1. Ontario Provincial Flood Forecasting and Warning workshop
Operational Monitoring of River Ice Break-up Conditions using RADARSAT-2.
A. Deschamps, J. van der Sanden, G. Choma, J.S. Proulx-Bourque, G. Marquis
Ontario Provincial Flood Forecasting and Warning workshop
Vaughan, Ontario
September 15, 2015

2. Outline Emergency Geomatics Service (EGS)
Brief history - river ice monitoring at Natural Resources Canada
Basic concept of RADARSAT-2 applied to river ice breakup monitoring
Examples from 2015 Pilot Study
Future Work

3. Since 2006, the Emergency Geomatics Service (EGS) supports emergency response activities in Canada by producing GIS-ready satellite-derived information products to a large community of emergency responders (federal & provincial)
Primarily support flood events
2015 Pilot Study, use RADARSAT-2 derived ice condition information during breakup season
Products provide near-real time situational awareness to assist emergency authorities with public safety and critical infrastructure risk

4. How is EGS Activated? (24/7) (24/7) (24/7) Provincial
EMO
Public Safety Canada, Gov. Operations Centre (GOC)
(24/7)
NRCan Dept. Emergency Operations Center (DEOC)
(24/7)
NRCan/ESS Canadian Hazard
Information Service (CHIS)
(24/7)
This is how we have been activated in past
DND
(DART)
CCMEO Emergency Geomatics
Service (EGS)
(International)
Note: Specific needs and the technical feasibility of requests are assessed on a case by case basis by the EGS team in collaboration with PS and partners.

5. Radar Remote Sensing of River Ice - NRCAN
Pilot Project for 3 rivers in Northern Ontario.
+“response” products for additional rivers in ON and NB.
At the request of Public Safety Canada (PS), EGS provided ice breakup products in demonstration mode for responses in ON and MB.
Continue on with a second year of pilot project to improve products and services.
Research on River Ice Breakup monitoring using R2 based on IPY funding to Dr. Van der Sanden at CCRS.
2013
2014
2015
2016
Development of IceBC tool at the request of Ontario MNRF.
IP development and released to the province of Ontario in 2013.
EGS continues to improve the automation of the IceBC tool.
Science to Ops
Transition
R&D
Operations
Geomatic Support

6. 2015 Pilot Study –RADARSAT for River Ice Breakup Monitoring
Conducted a Pilot Project in collaboration with Public Safety Canada and the Ontario Ministry of Natural Resources – Surface Water Monitoring Centre.
Goal: Demonstrate the benefit of RADARSAT-2 for operational monitoring of river ice-breakup on 3 rivers in Northern Ontario.
Moose River
Albany River
Attawapiskat River

7. Ice Jams – A Costly Hazard
The impact of ice covered rivers on Canadians typically peaks at the time of breakup; in particular, when ice jams form and cause flood and/or damage to nearby infrastructure.
The annual direct costs of ice jamming and associated flooding in Canada estimated at $60 million, indirect costs can be much higher. (Gerard and Davar, 1995)
Photo Credit
Embâcle sur le fleuve Saint-Jean à Perth-Andover Crédit photo : Étienne Dumont/Radio-Canada
Fort-Albany Flood Watch
Other ideas:
Cost reference for Kash:
hhttp://
Cost reference for St.Lawrence:
Morse, B., and Hicks, F., 2005, Advances in river ice hydrology, 1999–2003: Hydrological Processes, v. 19, no. 1, p. 247–
Fort Albany, ON (May 10, 2014)
Perth-Andover, NB (April 19, 2015)
Photo Credit: Fort-Albany Flood Watch
Photo Credit: Étienne Dumont/Radio-Canada

8. Radar Remote Sensing of River Ice
Satellite derived products provide a synoptic overview of ice breakup areas:
Can provide early warnings of problem.
Can help focus on areas of concern.
Best used in conjunction with ground and air surveys.
RADARSAT offers much potential as tool for collecting information about river ice condition. This can be explained from:
its weather and daylight independent imaging capability,
sensitivity to the presence of water,
sensitivity to surface roughness,
capability to penetrate frozen (dry) ice cover,
sensitivity to structural features within the ice volume
The backscatter of wet spring ice is governed by the surface roughness

9. Radar Remote Sensing of River Ice
Frozen (dry) winter ice and melting (wet) spring ice interact with radar waves in a distinctly different fashion!
Ice Breakup Classification (IceBC) algorithm exploits the sensitivity of SAR to differences in the roughness of breaking (wet) river ice cover.
The backscatter of wet spring ice is governed by the surface roughness

10. IceBC for Breakup Monitoring - Workflow
Post-processing: classification filtering, applying Land mask and map Generation.
Product validation
IceBC – Ice Classification (Standalone Executable)
Pre-processing: image ‘calibration’, speckle filtering, and orthorectification/mosaic, extract incidence angle.
Radarsat-2 Image +
DEM
Mask applied to ice cover condition classification product
KMZ Ice cover condition information product
Ice cover condition classification
Land/water mask, generated from NHN/Landsat
FULLY AUTOMATED!

11. IceBC for Breakup Monitoring - Workflow
Based on Python code + Matlab exe
Software: PCI Geomatica Global Mapper (KMZ)
Inputs:
RADARSAT-2 image (C-band SAR, HH-Polarization)
NOTE: Public Safety Canada can provide access to satellite imagery for Emergency Management purposes.
Water mask (NHN 1:5000)
DEM mosaic (CDED 1:50000)
Outputs:
Ice breakup classification in TIF and KMZ format.
Selecting images with larger incidence angles (>36°)
Tool Ingests selected beam modes (Fine, Standard, Wide, EH, Ultra-Fine and Wide Ultra-Fine)

12. IceBC - Products
Products discriminate between ‘water’ and three classes each for ice cover conditions identified as ‘sheet ice’ or ‘rubble ice’.
Sheet ice covers are characterized by smooth textures whereas rubble ice covers have rough textures.
Ice jams are a form of rubble ice with rough to very rough textures (shown in red & fushia) => Validated by field or repeat observations
Repeat observations or field validation is needed to determine if an ice cover that exhibits the texture of an ice jam is indeed jammed
Sheet Ice Cover (smooth) Rubble Ice Cover (rough)

13. Pilot Study: Validated Ice Jam on Moose River (ON) - April 23, 2015
Ground validation photo provided by Ontario MRNF

14. Pilot Study: Ice Condition on Moose River (ON) - April 30, 2015
Ground validation photo provided by Ontario MRNF

15. Operational Ice Breakup Monitoring - Challenges
Transition from dry winter ice vs. wet spring ice affect the type of SAR scattering.
Compile weather, streamflow, snow cover, optical images
R2 temporal signature & breakup onset
Diurnal effects are prevalent in early breakup phase
Cycle of freeze-thaw cycles prevalent in Southern Canada
High winds (>6 m s-1) and rapids
increases the roughness of water and complicates the discrimination of the open water and sheet ice classes.
Wet snow on ice or standing water on ice can be expected to be misclassified as water.
Dynamic ice condition at breakup complicates validation
The IceBC algorithm was developed to work with melting / wet ice conditions that are usually present in the spring breakup season.
Application of the method to SAR images acquired during winter / dry ice conditions will produce results that are not in line with the legend that accompanies the products.
Current Solution: Monitor temperature, streamflow, Modis/Landsat, snow cover in conjunction with and SAR backsatter behaviour over time to determine onset of breakup and start of product delivery.
(See example in next slides…)

16. Stage 1: High backscatter in winter
March 26, 2015 (F1 morning pass)
For illustration only (Fort McMurray)
Polygon #2 is at white crosshair!
Determining the onset of breakup using Radarsat-2 temporal signature (assumes acquisition with suffcient frequency).
Stage 1: You start with a high backscatter in winter (dry ice and snow cover; volume scattering)
Stage 2: The backscatter drops when the snow starts to melt in early spring (wet snow/water on top of ice cover; surface scattering)
Stage 3: The backscatter increases following snow melt (bare wet ice, surface scattering)
Stage 4: Start of breakup
IceBC outputs are only reliable in the stages 3 and 4 of the breakup process

17. Stage 3: The backscatter decreases following snow melt
April 02, 2015 (W3 morning pass)
For illustration only (Fort McMurray)
Polygon #2 is at white crosshair!
Determining the onset of breakup using Radarsat-2 temporal signature (assumes acquisition with suffcient frequency).
Stage 1: You start with a high backscatter in winter (dry ice and snow cover; volume scattering)
Stage 2: The backscatter drops when the snow starts to melt in early spring (wet snow/water on top of ice cover; surface scattering)
Stage 3: The backscatter increases following snow melt (bare wet ice, surface scattering)
Stage 4: Start of breakup
IceBC outputs are only reliable in the stages 3 and 4 of the breakup process

18. Stage 3: The backscatter drops when the snow starts to melt
April 19, 2015 (F1 morning pass)
For illustration only (Fort McMurray)
For illustration only (Fort McMurray)
Polygon #2 is at white crosshair!
Determining the onset of breakup using Radarsat-2 temporal signature (assumes acquisition with suffcient frequency).
Stage 1: You start with a high backscatter in winter (dry ice and snow cover; volume scattering)
Stage 2: The backscatter drops when the snow starts to melt in early spring (wet snow/water on top of ice cover; surface scattering)
Stage 3: The backscatter increases following snow melt (bare wet ice, surface scattering)
Stage 4: Start of breakup
IceBC outputs are only reliable in the stages 3 and 4 of the breakup process

19. Stage 4: Breakup April 26, 2015 (W3 morning pass)
For illustration only (Fort McMurray)
Polygon #2 is at white crosshair!
Determining the onset of breakup using Radarsat-2 temporal signature (assumes acquisition with suffcient frequency).
Stage 1: You start with a high backscatter in winter (dry ice and snow cover; volume scattering)
Stage 2: The backscatter drops when the snow starts to melt in early spring (wet snow/water on top of ice cover; surface scattering)
Stage 3: The backscatter increases following snow melt (bare wet ice, surface scattering)
Stage 4: Start of breakup
IceBC outputs are only reliable in the stages 3 and 4 of the breakup process

20. Future – Research Work
Integration of Sentinel-1 images in EGS workflows
Transition algorithms from Radarsat-2 based operational service to the RADARSAT Constellation Mission (RCM)
Diverse polarizations typically yields more information to help with target discrimination (i.e. wet/dry ice)
More frequent revisits (daily)
Standard coverages
Investigate the utility of Synthetic Aperture Radar (SAR) for river ice freeze-up monitoring
Improve robustness of flood algorithms and initiate research for flood mitigation
Including urban flood & flooded vegetation
Phase information from Compact Polarimetry (CP) to identify the transition from winter ice (dry – volume scattering) and spring breakup ice (wet – surface scattering).
RH/RV or HH/VV to improve ice mapping in difficult cases (e.g. high wind condition, steep incidence angles).
Standard Coverages:
-Operational monitoring of wide area
-Simplify the planning phase and eliminate conflicts.
-Facilitate change detection approaches (i.e. damage assessment)

21. RADARSAT Program Evolution
2018:
RCM
First operational civilian SAR satellite
Important R&D component
GoC owned
Ops extended to DND, DFO, NRCan, and EC
Numerous sci & ops modes
MDA owned (PPP)
Fully operational
Selected R2, Compact Pol. Coherent Change Detection
Enhanced ship detection
GoC owned
Fast tasking, fast delivery
Source: Canadian Space Agency

22. River Ice Breakup Pilot Study Phase II for 2016
Future – Operations
River Ice Breakup Pilot Study Phase II for 2016
Continue to improve the River Ice Breakup (IceBC) tools to facilitate internal operations.
Tools can be shared with our partners via agreements.
Continue to enable our partners to use SAR derived ice condition information during ice breakup season beyond response.

23. Thank You! Questions? Emergency Geomatics Service (EGS):
Director: Yvan Désy
Project Manager: Geneviève Marquis
Acting Operational Lead: Alice Deschamps
CCRS SAR and Ice Expert: Joost Van Der Sanden
EGS Team: Vincent Decker, George Choma, Pierre Gravel, JS Proulx-Bourque, Brigitte Belzil, Roger Bouchard, Charles Papasodoro