Presented at the Second Suomi NPP Applications Workshop

1
Flood Detection with S-NPP and JPSSIn the
context of the CEOS Disaster Risk Management
Flood Pilotthe GEO Regional End-to-End Disaster
Systems Task and the NOAA JPSS Proving Ground
Flood Project

• Presented at the Second Suomi NPP Applications
  Workshop
• Hosted by the University of Alabama at the Westin
  Huntsville
• Stu Frye NASA/GSFC/SGT (stuart.frye_at_nasa.gov)
• Mitch Goldberg NOAA/JPSS (mitch.goldberg_at_noaa.gov)
  
• 19 November 2014

Thanks to Pat Cappelaere (Vightel Inc.) and
Sanmei Li (George Mason University) for their
assistance in preparing this presentation
2
Agenda

• Context/Overview
• Why create flood maps
• Modeling (forecasts) versus monitoring (nowcasts)
• Flood monitoring from satellites
• Flood event overview (daily, global, low-moderate
  resolution, can capture entire event and detect
  flood maximum extent)
• Flood detail maps (periodic, targeted, high
  resolution, cover much smaller areas of interest
  over widely spaced moments in time
• International setting for implementation of
  satellites maps
• NOAA NWS Flood Applications

2
3
Context/Overview

• Purpose to improve delivery of satellite data
  and products for societal benefit in
  local/regional settings, but on a global scale
• Method - develop and infuse Earth Observation
  (EO) science applications for monitoring and
  modeling combined with advanced technology for
  data acquisition, processing, and product
  distribution for disaster applications
• Sensor Webs technology compatible with Global
  Earth Observation System of Systems architecture
• Experience NASA/NOAA collaboration built into
  Committee on Earth Observation Satellites (CEOS)
  and Group on Earth Observations (GEO) activities
• CEOS Actions and GEO Tasks address ground
  validation, crowd sourcing, and hand-held clients
  to validate disaster products and services
• Capacity building to infuse standardized web
  servers and clients that provide open access to
  critical disaster management information, data,
  and maps via the internet using common, open
  desktop tools
• Status - task objectives, PIs for each product,
  end user identification, team members, and
  accomplishments to date are contained herein
• Upcoming venues for showcasing USA satellite
  disaster applications UNISDR World Conference
  on Disaster Risk Reduction (WCDRR), World Bank
  Understanding Risk forum, AGU/EGU, IGARSS, ISPRS

3
4
Why Create Flood Maps

• Damaging flood events occur both over very large
  areas (e.g., major river valleys) and, for damage
  assessment purposes, small areas (e.g., urban
  flooding).
• After Hurricane Katrina (2005) along the southern
  Gulf Coast, USA, there was a need for very
  detailed, neighbourhood-level mapping in New
  Orleans of still-standing water over weeks of
  time.
• Yet also there was a need for an overview of the
  entire central Gulf coastline, as the storm had
  caused flooding of many small communities that
  were difficult to access by local authorities.
• NASA/NOAA collaboration producing Global Products
  for Regional Consumption - demonstrating how new
  flood data products can be generated, used, and
  validated through cooperation with regional and
  national agencies in international settings

4
5
Model Forecasts vs. Monitoring Nowcasts

• Typically models predict the amount of flooding
  that will occur, when it will occur, and where
• Run daily to produce outputs that are
  low/moderate resolution
• Based on geophysical parameters and inputs from
  satellite and ground measurements
• Satellite monitoring products are moderate/high
  resolution from imagery
• We need both cross-strapped, openly available,
  with legends imprinted
• Graphs/histograms of whats happening at a
  particular point can be produced from both for
  comparison with gauge data and crowd sourced
  validation inputs

5
6
Monitoring with Moderate Resolution

• Sensors such as MODIS and, potentially, SUOMI-NPP
  VIIRS can provide a kind of flood (and surface
  water resources) data "infrastructure", whose
  strengths are temporal and spatial coverage that
  are a useful base for higher spatial resolution
  sensors. They are always on, and public the data
  distribution obstacles that are sometimes present
  for other high resolution sensors are in this
  case absent.
• It would be very productive if VIIRS-sourced and
  Landsat 8-sourced surface water information could
  be added as services in addition to detection of
  surface water extent. VIIRS would add continuity
  and backup (the MODIS sensors are aging) Landsat
  increases spatial resolution about two orders of
  magnitude (100 - 30 m pixels per single MODIS or
  VIIRS 300 m pixel), at cost of temporal and
  spatial coverage.
• The strength of VIIRS is that
• The swath is 3000km with additional overlap
  between orbits (no gaps) even at the equator and
  with nearly constant spatial resolution across
  the scan (375m nadir 750m edge of scan)
• It provides daily coverage and will become the
  primary afternoon instrument once Aqua MODIS is
  retired
• The problem is the mapping capability is one-off
  in response to individual requests from higher
  upsthey are not routine and are not created with
  end user input except in very narrow settings

6
7
Comparison of MODIS and VIIRS Flood Products

• Issues with clouds and cloud shadows on single
  images
• Clouds can be detected and removed, but MODIS
  cloud mask is 1km resolution for corresponding
  flood images at 250m
• Composites over multiple days show better water
  detection
• Cloud shadows still problematic, shadows are
  detected as open water, but approaches exist such
  as height above nearest drainage (HAND) using DEM
  (e.g., Hydrosheds 90m SRTM) in addition to other
  geometric (height vs. sun angle) techniques
• VIIRS NOAA GMU (VNG) Flood Algorithm (VNGFLOOD)
  resolves most of the cloud shadowing issues
  (better resolution cloud mask plus combination of
  geometric and HAND approaches)
• VIIRS products are being proven more recently in
  a data rich environment (e.g. USA) where ground
  truth matching is being demonstrated - then we
  purvey them to our international partners and
  enlist them in ground truthing exercises in local
  settings

While great strides have been made, do not take
away that were done, the amount of the problems
we are facing and the urgency with which they
need to be dealt are still daunting if we want to
be serious about providing societal benefits to
humanity at large in the near future.
7
8
Higher spatial resolution VIIRS and better
handling of cloud shadows provides better
coverage than MODIS
Bangladesh, August 29, 2014, Left VIIRS, right
MODIS
8
9
VIIRS Flood Map from Namibia May 20 2014TIF
format, 9.5Mbytes
9
10
Example of Vectorized VIIRS Flood Map

• https//api.tiles.mapbox.com/v4/cappelaere.9bf9435
  4/page.html?access_tokenpk.eyJ1IjoiY2FwcGVsYWVyZS
  IsImEiOiJxSjM5MEt3In0.9PYNJ8PzRclvtEh1jkqBuA8/-16
  .093/23.898 (from Namibia May 20 2014 image)
• The compressed topojson file is only 161KB
  published on Mapbox using a simplistic base map.
  But you get a better idea of impact of flood on
  zoomable map.
• A real publisher would publish the data on
  request against a terrain map and with a legend
  and embedded tags to enable the sharing on
  Facebook/Twitter

10
11
MODIS Composite Namibia 19-20 May 2014
11
12
Landsat-8 Surface Water

• Also Northwestern Namibia (Zambezi River) May 20
  2014

12
13
Geo-Social Sharing of Science Data
Easy data dissemination, visualization and sharing
Facebook
Twitter
13
14
Example of Vectorized Frost Product for Kenya

• This data set comes from MODIS night time land
  surface temperature (LST)
• https//api.tiles.mapbox.com/v4/cappelaere.3107bfe
  3/page.html?access_tokenpk.eyJ1IjoiY2FwcGVsYWVyZS
  IsImEiOiJxSjM5MEt3In0.9PYNJ8PzRclvtEh1jkqBuA8/1.3
  68/38.276

15
NOAA JPSS Satellite ApplicationsRiver Ice and
Flooding Application for NWS

• JPSS Capabilities River Ice and Flooding Product

• Societal Impacts

• Floods are one of the most common hazards in the
  U.S.
• Coast-to-coast threat to the U.S. and its
  territories year round.
• More than 8 million people live in areas at risk
  of coastal flooding
• Strong need for near-real-time monitoring the
  evolution and dynamics of ice cover due to
• High frequency of river ice related events
• Significant risk of damages
• Accurate and detailed information on ice
  conditions is necessary for
• Better flood risk assessment
• Timely issuance of flood warnings
• Faster preventive or mitigation measures

• Frequent passes at high risk polar latitudes
• High VIIRS spatial resolution (375m)
• Near real time flood detection at regional scale
  in the USA
• Event-driven flood detection at global scale
• Case-driven high spatial resolution (from 10m to
  30m) flood map generation using digital elevation
  models
• Effective algorithms available in direct broadcast

VIIRS ice-jam flood detection in Galena, AK
during May, 2013
Capabilities Needed

• Evaluate state of river ice to prepare for
  break-up
• Identify flooding caused by ice jam or rainfall
  in remote and urban areas
• Estimate extent and depth of flooding waters

16
River Ice and Flooding Initiative

• NOAA JPSS River Ice and Flood Project Initiative
  began in Nov 2013
• Initiative Objectives
• Test new VIIRS River Ice and Flooding Products in
  operational River Forecast Center (RFC)
  environments
• Algorithms developed by NOAA JPSS Proving Ground
  Program
• Determine the value of these products in their
  use in response to real-world ice jam and
  flooding events
• Implement procedures to transition these research
  capabilities to operations
• Initiative has high interest in NWS River
  Forecast Centers (RFCs)
• Alaska-Pacific and North Central RFCs partners in
  initial actions
• Northeast and Missouri Basin RFCs new 2014
  partners
• Two more RFCs considering joining the initiative
• Products available from Direct Broadcast data
  through The Community Satellite Processing
  Package (CSPP) from direct readout in Univ.
  Alaska Fairbanks and University of Wisconsin, to
  RFC AWIP Systems
• Products evaluated during Winter 2013-14 Project
• Flooding Product has been applied year-round to
  US and international river basins (Paraguay
  request)

17
River Ice and Flooding InitiativeNWS RFC Partners
18
Chena River Flooding 4 Jul 2014
A significant flood occurred in the eastern
Tanana basin, including the Chena River near
Fairbanks AK Alaska Pacific RFC Validation of GMU
River Flooding Product
Heavy Precipitation
Rise in the River
Flooding Verification
GMU River Flood Product
19
Summary

• Flood maps are useful to disaster managers, civil
  protection, and the public at large, but are
  still not being provided routinely or made
  available from open repositories
• We can brag about the accomplishments, but need
  to emphasize the problems and why they are
  significant.
• We do need near realtime flood extents but we
  also need to composite the results with other
  products and models.
• We need automation for that... which means we
  need to have an API in place to generate those
  products on-demand
• We need to address product delivery to
  "disadvantaged" end-users (and not just
  scientists with high-end software)
• We need to address mobile platforms access and
  visualization with common open tools that users
  already have
• We need to figure out a way to fund incorporation
  of new capabilities (better DEMs, new radar
  satellites, high res optical, etc) into a
  seamless stream of delivery that can inform
  situational awareness for affected users

19
20
Conclusion

• NASA/NOAA collaboration has worked well and been
  demonstrated with both national and international
  flooding events
• That collaboration needs to be nurtured with
  further interactions on national and
  international fronts exploiting in-roads made
  through CEOS, GEO, SERVIR, and other foreign,
  regional, and international agencies to expand
  access to and use of products, return of
  validation results, and implementation of further
  product improvements
• Thank You

20