Advanced Satellite Imaging Applications in the NPOESS Era

1
Advanced Satellite Imaging Applications in the
NPOESS Era

• Joe Turk
• Naval Research Laboratory, Marine Meteorology
  Division
• Monterey, CA
• www.nrlmry.navy.mil/sat_products.html
• NRL Colleagues Steven D. Miller, Thomas F. Lee,
  Arunas P. Kuciauskas, Jeffrey D. Hawkins, Kim
  Richardson

2
Talk Outline

• Background on Sat-Focus
• NexSat
• VIIRS Day-Night Band (DNB) Overview
• Capability Examples
• Tropical Cyclones
• Hydrological Applications
• Summary/Conclusion

3
Audience and Scope
In the wake of 9/11, the ONR requested all Navy
RD agencies to accelerate tech-transfer
The Defined User Their Challenge Their
Requirements Our Task
Navy METOC analysts (e.g., Aerographers Mates)
operating in data-sparse/data-denied regions
Strike planning, making stoplight decision
charts based in part on environmental
characterization
Simple and comprehensive graphical products for
analysis and extraction of salient information
Exploit simple physical relationships (using high
spatial/spectral/radiometric resolution satellite
observations) to isolate key environmental
parameters from a potentially complex scene
4
Satellite Focus Web Page

• Co-registered product suite
• Customized image animations, mpegs, mosaic
  options
• Online image archive
• Satellite pass predictor
• Online training
• Hosted on Secure Internet (FNMOC)

SIPR and DoD-restricted NIPR Access Only!
Page 4 of 27
5
Introducing NexSat

• Building on the Satellite Focus heritage, and
  under the auspices of the NPOESS Integrated
  Program Office (IPO) NRL Monterey has developed
  NexSata public web site highlighting
  next-generation operational environmental
  satellite capabilities over the continental
  United States in near real time.
• NexSat demonstrates several new capabilities
  anticipated from the Visible/Infrared
  Imager/Radiometer Suite (VIIRS) instruments
  aboard NPP in 2006 and NPOESS beginning in 2009,
  using the current suite of RD and operational
  sensors Terra/Aqua-MODIS, DMSP-OLS, and
  NOAA-AVHRR, supplemented by GOES.
• NexSat products leverage data from NRLs
  numerical weather prediction models (NOGAPS and
  COAMPS) and the National Lightning Detection
  Network.

Page 7 of 27
6
Fast Sat Data Turnaround
Terra Aqua
TDRSS
Svalbard
Direct Capture Ground Station Users (X-band)
Multiple Ground Station Downlinks Per Orbit
(K-band)
White Sands
EOS Era (Today)
Uplink to Comm Satellites
NPOESS Era
Mission Center in Maryland
NOAA Pre-Processing in Maryland (L0-L2
Preparation)
Calibrated and Geolocated Radiance-Level Data for
DoD Users Within 3 hours
Calibrated and Geolocated Radiance-Level Data for
DoD Users Within 30 minutes
Interface Data Processing Segment
NOAA Server (NASA GSFC)
NOAA AFWA FNMOC NAVO NRL
7
Quantitative Products
Enhancements
NWP Models
ch1
ch4
ch3
Sat1
SatN
Sat1
Atmospheric Correction
x2 Up-Sampling
Example Rain Accumulation R(mm) f(ch1,
ch2,chN, u850, v850, etc)
RGB Scaling
Example Pseudo True-Color from MODIS
One week rainfall accumulations ending 25 Feb
2005 0 UTC from the NRL Blended Satellite
Technique (Turk and Miller, 2004)

• Numbers with the Pixels
• Validation and Quality Control
• Detection Efficiency? False Alarms?

• Fast and Simplified Visualization

8
Preparing for NPOESS-Era Capabilities of the
Visible/Infrared Imager/Radiometer Suite (VIIRS)
Page 26 of 27
9
VIIRS Channels
Page 26 of 27
10
Snow Cloud Discrimination

• The improved spatial and spectral resolution of
  VIIRS, leveraging the new 1.38 micron cirrus
  currently demonstrated on MODIS, offers new ways
  to classify elements of an otherwise ambiguous
  scene.

Page 10 of 27
11
True Color City Zooms

• NexSat high spatial resolution (250-m pixels)
  true color zooms over major U.S. cities provide
  users with an VIIRS-quality snap shot of weather
  conditions over their own backyard twice per
  day.

Page 8 of 27
12
Comparing VIIRS-DNB Against DMPS-OLS Nighttime
Visible

• Higher spatial resolution
• 0.74 km vs. 2-5 km OLS
• Higher radiometric resolution
• 4096 vs. 64 gray shades
• Superior temporal sampling
• Every 4 hrs vs. terminator only
• 3-gains for reduced saturation
• Higher SNR
• Calibrated data
• ? The VIIRS DNB represents a paradigm shift in
  the utility scope of nighttime visible data

Page 17 of 27
13
How Nighttime Visible Works

• Reflection Signatures
• Emission Signatures

14
Utilizing Night Time Observations
LEGEND whitesolar illumination, bluelunar
illumination in absence of solar
Northern Hemisphere Winter Moon 70
Northern Hemisphere Late Summer Moon 97
15
Understanding the Role of the Lunar Cycle
20041215 (0.24 full moon)
20041220 (0.75 full moon)
20041225 (full moon)
20041230 (0.81 full moon)
20050104 (0.33 full moon)
16
Lunar Reflection and Natural/Anthropogenic
Emissions
Low Clouds
High Clouds
NEXSAT Low Cloud Product (IR-only RED Low
Cloud)

• Under adequate lunar illumination conditions, the
  VIIRS DNB will provide quantitative information
  on cloud reflectance, allowing for markedly
  improved cloud characterization beyond IR-only
  approaches?carrier ops.

17
Nighttime Dust Mapping from Lunar Reflection and
Multi-Spectral IR
Dust Over Africa?

• Lunar reflection of dust plumes, shown here for
  enhancement of low level (warmer) bright targets,
  will be combined with multi-spectral VIIRS bands
  to provide an optimized day/night detection
  capability?pilot visibility.

18
Nighttime Sea Ice Mapping from Lunar Reflection

• Because reflected moonlight transmits through the
  thin Antarctic clouds, the day/night visible
  imagery reveals the edge of the ice sheet in
  several places indiscernible within the infrared
  image?navigation impacts.

19
Characterizing Nocturnal Thunderstorm Occurrence
Intensity

• An additional utility of the nighttime visible
  band is its ability to detect the lightning
  activity of thunderstorms. In the example above,
  white streaks from the DMSP/OLS correlate with
  NLDN detected strikes?aviation ops.

20
Active Fire Depiction
10/26/2003 0424Z DMSP F15
Fires Isolated
Fires City Lights

• Traditional 3.9 micron fire channel detects all
  hot spots (active smoldering). Flagging new
  light emissions via DNB decouples active fires.
  Requires creation of an up-to-date static city
  lights mask ?recon/surveillance.

21
VIIRS vs. MODIS Degradation
Page 26 of 27
22
Interdisciplinary Environmental Parameters

• Aurora detection and intensity characterization
  from calibrated DNB data. Polar latitudes? high
  temporal refresh to understand evolution in
  advance of GOES-R.

Aurora Borealis (Fall, 2001)

• Possible detection of specific varieties of
  marine bioluminescence from space currently under
  investigation using DMSP-OLS.

Dinoflagellate Emissions
? Impacts to be explored
23
NexSat Training Modules

• Online tutorials are designed to orient new users
  with NexSat products using simple and
  straight-forward illustrative examples, all the
  while tying into the general theme of future
  NPOESS/VIIRS capabilities.

Page13 of 27
24
Hydrological ApplicationsQuantitative Global
Precipitation and Validation
Page 26 of 27
25

• Two Major Factors Limit the Quantification of
  Precipitation from Space
• Revisit Time A single LEO satellite orbits once
  per 90 minutes and (usually) revisits at the
  nearly the same local time.
• Beamfilling The structure of the sensed
  hydrometeors is averaged across the antenna beam
  pattern.

Example from EOS-Terra morning overpass on June
26, 2003. Sunshine is indicated by the white
shades and the red stripe indicates the day-night
terminator. Note how Terra passes over at nearly
the same local time each orbit.
26
Current (10-Satellite) LEO Satellite
Constellation Revisit Time
Color Codes SSMI DMSP F-13/14/15 AMSR-E Aqua AMSU
-B NOAA-15/16/17 TMI TRMM Coriolis Windsat SSMIS F
-16
Revisit Scale White 0 hours Black 6 hours
(shaded boxes represent 15-minute coverage)
27
The Satellite Beamfilling Problem
We dont know the spatial pattern of the
underlying rainfall at the time that the
satellite flies over
Satellite movement
But its only raining in this fraction of the
sensors field of view (e.g., 25 mm/hour)
Therefore, when one interprets the satellite
signal (radiances), there will be a systematic
underestimate of the rainfall (e.g, 10 mm/hour)
Satellite system receives an added signal for any
rainfall that falls within this cone (field of
view)
Earths surface
50-km
(not drawn to scale)
28
Characteristics of the Satellite Sensor Scanning
rectangular map grid
29
Scan-Edge Effects from the Cross-Track
Scanners Changing pixel size changes the observed
precipitation rates
LAND BACKGROUND
scan edges
scan edges
scan center
OCEAN BACKGROUND
30
TRMM TMI/PR 15 Sep 2004 0509 UTC Over-Ocean
TMI cant delineate fine-scale structure
PR-estimated precip is displaced from the
TMI-estimated precip due to parallax
satellite motion
31
TRMM TMI/PR 17 Sep 2004 0636 UTC Over-Land
TMI cant capture the heavy isolated rain events
(the tail of the rainfall histogram, i.e. the
few big events) that the PR picks up
satellite motion
32
NRL Blended Technique Principles
0.1-degree grid map-projection of all data (3600
samples x 1200 lines)
Weight per-pixel instantaneous blended estimates
(blue) smaller as they approach an instantaneous
PMW estimate (red) Time-proximity threshold
decreases with increasing latitude (shorter
revisit at higher latitudes)
single 0.1-degree pixel with an instantaneous
rain averaged to this pixel size
Time (hours) ?
To 1-day Accumulate instantaneous blended and
instantaneous PMW observations 1-day to 1-week
Accumulate 3-hour blended accumulations 1-week to
1-year Accumulate 24-hour blended accumulations
33
24-hour Accumulated Precipitation Ending 10 Jan
2005 at 00 UTC (Top) NRL Blended-Satellite
Technique (Bottom) Navy NOGAPS Forecast
34
Overall Hurricane Isabel Total Rain
Accumulation Between 9-20 September 2003
Environmental factors limiting rainfall?
35
FABIANs cold SST wake crossed by ISABEL max
-2.5 C change observed by drifting buoys.
Region where Isabels rainfall accumulations
weakened
John Hopkins Univ
36

• INTERNATIONAL PRECIPITATION WORKING GROUP (IPWG)
• co-sponsored by
• Coordination Group for Meteorological Satellites
  (CGMS)
• and
• World Meteorological Organization (WMO)
• Endorsed during the CGMS XXIX Meeting
• 23-26 October 2001
• Capri, Italy
• Current Co-chairs
• Joe Turk, NRL-Monterey
• Peter Bauer, European Centre for Medium-Range
  Weather Forecasting, UK
• turk_at_nrlmry.navy.mil
  peter.bauer_at_ecmwf.int

37
Automated Processing Online and Available to the
World
Work is modeled after the pioneering effort of
Dr. Beth Ebert (BMRC/Australia) http//www.bom.g
ov.au/bmrc/SatRainVal/validation-intercomparison.h
tml Similar validation over Europe by Dr. Chris
Kidd (Univ. Birmingham, UK) kermit.bham.ac.uk/ki
dd/ipwg_eu/ipwg_eu.html U.S. Validation by John
Janowiak (NOAA/CPC) www.cpc.ncep.noaa.gov/produc
ts/janowiak/us_web.shtml
38
(No Transcript)
39
www.cpc.ncep.noaa.gov/products/janowiak/us_web.sht
ml
40
Continental Australia including Tasmania All
Latitude Regimes Jan 2003-Sept 2004 Daily
Correlation between Gauge Analysis and Estimates
15 Satellite Algorithms (blended PMW-IR,
PMW-only, Multi-Precip, IR-only)
4 NWP Models (AVN, ECMWF, NOGAPS, mesoLAPS)
summer winter summer winter
summer winter summer winter

• Wide variety in performance of satellite
  techniques
• NWP model performance is superior for winter
  season
• Similar performance in summer season

41
IPWG Validation Results So Far (Still Ongoing.)

• 1. Merging PMW IR estimates (i.e., GEO and LEO
  satellites) provides more accurate estimates of
  precipitation than the separate components can
• Two major systematic biases are apparent in the
  satellite estimates
• a. OVER-estimation over snow-covered regions
• b. OVER-estimation in semi-arid regions during
  the warm season
• When merging PMW IR data, more accurate results
  obtained
• when using IR to transport morph
  precipitation than to use IR to
• estimate precipitation directly
• 4. NWP forecasts generally outperform satellite
  estimates and radar
• during the winter season over the U.S.
• 5. Satellite estimates compare better with radar
  than gauge
• point estimates vs. less-direct / spatially
  complete
• gauges radar

42
NRL-Satellite Points of ContactSteve
Miller miller_at_nrlmry.navy.milJeff
Hawkins hawkins_at_nrlmry.navy.milJoe
Turk turk_at_nrlmry.navy.milTom Lee lee_at_nrlmry.nav
y.mil Kim Richardson kim_at_nrlmry.navy.mil
Arunas Kuciauskas kuciauskas_at_nrlmry.navy.mil
www.nrlmry.navy.mil/tc_pages/tc_home.html
www.nrlmry.navy.mil/nexsat_pages/nexsat_home.html
Please Stop By Soon!
Page 27 of 27
43
Moving Beyond TRMMThe Global Precipitation
Mission
Page 26 of 27
44
Occurrence of High Latitude Precipitation
(Distribution derived from COADS datasets)
Planned GPM Core Coverage
TRMM Coverage
1 mm hr-1 is taken as the approximate threshold
between light and moderate rain rate
45
Estimating Precipitation at Higher Latitudes and
Altitudes The primary difficulties
are lighter precipitation rates snowy/icy/fr
ozen surface defeats current microwave
schemes surface calibration/validation data are
sparse Complex terrain can induce variations
the satellites miss strong variations in
short distances warm rain enhancement on
windward slopes not retrievable GPM (and others)
have driven recent work evaluating additional
channels evaluating deployment of sounder
channels that dont see the surface
46
Global Precipitation Mission (GPM) Constellation
Architecture
Core Reference Satellite (similar to TRMM)
GPM Era
NPP (back up)
GPM Core Calibration-Reference
NPOESS-3
(ATMS)
(CMIS)
NASA-Partner
(GMI , Ku/Ka-DPR)
DMSP-F20
b
TRMM Era
(GMI)
(SSMIS)
METOP (back up)
TRMM
EGPM
NPOESS-2
DMSP-F19
Satellites of Opportunity
AQUA
DMSP-F16/18
(CMIS)
(AMSU)
Dedicated Constellation Members
(EMMR , Ka-NPR)
(SSMIS)
DMSP-F17
CORIOLIS
NPOESS-1
Megha Tropiques
(CMIS)
(MADRAS)
GCOM-B1
FY-3
AQUA (back up)
(PMWR)
(AMSR-F/O)
(AMSR-E)
47
CRL Dual Frequency (13.6,35 GHz)Ku-Ka Band Radar
(DPR)
Measurable range by 35GHz radar
Measurable range by 14GHz radar
DPR radar will measure intense rain in tropics
and weak rain snow in mid/ high-latitudes
DSD using differential reflectivity
tropical rain
mid- high- latitude rain snow
Frequency
strong rain
weak rain snow
Rainrate
new measurable range by addition of 35GHz radar
48
Continental Australia including Tasmania All
Latitude Regimes Jan 2003-Sept 2004 Bias Score
between Gauge Analysis and Estimates
15 Satellite Algorithms (blended PMW-IR,
PMW-only, Multi-Precip, IR-only)
4 NWP Models (AVN, ECMWF, NOGAPS, mesoLAPS)
summer winter summer winter
summer winter summer winter
Bias Score (hits false alarms)/(hits
misses) Range 0 to infinity Indicates
whether the system has a tendency to
underforecast (biaslt1) or overforecast (biasgt1)
49

• A satellite precipitation algorithm validation
  and intercomparison project
• Conducted by The International Precipitation
  Working Group (IPWG)
• Co-sponsored by the Global Precipitation
  Climatology Project (GPCP)
• Routine daily validation of several satellite
  precipitation algorithms against daily rain gauge
  analyses was begun in February 2003 at the
  Australian Bureau of Meteorology
• The NOAA Climate Prediction Center (CPC) began a
  similar validation of algorithms over the United
  States starting in May 2003, followed by a
  European validation in mid 2004
• Most of the algorithms currently being validated
  are "operational" or "semi-operational", meaning
  that they are run routinely in near-real time and
  their estimates are available to the public via
  the web or FTP
• Short-term rain forecasts from a small number of
  numerical weather prediction (NWP) models are
  also verified for comparison

50
Research Baseline Activities

• Enlisting of lunar model for assignment of
  top-of-atmosphere downwelling lunar spectral flux
  as a function of time, location, and lunar phase
  (e.g., new, quarter, half, full)
• Development of a forward model for VIIRS DNB ?
  radiances for quantitative applications.
• Obtaining a near real-time feed of NPP/VIIRS-DNB
  datapossible through NOAA/NASA/DoD NRTPE.
• Incorporation of DNB satellite demonstration
  products in the multi-sensor/model-fusion NexSat
  near real-time processing framework.

NRTPE
51
MODIS Views of Tropical Cyclones
Click here for NRL TC-Web
Hurricane Isabel 1-km, 500-m, 250-m zoom