variability of DSD parameters

•How can we extend the DSD information retrieved by Ku+Ka-band observations in the inner swath to the Ku-only outer swath?

•What is the relationship between particle size above, within, and below the melting layer?

•What is the behavior of the DSD in the


•What is the behavior of the DSD in the ground clutter-masked region (as high as 2km)?

•How strongly should DSD parameters be constrained in adjacent pixels?

•Dual-polarized radars can retrieve particle size (D0 or Dm) over large areas and with fine vertical resolution to answer these questionsBringi et al.

Var(Rr – R) = Var(Rr - Rg) – Var(Rg-R)

ObjectiveCharacterize uncertainties in satellite products using radar

and/or gauge data

25 km range rings

Precipitation GV Science Research Facility at NASA/WFF

Location: NASA/GSFC Wallops Flight Facility


• Approach: Dense long-term gauge/disdrometer network under radar coverage• Stage 1: - Dense gauge network and multi-parameter/frequency radars

- 25 gauge pairs, 5 x 5 km2 area. Total inventory 70+ TB rain gauges- 4 existing locations with gauge pairs along the Eastern Shore (range studies)- NPOL (S-band) , SPANDAR (S-band) , WSR-88D (S-band), TOGA (C-band), D3R (Ka-Ku band) to quantify radar reference accuracy as f(scale, measurement type);

• Stage 2: 20+ disdrometers, 5 2DVDs, 20+ Parsivel, ~4 Joss; DSD variability studies +6 MRR’s

• Addressing precipitation regime diversity via partnerships and collaboration: - Coastal land/oceanic and seasonal regime gradients;- long term observations between IOPs- Leverage partnering activities to expand regimes; e.g., Iowa Flood Center, HyMeX, S. Korea

25 km range ringsWallops Flight Facility

(See poster by Walt Petersen)

International Science Collaboration

� Pre-CHUVA with Brazil on warm rain retrieval over land in Alcântara, 3-24 March 2010.

� Light Precipitation Validation Experiment (LPVEx): CloudSat-GPM light rain in shallow melting layer situations in Helsinki, Finland, 15 Sept - 20 Oct 2010.

� Mid-Latitude Continental Convective Clouds Experiment (MC3E): NASA-DOE field campaign in central Oklahoma, 22 Apr 22 – 6 June 2011.

� GPM Cold-season Precipitation Experiment (GCPEX): Joint campaign with Environment Canada on snowfall retrieval in Ontario, Canada, 17 Jan – 29 Feb 2012.

• GPM participation in 2012 HyMeX SOP later this year.

• 3 more campaigns in the U.S. being planned for 2013-2016.

Active Joint Science Projects With:


Active Joint Science Projects With:

• Argentina (U. Buenos Aires) • Australia (BOM)• Brazil (INPE)• Canada (EC)• Ethiopia (AAU)• European Organization (ECMWF)• Finland (FMI)• France (CNRS & Obs. de Paris)• India (ISRO)• Israel (Hebrew U. Jerusalem)• Italy (CNR-ISAC & Sapienza U. Rome)• South Korea (KMA)• Spain (UCLM)• United Kingdom (U. Leicester)

Pre-CHUVA (2010)

MC3E (2011)

NASA-EC GCPEX (2012) LPVEx (2010)

• Dynamic downscaling using cloud-resolving WRF ensemble data assimilation for hydrological applications

WRF-GSI(no AMSR-E,TMI)

WRF-EDAS(with AMSR-E, TMI)

NOAA Stage IV(Verification) mm

Rain accumulation for 15-22 Sept. 2009 over Southeast US flood region

- Assimilation of rain-affected MW radiances (TMI, AMSR-E, MHS) into the NASA Unified WRF Ensemble Data Assimilation System improves precipitation analysis and short-term forecast.

• Variable-resolution global precipitation product via data fusion of satellite and

NASA/GSFC & CSU

Downscaled High-Resolution Precipitation Products


• Variable-resolution global precipitation product via data fusion of satellite and ground measurements within the framework of probabilistic estimation and sparse representation

- Optimal non-Gaussian estimation and filtering using wavelet decomposition that preserves precipitation extremes at native measurement scales.

- Statistical characterizations of observation errors at native scales as a function of surface type, geographic location, season, and rain intensity.

- Data fusion at multi-scales according to uncertainties of individual data types

NASA/GSFC & U. Minnesota

TRMM PR (4km)

Data Fusion Product (1km)

NEXRAD (1km)

dB

� Advanced active/passive sensor capabilities- Higher sensitivity to light rain and solid precipitation than TRMM instruments

- Insights into precipitation physics with quantitative estimates of PSD parameters

� Next-generation unified global precipitation data products- Inter-calibrated radiometric data from a constellation of MW sensors

- Unified precipitation retrieval using a common hydrometeor database consistent with combined active/passive sensor measurements

� Near real-time data for operational use and societal applications

Summary

GPM is an international satellite mission that will unify and advance precipitation measurements from a constellation of microwave sensors for research and application


� Near real-time data for operational use and societal applications

� GV is key to refining algorithm assumptions and characterizations of uncertainties in precipitation estimates to improve GPM data products and utilization:

- NASA is conducting a series of focused GV field campaigns in collaboration with domestic and international partners to improve GPM satellite algorithms

- Establishing GV research facilities to characterize uncertainties in satellite and ground-based precipitation estimates

� GPM offers a framework for international collaboration on precipitation science:- NASA Precipitation Science Team currently has 21 International Pr. Investigators from 13 nations.

- Environment Canada hosting the 5th International GPM GV Workshop in Toronto, 10-12 July 2012.

- CNR/ISAC of Italy will host the 6th International GPM GV Workshop in Rome, Nov 2013.


http://gpm.nasa.gov

Additional Slides


Baseline Constellation Schedule GPM (2015)

Baseline Constellation Coverage & Sampling Capability

(Hour)


Reduce the global mean revisit time from 2.2 to 1.6 hrsPrime Life Extended Life

Cumulative Distribution of Observations (2015)

(%)

(Hour)

• More than 50% of observations are << 1 hr apart at all latitudes

• Percent observations that are < 3 hrs aapart:

- 80% in the tropics

- 70% in the midlatitudes (30o-45o)

- 90+% in polar regions

GPM Core Launch

GPM Near Real-time Data Products

• GMI L1 and L2 swath products within 20 min. of data collection

• Selected DPR L2 (e.g., reflectivity and precipitation rate) swath products within 120 min. of data collection

• Combined GMI and DPR L2 swath products within 120 min. of data collection

• L1C intercalibrated brightness temperature swath products and L2 GPROF precipitation products for partner radiometers within 10 min. of


GPROF precipitation products for partner radiometers within 10 min. of receiving L1B data from data providers

• L3 merged MW+IR, 0.1o x 0.1o gridded, half-hourly global precipitation products:

- Low-latency, quick-look products (with relatively high IR data content) near data collection time

- Late-look products with all available MW data within the collection window

precipitation estimation using combined radar/radiometer measurements within … · 2013-01-16 ·...

Documents