The 10^th International Circumpolar Remote Sensing Symposium and
29^th Canadian Symposium on Remote Sensing
June 2-5, 2008

Abstracts of verbal presentations are listed according to Session, and poster presentations are listed below.

Session 1 - GIS, Data Integration and Planning

Yukon Government Corporate Spatial Data Infrastructure

Lauren Crooks, Spatial Data Administrator, Geomatics Yukon Information, Communication & Technology Branch, Government of Yukon, Box 2703, Whitehorse, Yukon Y1A 5T8, phone 867-393-7084, fax 867-667-5304, email Lauren.crooks@gov.yk.ca

Nicole Parry, Spatial Data Administrator, Geomatics Yukon Information, Communication & Technology Branch, Government of Yukon, Box 2703, Whitehorse, Yukon Y1A 5T8, phone 867-667-5844, fax 867-667-5304, email Nicole.parry@gov.yk.ca

During the past 4 years, the Yukon Government has redefined and rebuilt their spatial data infrastructure.  Geomatics Yukon provides service to Yukon Government, external partners, and the public.  The Yukon Government’s Corporate Spatial Data Infrastructure (CSDI) includes the supporting technology, tools, policies, standards, and partnerships associated with managing and distributing spatial data for the Yukon Government.  All these components, including spatial data warehouse, operational database, internet mapping, and metadata, are integrated to meet their clients’ needs.  Key components, their interactions, reasons for decisions, and how clients benefit will all be discussed.

The Corporate Spatial Warehouse (CSW) is the core of the CSDI. This is a “one copy” read only central repository of spatial data. All data housed in the CSW is modeled, fully described, and has a designated custodian who maintains their data. The CSW does not replace the custodian’s operational spatial database. There is a corporate operational database that is currently being developed which will eventually house the majority of the Yukon operational data, using the same technology and standards as the CSW.  The CSW is distributed internally through ArcSDE, and distributed externally to external partners and the public, through a CSW catalog web mapping service, data download files, and through web mapping.

The web mapping component is the primary method for distributing Yukon spatial data and tools to external partners and the public.  The web applications provide a series of windows into the CSW, based on the business information required, such as Mining Lands, Oil and Gas, or Geology. External agencies, such as the Yukon Environmental Socio-economic Assessment Board (YESAB), have been able to utilize this infrastructure to build their own web application. These web applications enable clients to use GIS functionality and tools to interact with this data.  Standards and protocols have been developed to guide application development, ensure a common look and feel, create continuity of spatial data and tools, and minimize web mapping administration.   This allows everyone to benefit from minimal software and hardware costs, administration, and shared tool development.

The Yukon Government has 500 GB of imagery on a shared image server. They are currently experimenting with an improved method for distributing imagery internally. The primary option being considered is using the current CSW and ArcSDE technology. Imagery data, especially satellite imagery, is an important GIS data source in the Yukon as it aids visualization. The current CSW is being considered due to its multi-user access from the central repository and the ArcSDE pyramid functionality.

North Yukon Regional Planning Database - Yukon Planning Atlas

Jeff Hamm, Senior Planner, Yukon Land Use Planning Council, 201-307 Jarvis St, Whitehorse, Yukon Y1A 2H3

Shawn Francis, Senior Planner, North Yukon Planning Commission, 201-307 Jarvis St, Whitehorse, Yukon Y1A 2H3

Presented by: Jeff Hamm, phone 867.667.7397, fax 867.667.4624,  email jeff@planyukon.ca

The Yukon Land Use Planning Council has developed a web based Planning Atlas to meet two objectives:(1) To provide public access to information assembled for preparation of Yukon regional land use plans; (2) To provide simple web-based tools for  integration and analysis of planning themes. The data assembled by the North Yukon Planning Commission included biophysical land classifications, land use footprints and important ecological and cultural areas identified through science and traditional knowledge approaches. These regional data, and similar information from the Peel Watershed, are being integrated with the Yukon Planning Atlas. Thematic maps, including Land Status, Heritage Resources, Sensitive Environmental Areas, can be displayed, manipulated and printed. The (draft) North Yukon Land Use Plan recommends a cumulative effects framework for monitoring ecological integrity of the landscape. The Yukon Planning Atlas has advanced geographic processing features to perform vector/vector and vector/raster data summary. This allows CE indicators such as Linear Density (km/km2 of linear disturbance) and Surface Disturbance (as %) to be calculated for Landscape Management Units or user defined areas. Advanced features for up/dowloading, multi-media access and advanced query are available to authorized users.

Session 2 - Environmental Monitoring

Satellite Analysis Procedures for National Scale Land Cover Map Update

Donald G. Leckie¹, Morgan Cranny¹, Michael Henley¹, Joan Luther², and Olivier Van Lier²

¹Canadian Forest Service, Natural Resources Canada, 506 West Burnside Road, Victoria, BC  V8Z 4X2, phone 250 363-0624, fax 250 363-0775, email dleckie@pfc.forestry.ca
²Canadian Forest Service, Natural Resources Canada, Corner Brook, Newfoundland

National scale land cover maps generated from medium resolution satellite data such as Landsat have been produced for various countries and regions of the world.  There is now a need to update them.  It is desirable to have the new updated map compatible with the original map such that, if land cover of an area is different on the updated map it reflects a true change in land cover rather than a spurious difference related to the vagaries of land cover classification.  It is also desirable to capitalize on the work done in the original land cover classification to make the update process as efficient as possible.  Land cover update procedures were explored, developed and tested.  Update of the circa 2000 land cover map of the Earth Observation for Sustainable Development of Forests (EOSD) project of the Canadian Forest Service, Canadian Space Agency and partners is the target or example application.  The principle is to identify areas of change and classify the new cover types within these areas.  Non-change areas will remain the same class as the previous land cover classification.  Efficiencies are gained as often on only a small fraction of the land base needs to be examined and given a new land cover class, and the nature of these new land covers is limited.

Various change mask procedures to determine potential areas of land cover change were examined including spectral change indices and 2-date unsupervised classification.  A combined wetness-greenness change mask was an effective spectral mask, but a change mask from a K-means clustering algorithm using all 6 bands from each date was selected as the nominal approach.  It gave more flexibility than threshold methods and provided information during the change labelling process that was useful to later land cover classification.  It was determined that the change mask should be generous (i.e. be inclusive of change at the risk of incorporating non-change areas).  The standard EOSD K-means single date (time 2) land cover classification was then applied using the change mask as input and clusters within the change mask were labelled as to land cover.  This was a simple and effective procedure and permits the land cover to be the same as the time 1 classification even if there was no change in cover class under the change mask.  The method was applied to four diverse trial sites (southwestern Northwest Territories, Prince George, BC, Central Quebec and western Newfoundland).  The method worked well, was able to account for ephemeral changes that do not change cover class such as moisture conditions and phenology, was compatible with the T1 land cover map, and was quick and cost effective.  Such procedures are viable and suitable for operational implementation at national, regional and local levels.

20 Metre Bio-Physical Variability of the Northern Mixedgrass Prairie

Joseph M. Piwowar, Department of Geography, University of Regina, Regina, SK  S4S 0A2, Canada, +1.306.585.5273, email joe.piwowar@uregina.ca

The northern Great Plains of North America are significant for three reasons: (i) They are the source for much of the food produced in North America; (ii) They encompass the last remaining native habitats of many endangered species; and (iii) Their vulnerability to climate change is second in North America only to the Arctic.  Paleoclimate records for the northern Great Plains show prolonged droughts far more extreme than those that have been experienced since European settlement.  There is concern that one of the most immediate impacts of global warming in this region will be a return to past conditions, putting tremendous strains on the sustainability of natural, physical and social prairie infrastructures.

During the summer months of  2006 and 2007 field data were collected from over 50 sites in Grasslands National Park of Canada.  These sites were selected from 15 spectral and temporal classes that were found to be representative of the vegetation found within the northern mixedgrass prairie ecosystem.  At each site, measurements were made of the vegetation type, percent ground cover, soil colour, soil moisture, and in situ spectral reflectance at 5 locations within a 10 m radius.

This research documents the variability of the specific measurements from within each site, between sites of the same class, and between different site classes.  These observations are then compared to contemporary SPOT HRV/HRG images to document the within-pixel variability on spaceborne spectral reflectance.

An operational system for deforestation estimation for Canada

Donald G. Leckie, Canadian Forest Service, Natural Resources Canada, 506 West Burnside Road, Victoria, BC  V8Z 4X2, phone 250 363-0624, fax 250 363-0775, email dleckie@pfc.forestry.ca

Canada has producing estimates of deforestation area and related greenhouse gas emissions since 2006 based on a system dependant on remote sensing data.  Area estimates are derived using Landsat imagery from circa 1975, 1990, 2000 and recently 2007.  Change enhancements highlight potential areas of forest clearing and further examination of the imagery and ancillary data confirms that the sites were forest at time 1 and another land use at time 2.  Deforestation here uses the international definition of a permanent loss of forest land to another land use.  Thus forest harvest is excluded.  Ancillary data includes circa 1990 winter Landsat imagery, near 1990 aerial photography, and GIS coverages of roads, settlements, pits and quarry license areas, and sometimes forest inventory.  Field work (mostly aerial observation) is conducted to calibrate the interpretation, validate results and resolve selected problem interpretations.  Much of the initial mapping is done commercially on contract with various quality control steps, field validation and final review.  The deforestation mapping is done on a stratified sample across the country and scaled to give estimates on an ecoregion and country-wide basis.  Records data such as reservoir flooding, hydro lines, forest road and pipelines are also assembled, vetted and used where appropriate.

To date, almost 70 million ha has been mapped for deforestation, yielding approximately 260,000 ha of deforestation in 58,000 events.  The frameworks, methods and operational considerations of the deforestation monitoring program are discussed.

Time series analysis of the mountain pine beetle outbreak in British Columbia

Nicholas R. Goodwin^*, Department of Forest Resource Management, 2424 Main Mall. University of British Columbia, Vancouver, Canada. V6T 1Z4, phone 604-822-6452, fax 604-822-9106, email Nicholas.goodwin@ubc.ca
* Presenting author

Nicholas C. Coops, Department of Forest Resource Management, 2424 Main Mall. University of British Columbia, Vancouver, V6T 1Z4, Canada

Michael A. Wulder, Canadian Forest Service (Pacific Forestry Center), Natural Resources Canada, 506 West Burnside Road, Victoria, V8Z 1M5, Canada

Steve Gillanders, Department of Forest Resource Management, 2424 Main Mall. University of British Columbia, Vancouver, V6T 1Z4, Canada

Todd A. Schroeder, Department of Forest Resource Management , 2424 Main Mall. University of British Columbia, Vancouver, V6T 1Z4, Canada

Trisalyn Nelson, University of Victoria, Department of Geography, Victoria, V8W 2Y2, Canada

Time series analysis of satellite imagery provides a valuable tool for environmental and forest managers to evaluate forest disturbance events as well as land cover change at broad spatial scales. In this study, we utilise eight Landsat scenes collected over a 14 year period (1992 to 2006) to examine the spectral changes due to an outbreak of mountain pine beetle outbreak in north-central British Columbia, Canada. After pre-processing and normalizing the eight scenes using a Multivariate Alteration Detection (MAD) transformation, decision rules were applied to classify spectral trajectories of the Normalised Difference Moisture Index (NDMI). From the classified trajectories, key parameters were extracted including the presence of beetle disturbance and timing of stand decline. The accuracy of discriminating beetle attack from healthy forest stands was assessed both spatially and temporally using three years of aerial survey data (1996, 2003, and 2004) with results indicating overall classification accuracies varying between 71 and 86%, which is comparable to previous efforts. As expected, the earliest and least severe attack year (1996), recorded the lowest overall accuracy. The relationship between the timing of stand attack (i.e. moderate to severe beetle infestation) and NDMI (year of minimum NDMI and initial year of detected disturbance) was also explored. Overall, we believe there are a number of advantages with using an approach based on decision rules to assess insect infestation dynamics for time series data. This includes the capacity to integrate our knowledge of the physiological behaviour of mountain pine beetle infested forest stands with the form of spectral trajectories, and to assess the cumulative area of mountain pine beetle attack over a number of years using a single classification process. However, to quantify the timing of mountain pine beetle attack a higher temporal frequency of images is required.

SAR and Optical EO Data for Monitoring Permafrost in the North

Brisco, B., Budkewitsch, P., Short, N., Murnaghan, K., and Perrott, T.

Canada Centre for Remote Sensing, 588 Booth St., Ottawa, ON, Canada K1A 0Y7

With the rapid warming in the North the subsequent degradation of permafrost is becoming an increasing concern leading to widespread hydrological change.  Due to the remoteness and large areas being affected remote sensing provides key sources of data for mapping and monitoring the changing surface conditions.  Synthetic aperture radar and optical data provide complementary information and can be used synergistically for this application.  In this presentation the use of multi-frequency SAR and optical data from a range of platforms are reviewed for a number of test sites in the North.  Sites being investigated include Herschel Island and Old Crow Flats in the Yukon Territory and parts of the Fosheim Peninsula on Ellesmere Island, Nunavut.  The work includes the use of change detection techniques using multi-source data as well as InSAR investigations with both C and L band SAR.  The current status and results of these investigations will be presented at the symposium.

Session 3 - Arctic Vegetation

Modelling Habitat Distribution of Arctic Plant Species at a Regional Scale: An Example from Svalbard in the European Arctic

G. Arnesen¹, D. Joly², L. Nilsen¹

¹University of Tromsoe, Department of Biology, Tromsoe, Norway
²University of Franche-Comte, Besancon, France

geir.arnesen@ib.uit.no

The aim of this project is to develop a model of regional patterns of distribution among selected species of arctic vascular plants. Study area is the island of Spitsbergen, which is the largest island in the archipelago of Svalbard in the European Arctic. The Spitsbergen Island covers about 37 800 km², and provides diverse ecological conditions for plant growth. The interior parts are, for instance, dissected by several fiords surrounded by lowlands that to some extent are protected from direct ocean influences, and thus have less precipitation and longer periods of sunshine during summer. On the other hand, the coastal parts of the island are characterized by colder, wetter and more foggy climate, the east coast being the coldest. The island also displays a considerable variability of bedrock, ranging from hard, felsic rocks forming acidic substrates, to limestones and evaporites producing alkaline soil conditions. Various sedimentary rocks (pelitic shales and schists) develop substrates of intermediate acidity.

In order to model plant habitat distribution we utilize the niche based approach using the ecological niche expressed by the available environmental variables as an estimate for the habitat. The model is established based on spatial distribution of observed plant species and quantitative responses to environmental variables such as temperature, bedrock, hydrology, aspect and slope.

We assume that temperature during the growth season, and soil chemistry accounted for by the local bedrock, are the two most important environmental factors determining plant distribution at a regional scale in Spitsbergen. Hence, we emphasised the development of raster layers representing the ecological features soil chemistry and temperature.

To address soil chemistry, a bedrock map was reclassified into four categories of bedrock representing the soil acidity gradient. The reclassification was based on the likeliness of the different bedrocks to provide the anions CO₃^{– –} and HCO₃^– to the carbonate buffer system. This is connected to the mineralogical composition of the bedrock as well as its resistance to chemical weathering.

Regarding temperature, different approaches were used to find relations to the distribution of plant species. In Svalbard, cumulative positive daily or monthly temperature during growth season seems to show the most promising results. Rather than using mean temperature we made temperature layers representing daily cumulative temperature.

Floristic data were sampled from selected spots across the island using an adjusted version of the grid frequency method. Additionally, a large amount of presence/absence data derived from arctic herbaria databases in Norway are available. Preliminary results show good correlations between the distribution of some species and the selected environmental data. Modelled habitat distribution maps of selected plant species in Spitsbergen Island, at approximately 100 by 100 m resolution, are presented.

A New Circa-2000 Land Cover Map of Northern Canada at 30m Landsat Resolution

Dr. Ian olthof, Dr. Rasim Latifovic, Dr. Darren Pouliot

Canada Centre for Remote Sensing, 588 Booth Street, Ottawa, ON K1A 0Y7

Proposed presenter: Ian Olthof, email iolthof@ccrs.nrcan.gc.ca, phone (613)-947-1233, fax (613) 947-1406

The Centre for Topographic Information (CTI) is currently assembling a complete land cover of Canada at Landsat (30-m) resolution by combining separate maps of forest and agriculture regions. Previously, land cover was missing over Northern Canada at the required resolution to complete the coverage, therefore CTI contacted the National Land Cover Characterization (NLCC) project at the Canada Centre for Remote Sensing to provide the missing northern portion. Orthorectified circa-2000 Landsat data from CTI Geogratis were acquired for Northern Canada from the treeline to the northern tip of Ellesmere Island and were combined into several radiometrically-balanced large-area mosaics. Literature on northern land cover and vegetation mapping as well as numerous northern vegetation surveys were examined to determine an optimal set of land cover classes to map and provide some reference information to assist class labelling. Field data gathered during numerous northern campaigns over the past few years were combined with land cover information from maps of protected areas generated by other government agencies such as Parks Canada, the Geological Survey of Canada and Territorial Governments to form a reference dataset for training and validation. In addition to achieving CTI’s objective of releasing a national 30m land cover product to the public, certain emerging northern issues related to wildlife conservation and resource development can now be addressed where previous national and northern maps were too generalized spatially or thematically to meet these needs. The new fine-resolution northern map offers better thematic detail than previous national products and better spatial detail than previous national and northern-specific vegetation maps that were generated from coarse resolution satellite imagery.

Vegetation Mapping and Estimation of the Extent of Near-surface Permafrost in the Mackenzie Delta, Northwest Territories

T-N. Nguyen, D.J. King, C.R. Burn

Department of Geography and Environmental Studies, Carleton University, 1125 Colonel By Drive, Ottawa, ON, Canada, K1S 5B6

Correspondence to: Thai-Nguyen Nguyen, Geological Survey of Canada, Natural Resources Canada, 601 Booth Street, Ottawa, ON, Canada K1A 0E8, email: tnnguye2@connect.carleton.ca, phone (819) 771-1509

This research investigates the proportion of the Mackenzie River delta, underlain by Near-surface permafrost (NSP). NSP, defined here as permafrost within 3 m of the ground surface, is widely encountered beneath exposed ground of the Mackenzie Delta. It is a key component of northern environmental systems because it influences terrain stability, and surface hydrology. With increasing industrial development, and a rapidly changing climate in the North, knowledge of the spatial distribution of NSP is critical for land-use planning, as terrain behaviour varies significantly between frozen and unfrozen ground. Climatically, permafrost should be continuous in the Mackenzie Delta, and underlies more than 90% of the exposed ground. However, the most recent Permafrost map of Canada, using sparse ground temperature data, classifies the delta as discontinuous permafrost. There have been few field and remote sensing studies investigating permafrost extent throughout the delta. The objectives of this research were to assess if the distribution of near-shore vegetation communities can be used to predict NSP presence, using first an intensive field campaign, and second remote sensing classification methods. The resulting maps could then be used to estimate the extent of NSP. The field component of this research confirmed that permafrost is ubiquitous beneath spruce forests in the delta, but there is some variation in its occurrence near channels, where snow drifts may accumulate. On point bars and alluvial islands, NSP was absent beneath Willow-horsetail vegetation communities throughout the delta, and beneath Horsetail communities in the southern and central delta. NSP was present beneath all other vegetation communities as well as in other land surface types. In the remote sensing analysis, NDVI and MSAVI as well as texture information were found to be useful for discriminating between vegetation communities. Three classifiers were tested: Maximum Likelihood (ML), Artificial Neural Networks (ANN), and Linear Spectral Unmixing (LSU). ML achieved the highest overall classification accuracies of 84%, 82%, and 83% for the southern, central, and northern delta image, respectively. LSU was useful in studying vegetation gradation from one community to another. The vegetation maps produced from the ML classification showed that NSP lies beneath 93%, 95%, and 96% of the land surface in the southern, central, and northern delta, respectively. The Mackenzie Delta is therefore part of the continuous permafrost zone.

The Effects of Landscape Age on Circumpolar Distribution of Arctic Vegetation

Martha K. Raynolds, PhD candidate and Donald A. Walker, Professor, University of Alaska Fairbanks, 311 Erving, Box 757000, University of Alaska, Fairbanks, AK 99775

Proposed Presenter: Martha Raynolds

Session Topic: Arctic vegetation

Phone, fax, email: 907-474-6720, 907-474-6967, fnmkr@uaf.edu

An understanding of the many different factors controlling the distribution of arctic vegetation factors will allow better predictions of changes expected to occur under different climate change scenarios. Vegetation communities in many parts of the Arctic are relatively young, having established only since landscapes were deglaciation in the Late Pleistocene. This study examines the effect of the age of landscapes on the distribution of arctic vegetation.  We compared the distribution of vegetation types as mapped by the Circumpolar Arctic Vegetation Map with AVHRR satellite measures of greenness (NDVI) and age since glaciation as mapped in Ehlers and Gibbard’s 2004 summary of Quaternary glaciations. The time since deglaciation varies from years, in areas where ice caps and glaciers are presently melting, to millions of years in areas that were not glaciated during the Quaternary.  Some Arctic landscapes are also young because they emerged relatively recently from the ocean due to glacial rebound, or formed from relatively recent sedimentation such as on river deltas. Most of the older arctic landscapes occur east of Greenland and west of the Taimyr Peninsula in Russia.  The age of landscapes generally decreases with elevation. The vegetation types most commonly associated with the oldest landscapes include tussock-sedge, dwarf-shrub, moss tundra, and sedge-shrub wetlands.  Most of the Arctic, including most bioclimate zones and most vegetation types, showed increasing NDVI with landscape age up to around 100,000 years, followed by a decrease.  Landscape age accounted for 40-50% of the variation in NDVI for landscapes younger than 100,000 years. The coldest parts of the Arctic (Subzone A) and vegetation types that grow primarily in these areas did not show any trend with landscape age.  The decrease in NDVI with time is likely due to the effects of paludification, whereby plant communities build-up organic material (live and dead) over time and insulate the soil from summer warming.  This results in a thinning of the active layer, restriction of drainage, and acidification of soils.  These conditions favor the vegetation types shown by this study to be most common on older landscapes.

Session 4 - Data Integration and Data Mining, Modeling

Northern Mapping – A Yukon Perspective

Suzanne Brunke, Geospatial Project Specialist, MDA GSI, 13800 Commerce Parkway, Richmond, BC, phone 604-231-4910, fax 604-231-4900, email sbrunke@mdacorporation.com

Peter von Gaza, Remote Sensing Specialist, Pixelmapper Geospatial Consulting, Box 32011, Whitehorse, Yukon, Y1A 5P9, phone 867-668-4289, email peter@shadowlynx.com

Lauren Crooks, Spatial Data Administrator, Geomatics Yukon Information, Communication & Technology Branch, Government of Yukon, Box 2703, Whitehorse, Yukon Y1A 5T8, phone 867-393-7084, fax 867-667-5304, email:Lauren.Crooks@gov.yk.ca

The Yukon Government is in the first phase of an ambitious multi-year project to improve the existing Yukon 1:50,000 NTS and NTDB datasets.  This first phase of the project is to determine the methodology and best practices for producing 1:25,000 scale digital image and vector base map data for the entire Territory.  MDA GSI is a partner in the project, tasked to evaluate and make recommendations on potential satellite imagery solutions to be used as inputs to support the Yukon’s requirements for very large area mapping and for producing both image and vector products.

Some of the satellites being evaluated are SPOT, CARTOSAT, and IKONOS for their stereo capability as well as speculative evaluations of WorldView-1, GeoEye-1 and RADARSAT-2.  The final project objectives are to document methodologies and develop a model for the purposes of estimating future costs for vector base products, digital elevation models and a seamless image mosaic for the entire Yukon.

While mapping at relatively small scales of 1:25,000 is an accepted and routine endeavor in areas south of 60° latitude, these methodologies cannot often be directly extrapolated in areas to the north.  A multitude of factors pertaining to data availability, ground control collection, and digital elevation model creation, among others, have to be resolved in order to achieve this scale.  This paper describes methodologies evaluated, results achieved, difficulties encountered and potential solutions found that could be used for similar strategies for mapping in northern environments.

High Resolution Lake Edge Extraction from Colour Orthophotography

Barry Pierce and York Law

EBA Engineering Consultants Ltd., 1066 West Hastings Street, Vancouver, BC, phome 604-685.0275, fax 604-684-6241, email bpierce@eba.ca

Permitting and development in the Canadian Arctic is made more difficult due to the low availability of quality baseline information. An overland route of approximately 160km is currently being developed for the southern portion of the Tibbitt to Contwoyto winter road, which has suffered from early thaws in recent years. Engineering designs were produced with the assistance of high resolution Lidar data; however, hard breaklines were required at water's edge to produce high quality elevation models. These lake positions were previously unknown with sufficient accuracy.

Multipath scatter of the raw Lidar returns over water prevented easy water boundary extraction from the Lidar data itself. Fortunately, the Lidar data also came with a series of 453 colour orthophotos. Water in these images were of variable siltiness, brightness and shallowness. An Earth Resource Mapper (ERMapper) algorithm was developed to select the water/land interface by choosing thresholds based on three derived measures, roughly described as Greyness, Blueness, and Smoothness. Ratios of grayness and blueness to blueness and smoothness helped reveal transitions between water and land. Minimum smoothness criteria further helped reject strong edge transitions over land. Greyness was further used to remove surrounding black representing no-data.

Sections of the image sequence were captured as ERMapper mosaic algorithms. This allowed several images to be sequentially processed as one large virtual image. The image processing steps could in theory be run on one enormous virtual image mosaic containing all the individual scenes. It was experimentally found that vector conversion from ERMapper's .erv format to shapefiles seemed to be a memory-bound step, and limited the number of mosaic images in practice to about 20 at one time. Segmentation was performed on each mosaic section and then converted to vectors. The image segmentation cleaning was performed in ESRI’s Arc/Info environment on vector polylines converted to coverages. Segment refinement was performed by rejecting line segments less than 50m in length, corresponding to erroneously selected land and rough water areas. The resulting cleaned vectors were then manually reassembled to produce lake edges.

The algorithm performed well over a range of water surfaces. Thresholds conservative enough to returned unbroken water edges also falsely selected small terrain features, choppy water and seams of adjacent ortho tiles as water’s edge. The pre-cleaning steps made selection of the appropriate segments much easier. Occasionally, segments had to be drawn by hand where silt at water's edge confounded the segmentation process. Final elevation models were produced from Lidar with hard breaklines at the extracted water’s edge. These models were successfully employed for road engineering design and ecosystem and habitat mapping.

PROSPECT+SAILH canopy analysis using Python

Qingmou Li¹, Rasim Latifovic¹, Richard Fernandes¹, Baoxin Hu², Francis Canisius¹

¹Canada Center for Remote Sensing (CCRS), Natural Resources of Canada, Ottawa, Canada, emails qli@NRCan.gc.ca; rlatifovic@NRCan.gc.ca; rfernand@NRCan.gc.ca; fcanisius@NRCan.gc.ca

²Department of Earth and Space Science and Engineering, York University, Toronto, Canada, email baoxin@yorku.ca

Retrieval of biophysical parameters, such as leaf area index (LAI) and chlorophyll content (chl) using optical indices, physical models and their sensitivity analysis are widely considered in remote sensing and application field. However, still there is a difficulty in finding a flexible tool currently available to fulfill these analyses. This paper presents a scripting tool based on an integrated physical model, PROSPECT+SAILH, to carry out the analysis of the reflectance spectrum response on parameters, designing optical indices, and biophysical parameter inversion with Marquardt-Levenberg (ML) method.

The used PROSPECT+SAILH (ProSailH) radiative transfer model in this study has been validated by many lab/field/space experiments over a decade and widely used in agriculture, plant physiology, and forestry. The generic ProSailH forward model and (ML) inversion algorithm are implemented as C++ objects with carefully designed encapsulation and stubbed with a Python (PyProSailH) module. With this scheme, ProSailH based applications might be implemented very easily without any requirement of programming experiences.

With the developed PyProSailH module, spectral response of biophysical parameters, selected optical indices, ML inversion results of forward modulated spectra, and Python code examples are presented in this paper. As the major algorithms are in C++, the Python module shows similar performance as C++.

Key words: PyProSailH; PROSPECT; SAILH; vegetation index; Inversion; Python

Session 5 - LIDAR and Geology

Use of LIDAR for Characterising the July 24 2007 Rock and Ice Avalanche at Mount Steele, St. Elias Mountains, Yukon, Canada

Panya S. Lipovsky, Chris Hopkinson, Michael N. Demuth, Stephen G. Evans, John J. Clague

P. S. Lipovsky (proposed presenter), Yukon Geological Survey, 2099 2nd Avenue, Whitehorse, Yukon, Y1A 2C6, Canada, email Panya.Lipovsky@gov.yk.ca, phone 867-667-8520, fax 867-393-6232

C. Hopkinson, Applied Geomatics Research Group, Centre of Geographic Sciences, Nova Scotia Community College, 50 Elliott Rd, RR1, Lawrencetown, Nova Scotia, B0S 1M0, Canada

M. N. Demuth, Geological Survey of Canada (Glaciology), Natural Resources Canada, 601 Booth Street, Ottawa, Ontario, K1A 0E8, Canada

S. G. Evans, Landslide Research Programme, Department of Earth and Environmental Sciences, University of Waterloo, 200 University Avenue West, Waterloo, Ontario, N2L 3G1, Canada

J. J. Clague, Centre for Natural Hazard Research, Department of Earth Sciences, Simon Fraser University, 8888 University Drive, Burnaby, British Columbia, V5A 1S6, Canada

A catastrophic rock and ice avalanche occurred on the north face of Mount Steele, Yukon Territory, on July 24, 2007, depositing debris across Steele Glacier below. Mt. Steele is Canada's fifth highest mountain and is located in the extremely rugged and remote Icefield Ranges of the St. Elias Mountains. The avalanche was one of the largest recorded landslides onto a glacier in the Canadian Cordillera. On August 12, 2007, a high-resolution, airborne LiDAR survey was performed above Mount Steele and Steele Glacier by the Geological Survey of Canada in partnership with the Applied Geomatics Research Group (Centre of Geographic Sciences, Nova Scotia Community College). The aircraft flew at altitudes between 4000 and 6100 m asl using a pulse frequency of 33 kHz, an IR laser wavelength of 1064 nm, and a scan angle of ± 24 degrees (48 degree field of view). A digital elevation model derived from the LiDAR data and supplemented by field observations was used to characterise the morphology of the landslide source area and deposit. The LiDAR survey provided valuable information for modelling the landslide behaviour and allowing comparisons with similar mass movements that have occurred around the world. The results of this work have important implications for landslide risk management in more populated regions that are located near steep rock slopes in glacial environments. Our work highlights the value of airborne LiDAR surveys for quickly determining important landslide morphological characteristics in rugged areas with difficult and expensive access. It also demonstrates that airborne LiDAR surveys can be performed successfully at high altitudes.  Field studies are planned in 2008 to better constrain uncertainties in some of the data derived from the LiDAR survey.

Session 6 - Ground, Airborne and Satellite Imaging

The Alaska Volcano Observatory Monitoring System: A Possible Model for the Changing Polar Environment

Ken Dean¹, Jon Dehn¹, Peter Webley² and John Bailey²

¹Alaska Volcano Observatory, Geophysical Institute, University of Alaska Fairbanks
²Alaska Volcano Observatory, Arctic Regions Supercomputer, University of Alaska Fairbanks, Fairbanks, Alaska  99775-7320

Ken G. Dean (Presenter)  [Oral], Session Topic: Environmental Monitoring, phone (907) 474-7364, fax (907) 474-7290, email ken.dean@gi.alaska.edu

Climate change and its impacts on the environment are of increasing concern especially to people living and working in the Arctic. In Alaska melting permafrost, decreasing Arctic Ocean sea ice and increased storm activity has already threatened coastal villages, industrial infrastructure, and habitats.  One can expect that changes in fluvial erosion, transport and deposition will have further impacts on the Polar region. Satellite data has been used for years to study many of these processes but not always in a holistic program of systematic daily monitoring and analysis. In volcanology the Alaska Volcano Observatory (AVO) has developed a sophisticated “near real-time” monitoring system that could potentially act as a model for a system to monitor, analyze and/or catalog the effects of climate change on the Arctic environment. The purpose of this presentation is to describe the AVO system in terms of its data collection, data processing, information cataloging, reporting of observations to emergency response agencies and to introduce the analysis tools used to assess volcanic activity.  Data with high temporal resolution, such as GOES, MTSAT, AVHRR and MODIS, are the primary sources for hourly assessments but high spatial resolution ASTER and SAR data are incorporated into the analysis to identify specific processes and to assess impacts. Automated processing has been developed to optimize the detection and analysis of changes in surface temperatures and plume detection related to volcanic processes. Observations are entered into a database that can be easily accessed and used for more holistic analyses.  Finally, a sophisticated warning system has been developed to alert emergency response groups and directly affected parties.  The AVO system is designed to have a response time measured in minutes but this would not necessarily be required for climate change studies.  AVO is primarily focused on relative change using current real-time data but climate change would be assessed by examining new data and comparing it to older historical data. The AVO monitoring database system and the automated processing and analysis tools should be applicable for these studies. For most climatic change studies, higher spatial resolution will probably take precedence over high temporal resolution and so new derived products will need to be generated.

Real time MERIS image geo-projection and multifractal interpolation with C++

Qingmou Li¹, Rasim Latifovic¹, Richard Fernandes¹, Shahid Khurshid¹

(3)Canada Center for Remote Sensing (CCRS), Natural Resources of Canada, Ottawa, Canada, qli@NRCan.gc.ca, rlatifovic@NRCan.gc.ca, rfernand@NRCan.gc.ca, Shahid.Khurshid@NRCan.gc.ca

The MEdium Resolution Image Sensor (MERIS) on the ENVISAT of Europe Space Agency (ESA) has 15 narrow bands at visible and very near infrared wavelength with 260 and 1000m resolution, making it has specific advantages for space-borne based land cover classification, water color characterization, forest fuel type identification, agriculture investigation or other purposes at national or global scales. The starting point of these applications is the generation of geoprojected and cloudy free composite GIS-ready products. However, the huge volume of MERIS orbit dataset, characterized from 600 to more than 1300 mega bytes, challenges the currently available geoprojection and interpolation algorithms.

This paper presents a real time geo-projection and multifractal interpolation algorithms developed to generate reflectance at the top of atmosphere using MERIS L1B radiance data sets.

In the developed procedure, MERIS orbit is modeled at first with the Latitudes/ Longitude tie information after ground control point (GCP) correction with automatic adaptable polynomial equations tile by tile. The orbit is divided into a series of tiles to optimize the modeling and balance the memory requirement and performance. With build-up orbit models, the points and its vicinity in the L1B tie point space which is required for interpolation for unknown points in the geoprojected space points are positioned directly, not by any kind of searching algorithms.  The orbit modeling gives pixel accuracy less than 0.5 pixels with real time performance.

The nearest neighbor, inverse distance weighting (IDW), SinX/X weighting, bilinear, and multifractal interpolation methods are implemented. It is found that the multifractal interpolation has advantages compared to other methods, such as less notched edges and reservation of subtle changes.

MERIS reduced resolution radiance over Canada of 2003 is calibrated with sensor information to reflectance and then geoprojected into the national standard LCC space of Canada. The projected images are composited for cloudy free images. The composited image has accuracy less than 1 pixel.

Further work, such as atmospheric correction, Leaf Area Index and FPAR retrieval are under development.

The processor is developed with objected oriented programming concept with C++. Real time images of different format are generated with Template classes (ATL) in real time.

Key words: MERIS, Orbit modeling, Geoprojection, Composite, multifractal interpolation, real time

Comparison of Temporal Filtering Methods for Error Minimization of 250m MODIS Spectral Time Series

Dr. Darren Pouliot, Dr. Richard Fernandes, Dr. Rasim Latifovic, Dr. Ian Olthof

588 Booth Street, Ottawa, ON K1A 0Y7

Proposed presenter: Darren Pouliot, email: Darren.Pouliot@ccrs.nrcan.gc.ca, phone (613) 947-1267, fax (613) 947-1406

Spectral time series acquired by satellite sensors have high variability in measured reflectance making robust feature extraction a relatively difficult task. Major sources of variability include atmospheric conditions, clouds, shadows, unfavourable illumination conditions, sensor noise, viewing geometry, and spatial misregistration.  Therefore, a requirement for feature extraction\identification is noise reduction.  This is usually accomplished by filtering in order to clearly reveal the characteristic properties of the time series such as snow\ice free, vegetation green-up, vegetation senescence, or snow\ice cover. In this research four local time series filtering approaches were compared. Of these rank based filtering was found to be optimal maintaining critical points (e.g. initiation of snow melt, snow/no-snow) and minimizing temporal and spatial noise.

StereoSATs: a satellite-based approach that optimizes mapping processes

Author(s): Mathieu Benoit, Director – Earth Observation group, VIASAT GeoTechnologies Inc.

Proposed presenter: Mathieu Benoit

Session Topic: “Other:”

Phone, 514-495-6500 #106, fax: 514-495-4191, email: mbenoit@viasat-geo.com

Land management relies on complete up-to-date knowledge of the available geographic information. In remote areas or developing countries, where technical knowledge and expertise are usually more limited, such information is not accurate enough and is often incomplete or out of date. Although aerial photography was traditionally used as a source of information for producing these maps, the high costs and technical difficulties of acquisition encountered in many areas of the world have shifted the focus more towards technologies that make use of satellite imaging. Currently, more than 30 commercial satellites scan and analyze Earth’s surface every day. These satellites capture images that are more and more like aerial photos, and many of them have also the ability to capture stereoscopic images (IKONOS, SPOT, etc.).

StereoSATs^TM is a solution using satellite imagery, independently if imagery was acquired in stereoscopic mode or not, to provide a 3D representation of the countryside. Depending on the spatial resolution of the images and the data extraction technologies used, information can be produced in order to meet topographic and thematic mapping needs for a wide range of scales. By enabling an analysis of the territory that integrates spectral values, texture and relief, StereoSATs^TM is providing the various land specialists with several advantages compared to traditional remote sensing data.

The StereoSAT^TM approach was developed by VIASAT GeoTechnologies on applied research projects conducted in Northern Quebec and Africa in collaboration with The Canadian Space Agency. It has also been used in operational mode to produce topographic data for Quebec’s Ministry of Natural Resources, and Wildlife (MRNF) and is currently deployed in Peru on the “Vitrine StéréoSat Pérou” project being conducted in the GéoQuébec group. The goal of this project is to demonstrate the technical, economic and operational benefits of the Earth observation StereoSAT^TM for producing topographic, thematic and cadastral data intended for land-based knowledge in external markets.

Session 7 - Radar

RADARSAT-2: SOAR Program and Benefits for the Canadian Government

Daniel De Lisle and Jill Smyth, Canadian Space Agency, 6767 route de l'Aéroport, St-Hubert, Quebec, Canada

RADARSAT-2 was successfully launched in December 2007. RADARSAT-2 is the most versatile commercial SAR Earth observation system with a high spatial resolution mode of 3 meters and a fully polarimetric mode that will be available on an operational basis. The advancement in satellite technology do not only profit the imaging modes, other innovations on RADARSAT-2, such as high downlink capacity, onboard digital recorders, and high performance processing will increase the quality of the service and products.

RADARSAT-2 is the product of public-private partnership. The private sector owns and operates the spacecraft, and in return for their investment, the Government of Canada has obtained a credit for data to be drawn against over the lifetime of the spacecraft. RADARSAT-2 data will be used by the Government of Canada to meet the current and evolving priorities of the Government and the needs of Canadians.

Therefore, beyond the commercial benefits, the Government has a vital interest in the public good aspects of RADARSAT-2. The CSA has committed to objectives that pertain to natural resource management, environmental monitoring, ice mapping and marine monitoring. RADARSAT-2, together with RADARSAT-1 help assure Canada’s safe navigation in icy waters, patrol of coastal waters, support for pollution and fisheries interdiction, and, sustainable development of the far north. RADARSAT-2 will be an important space asset, providing Government departments with the continuity of critical and timely data for the active management of natural resources and monitoring of the environment.

Among the Applications Development Support Programs, the Science and Operational Applications Research for RADARSAT-2 (SOAR) was put in place. The main objective of the SOAR program is to provide data to explore operationally and commercially viable solutions to current problems and issues. SOAR provides an opportunity to explore the enhanced capabilities of  RADARSAT-2 and their potential contributions to applications, operational requirements, and business opportunities.

This paper will briefly present the capabilities of RADARSAT-2, and will focus on the Scientific and Operational Applications Research (SOAR) Program and activities that the Canadian Government is undertaking to optimize the use of RADARSAT-2 in support of its mandate and priorities.

Session 8 - Agriculture/Forestry/Wetlands

Mountain Pine Beetle Structural Change and Damage Assessment Using MFM Canopy Reflectance Inversion Modeling

Derek R. Peddle*, Sarah Boon, and Aaron Glover

Department of Geography, University of Lethbridge, 4401 University Drive West, Lethbridge, AB. T1K 3M4 Canada, *phone 403-329-2524; fax 403-329-2016, email derek.peddle@uleth.ca

The recent mountain pine beetle (MPB; Dendroctonus ponderosae) outbreak in British Columbia (BC) began in 1999 and rapidly reached epidemic proportions. As of 2007, over 13 million hectares of BC Crown forest were damaged, and the beetle infestation had spread to Alberta. MPB significantly impacts forest canopy structure, as infestation and death of lodgepole pine (Pinus contorta) results in needle and branch loss, and subsequent blowdown. Forest canopy changes have secondary impacts on local meteorology, hydrological processes, understory regeneration, and wildlife habitat. Given the extensive area affected by the MPB epidemic, the regional and repeat coverage of satellite imagery plays an important role in comprehensive monitoring and map updates that are otherwise labour intensive and costly over large areas. Of particular importance is the ability to retrieve biophysical structural information (BSI) that is essential for regional scale estimates of structural change, yield status, salvage operations and damage assessment. Canopy reflectance (CR) models provide a powerful, physically-based approach to deriving forest BSI. The Multiple-Forward-Mode (MFM) approach has been developed and validated over a wide variety of ecosystems including extensive analyses in boreal and montane terrain using different CR models, airborne and satellite sensors and shown to be superior to conventional BSI estimation approaches. MFM provides a robust inversion model capability to derive forest structural information for a variety of CR models, including more complex models that are not invertible by conventional means. MFM modeling was applied to two 2500 m² study plots - alive and beetle-killed (‘grey attack’) - in the dry-cool sub-boreal spruce zone (SBSdk), 50 km southwest of Fraser Lake, BC (53.72ºN, 124.92ºW). Measurements of stand age/decay class, species mix, and canopy structure were collected in 2007 in each forested plot. Data were used to validate model output derived from Landsat imagery acquired during the 1999-2007 period. Good correspondence was found between MFM forest structure outputs and field measured stand density, horizontal and vertical crown radius, and tree height and distribution for the sites tested, which represent diverse MPB stand conditions. This work has demonstrated a potential for deriving forest structural information over larger areas, with future work to involve increased sample sizes and larger area validation, as well as hydrological applications such as assessing altered canopy interception regimes as a function of mountain pine beetle damage.

A Medium-Resolution Remote Sensing Classification of Agricultural Areas in Grizzly Bear Habitat

Adam Collingwood, Steven E. Franklin, Xulin Guo, and Gordon Stenhouse

Proposed presenter: Adam Collingwood, M.Sc. candidate, University of Saskatchewan, Environmental Remote Sensing Laboratory, Kirk Hall Rm. 2, 117 Science Place, Saskatoon, SK S7N 5C8, phone 306-880-2107 or 306-966-1488, email Adam.Collingwood@usask.ca

Habitat loss and human-caused mortality are the most serious threats facing grizzly bear (Ursus arctos L.) populations in Alberta, with conflicts between people and bears in agricultural areas being especially important.  To help manage and mitigate these effects, current habitat maps are needed.  The objectives of this research were to find the best possible classification approach from a limited selection of methods for determining multiple classes of agricultural and herbaceous land cover, and to create land cover maps of agricultural and herbaceous areas which will be integrated into existing grizzly bear habitat maps for western Alberta.  Spectral and environmental data for five different land-cover types of interest were acquired in late July, 2007, from Landsat TM satellite imagery and field data collection in two study areas in Alberta.  Three different classification methods, one unsupervised and two supervised methods, were analyzed with these data to determine the most accurate and useful method.  The best method was the Supervised Sequential Masking (SSM) technique, which gave an overall accuracy of 88% and a Kappa Index of Agreement (KIA) of 83%.  Three of the 5 classes had an average KIA of greater than 95%, with the other two classes being above 72%.  This classification was then expanded to cover 6 more Landsat scenes, and when combined with bear GPS location data, it was discovered that bears in agricultural areas were found in grasses / forage crops 77% of the time, with small grains and bare soil / fallow fields making up the rest of the visited land-cover.  The bears were found in these areas primarily in the summer months.

The results of this research will allow for the creation of a more accurate and detailed land cover map covering areas of grizzly bear habitat.  This map could contribute to more accurate resource selection models and would give a better understanding of bear activity in agricultural areas.  The increased thematic accuracy of this map compared to current maps could also contribute to more robust calculation of landscape metrics in agricultural areas.

The Canadian Wetland Inventory Phase I

Brian Kazmerik, National GIS Manager, Ducks Unlimited Canada, P.O. Box 1160 Stonewall, Manitoba, Canada, R0C 2Z0, phone (204) 467-3247, fax (204) 467-9028, email b_kazmerik@ducks.ca

Robert Hélie, National Coordinator, Canadian Wildlife Service, Environment Canada, Place Vincent Massey – Floor:03, 351 St Joseph Boulevard, Gatineau, Quebec K1A 0H3, phone (819) 953-7935, fax (819) 994-4445, email Robert.Helie@ec.gc.ca

Canada is estimated to have between 23 to 28 percent of the world’s wetlands, yet it currently has no national inventory for wetlands.  In 2002 the initial phase of the Canadian Wetland Inventory was launched with a partnership approach.  The objective of the CWI is to provide a national wetland inventory that can be used for the conservation and sustainable management of wetlands for environmental and societal benefits.  Phase I of the CWI has recently been completed.  This consisted of inventory methodology development on study sites which represent the diversity of wetlands across Canada.  Partnerships and support was cultivated by working towards national standards for wetland classification, scale, and level of accuracy, while respecting the needs of regional earth cover initiatives.  The CWI Project Team recommends that the majority of Canada (93 percent) is mapped at medium resolution using satellite-based methods refined to Canada’s regional variability and joined within a common framework of standards.  A small portion of Canada (7 percent) is mapped at high resolution using a different method, but with a common framework hierarchy.  The end product is consistent with the needs of a large number of partners (including federal departments, provinces, territories, Aboriginal communities, and municipalities), and communities of interest.  A business case to complete the inventory has been drafted and is presently under consideration by the lead agency, Environment Canada.  Other national partners include the Canadian Space Agency, Ducks Unlimited Canada, Natural Resources Canada, Agriculture and Agri-Food Canada, and the North American Wetlands Conservation Council (Canada).  Several provinces and territories and two academic institutions have also participated in Phase I of the CWI.  This presentation will focus on the accomplishments of the first phase of the CWI, and the recommendations to complete a full national wetland inventory.

Session 9 - Snow and Ice, Glaciers

Development of passive microwave snow water equivalent retrievals for tundra environments

Peter Toose¹, Chris Derksen², Anne Walker³

Climate Research Division, Environment Canada, 4905 Dufferin St., Toronto, Ontario, M3H 5T4

Peter.Toose@ec.gc.ca¹, Chris.Derksen@ec.gc.ca², Anne.Walker@ec.gc.ca³

The tundra landscape encompasses a large proportion of the terrestrial environment in northern latitudes, and yet systematic datasets of the winter season distribution and magnitude of snow water equivalent (SWE) within this landscape are generally unavailable. Only a very sparse conventional snow monitoring network exists across this region, therefore satellite derived estimates may provide a potential solution to this data inadequacy. In recent years, Environment Canada has conducted intensive field campaigns in the central Northwest Territories (2004-2007) and northern Manitoba (2005-2006) to acquire snow cover measurement data sets in support of the development and evaluation of tundra specific SWE retrieval algorithms for satellite passive microwave data.

Multi-frequency (6.9, 19, 37 GHz) and multi-scale (1 m to 25 km) passive microwave measurements were acquired from ground, airborne, and satellite radiometers along with coincident in situ snow surveys. Analysis of these datasets has shown that satellite passive microwave SWE estimates using existing algorithms consistently underestimate tundra SWE due to the combined effects of a heterogeneous snow distribution controlled by terrain and the high percentage of lake cover typical to the tundra environment. The very large snow amounts deposited in spatially constrained drift features has little influence on the microwave emission measured by coarse resolution radiometer sensors. Lake cover fraction has a significant influence on measured brightness temperatures due to the presence of liquid water below the ice. The influence of lake cover, however, is frequency dependent and evolves seasonally as the ice thickness increases and is largely removed if the ice completely freezes to the lake bottom (a common occurrence in some tundra regions). If drift storage and lake fraction effects are properly accounted for, microwave brightness temperatures do exhibit sensitivity to changes in tundra SWE.

This paper will provide an overview of the progress being made in the development of satellite derived SWE information for the tundra region of northern Canada.  These results are based on datasets from Canadian shield tundra (central Northwest Territories) and Hudson Bay lowland tundra (northern Manitoba).  Environment Canada will conduct additional airborne passive microwave surveys with coincident in situ snow and lake ice measurements across the Canadian shield tundra of northern Quebec (February/March 2008) and in the upland tundra near Inuvik in the Northwest Territories (April 2008) as part of the International Polar Year project “Variability and Change in the Canadian Cryosphere”. Analysis of these datasets will help refine quantitative relationships between tundra SWE, lake cover fraction, and multi-frequency microwave brightness temperatures, and determine the applicability of new algorithm development across the various Canadian tundra sub-regions.

Formosat, automatic cameras and GPS to survey a polar glacier (Austre Lovénbre, 79°N, Svalbard: first results of the IPY field trips

Madeleine Griselin¹, Dominique Laffly³, Christelle Marlin², Jean-Michel Friedt⁴, Eric Bernard¹, Gilles Martin⁴

¹University of Franche-Comté, CNRS UMR ThéMA, Besançon, F
²Univesity of Pau, UMR SET, Pau, F
³University Paris-Sud, UMR IDES, Orsay, F
⁴CNRS UMR FEMTO-ST, University of Franche-Comté, Besançon, F

Since the sixties, French scientists have followed the hydrology of the East Loven glacier basin (Austre Lovénbre), which is considered as a « school-site », equipped with hydrological stations by Spanish scientists since some years.

Forty years of hydrological observations (even discontinuous in time and in space) give an historical view on the rapid reactions of an environment in constant evolution like in the polar regions. The hydrological response of this glacio-hydrosystem to climatic changes, even if they are minor, is very short. The observation of this response as well as its monitoring is necessary to understand, to quantify, to qualify, to spatialize the flows (liquid and solid) and their dynamics (in space and time), and to estimate the temporal variability of the different components of the runoff due to the very recent climatic changes (40 years). Compared to the 40 years of continuous climatic records at Ny Alesund station, the hydrological approach of the Loven East Glacier Basin will provide a better understanding of the importance of water in the global dynamics of the polar margins of this area, where an accelerated ice retreat is clearly observed since the 70s.

Beyond the hydrology and climatology experiments performed in Spitsbergen, the group involved in this project is also expert in geomatic (DEM) and satellite data processing. The crossing of informations « from above » and « from inside » is necessary for the spatialisation of the climatic data and their dynamics. Through the Hydro-Sensor-FLOWS program (IPY # 16), a continuous survey is carried out for 2006 to 2010.

This program lays on the survey during 4 years (minimum) of an arctic hydro-system (Austre Lovénbre) through the crossing of informations coming from image loggers (Formosat and digital cameras on the ground), information coming from physical loggers recording hydro-climatic data, and from samplings in situ: this sensor system has been presented during the 9th Circumpolar Remote Sensing Symposium.

The program began in September 2006: four field trips have been already conducted, allowing us to set up the sensor network and to obtain first results. After one year of running, we will present the first results obtained during the 2007 field trips. We will focus on the automatic digital camera system (3 photos per day on 8 stations since April 2007).

We will show the limits of the instruments and of the data processing. We will show the first real results crossing the air temperature data (30 loggers) with the photos of the digital cameras which complete the information obtained through the 11 Formosat images received during that time (April-September 2007).

New geopositioning measurements compared to DEM data of 1995 give a very accurate variation of the glacier elevation and of its volume. The front retreat has been determined through aerial pictures and in situ GPS measurements.

Landsat Image Mosaic of Antarctica (LIMA) – A Large Scale Data Solution for Science

Susan Parks, Senior Technical Engineer, ITT Visual Information Solutions, 4990 Pearl East Circle, Boulder, CO, USA 8030480301, Phone 303-413-3970, fax 303-786-9909, email: sparks@ittvis.com

The Landsat Image Mosaic of Antarctica (LIMA) provides a base map of Antarctica in support of activities surrounding the International Polar Year (IPY 2007-2008). The LIMA mosaic provides the most digitally accurate, true color, virtually cloudless, seamless and high resolution view of the Antarctica earth.

This abstract proposes a talk about the LIMA development process as an example of a large scale, distributed development effort that was essentially a collaboration of science and software.  The collaborators included the U.S. Geological Survey (USGS), the British Antarctic Survey (BAS), the National Aeronautics and Space Administration (NASA), and ITT -Visual Information Solutions.

The development of the LIMA mosaic presented a number of unique image processing challenges.  Many of the problems encountered during the LIMA creation process are common to researchers working with remotely sensed data in the polar regions.  For example, a mosaic of the polar regions is traditionally difficult in terms of color-balancing and calibrating the data.  Large variations in sun angle for each Landsat scene created the need for custom built software to calculate a sun elevation for each pixel.  Another issue experienced during the LIMA creation process was the shear volume of data involved; the mosaic was created from more than 1,1000 Landsat ETM+ scenes.

This comprehensive base map will benefit polar research being undertaken around the world. In addition, lessons learned during the creation of this mosaic will lead the way for similar efforts in the future.

Initial Assessment of RADARSAT-2 for Sea Ice Monitoring

Matt Arkett, Dean Flett, Roger De Abreu and Marie-France Gauthier

Canadian Ice Service, Meteorological Service of Canada, Environment Canada, Ottawa, Ontario K1A 0H3, phone (613) 947-7514,  fax (613) 996-4218, email matt.arkett@ec.gc.ca

The Canadian Ice Service (CIS) promotes safe and efficient maritime operations and protects Canada’s environment by providing reliable and timely information about ice and iceberg conditions in Canadian waters.  The CIS relies on a suite of both airborne and satellite sensors to operationally monitor ice conditions in Canadian coastal and inland waterways.  Satellite SAR, mainly from RADARSAT-1 and Envisat ASAR, are the primary satellite datasets for coastal monitoring.

RADARSAT-1’s successor, RADARSAT-2 was successfully launched on December 14^th, 2007.  In the winter and spring of 2008, the CIS will be performing an evaluation of the performance of RADARSAT-2 in support of its ice operations.  In this work, we will provide a preliminary assessment of the use of the new SAR sensor for monitoring sea ice conditions.  We will compare its performance against RADARSAT-1 while also evaluating the utility of the new advanced SAR modes (eg. polarization).  The results from this comparison and evaluation will assist CIS in determining preliminary recommendations on mode selection for ice monitoring in the future.

Session 10 - International Polar Year

A GeoNetwork discovery portal for North Yukon IPY Data Collections

Jeff Hamm, Principal, Geoplan Consulting, 23 Balsam Cr, Whitehorse, Yukon Y1A 4V6

Jennifer Lee, Lands Manager, Vuntut Gwitchin Government, Box, 94, Old Crow, Yukon Territory, Y0B 1NO

Presented By: Jeff Hamm, phone 867.667.7397, fax 867.667.4624, email jeff@geoprism.ca

An IPY project (Yeendoo Nanh Nakhweenjit K'atr'ahanahtyaa - Environmental Change and Traditional Use of the Old Crow Flats in Northern Canada. IPY Initiative #292) undertaken jointly with the Vuntut Gwitchin First nation (VGFN), Vuntut Gwitchin Government (VGG) and a multi-disciplinary team of scientists is currently conducting scientific and traditional knowledge data collection and analysis in the Old Crow Flats area of Northern Canada. Themes being investigated include (Paleo)Hydrology, Food Security, Paleoclimate, Permafrost, Quaternary History, Vegetation and Wildlife. A GeoNetwork web portal developed using to meet two project objectives: (1)To meet IPY Data Policy requirements for discoverable, FGDC compliant metadata records; (2) To provide on-going access to project data collections for long-term environmental monitoring by VGFN. The portal allows project investigators to easily create FGDC compliant metadata records and share geographically referenced thematic information between different projects. Metadata can be harvested from existing collections using standards based web  services, including WMS/WFS, GeoNetwork nodes, CWS services or WebDav can be automatically harvested.

The Polar Metadata Catalogue as a Resource for Canadian IPY Scientists

E.F. LeDrew¹, P. Yoon², C. Barnard³, W. Vincent⁴, S. Latour⁵, S. Tomlinson⁶

¹Department of Geography, University of Waterloo, Waterloo, ON, Canada
²Canadian Cryosphere Information Network, University of Waterloo, Waterloo, ON Canada
³ArcticNet Inc., Universite Laval, Quebec, Canada
⁴Centre d Etudes Nordiques et Departement de biologie, Universite Laval, Quebec, Canada
⁵NRCAN, Government of Canada, Ottawa, ON, Canada
⁶Federal IPY Secretariat, Gatineau, Quebec, Canada, email ells@watleo.uwaterloo.ca

The International IPY Data Management Committee is working towards an IPY Master Directory that will provide a discovery portal for metadata using standard protocols.

In Canada, we have been developing a Metadata discovery portal for the ArcticNet program which is currently operational. This program includes scientists working in the natural, social and health sciences and, as such, addresses a wide range of issues associated with Metadata cataloguing within a multidisciplinary program.

This Metadata portal has been modified to address many of the needs of the Canadian IPY scientists. In collaboration with other national programs the system is evolving towards a ‘Polar Metadata Catalogue’ that will include flexibility to address the needs of scientists in a wide range of programs dealing with Polar science issues.  In doing this, we will maintain interoperability with other Metadata catalogues to ensure a wide-ranging discovery experience for the scientists, no matter what the discipline, provide long term support of the information beyond the IPY sunset years, and provide information and tools for ongoing outreach and communications.

In this paper we will describe the evolution of this portal and potential new directions, as well as provide a hands-on demonstration of the existing portal.

Session 11 - Land/Water/Wildlife

Multi-Species Habitat Modelling for Land Use Planning

Sam Skinner, Land Use Planner, Peel Watershed Planning Commission, 307 Jarvis St, Suite 201, Whitehorse, YT, Y1A 3P4, Canada, phone 1-867-667-2374, fax 1-867-667-4624, email sam@planyukon.ca

Describing and mapping several conservation values is an important and challenging step in the development of a regional land-use plan in the data-poor, remote and largely uninhabited Peel watershed in north east Yukon, Canada.  Existing polygonal traditional use and wildlife key area data does not uniformly cover the entire planning region, so other blanket habitat maps were required for uniform coverage.  An enhanced biophysical base map was first developed using existing satellite-derived vegetation mapping (Earth Observation for Sustainable Development of Forests – EOSD) , DEM, topographic (NTDB), and polygonal soil landscape data.  Next, data on perceived habitat suitability of the resulting biophysical units for four herds of caribou, moose, sheep, marten, breeding birds and grizzly bears were collected at workshops with relevant biologists and people from the three adjacent communities.  These habitat suitability indices were then applied to the biophysical base map to generate preliminary habitat maps.  These maps were modified to  reflect the influence on snow depths, recent fire history, wind fetch, and local environmental conditions on habitat quality, to produce final habitat maps.  Concurrently, habitat models of waterbirds, Peregrine Falcons and fish were created using topographic data and Landsat 7 image interpretation.  All these maps were first integrated together to generate composite conservation values, then used with conservation area planning software (MARXAN and Zonation) to generate a suite of conservation network plans.  The Peel Watershed Planning Commission will then weigh these plans against other values and interests in the planning region to create a comprehensive land-use plan for the region.

Seasonal and long term changes in forage availability to the Bathurst caribou herd detected using remote sensing time series and field measurements

Wenjun Chen^1*, Junhua Li¹, Klaus Keohler², Yu Zhang¹, Weirong Chen¹, Brad Griffith³, Bruno Croft⁴, and Don Russell⁵

¹Canada Centre for Remote Sensing, Natural Resources Canada
²Canadian Food Inspection Agency
³United States Geological Survey and University of Alaska
⁴Government of Northwest Territories and Bathurst Co-management Board
⁵Yukon College and CircumArctic Rangifer Monitoring and Assessment network (CARMA)

Presenter: Wenjun Chen, Canada Centre for Remote Sensing, Natural Resources Canada, 588 Booth Street, Ottawa, ON, Canada, K1A 0Y7, phone (613) 947-1286, fax (613) 947-1383, email wenjun.chen@nrcan.gc.ca

The population of the Bathurst caribou herd has being decreased by over 70% during the last decades. Many aboriginal communities, regional caribou co-management boards, and the Government of Northwest Territories have expressed strong concern about the decline. Changes in forage availability, predators, diseases/parasites, harvest, cyclical caribou population dynamics, climate trends and weather events, and industrial developments have all been suggested to be potential factors contributing to the decline. To fully understand the causes of the decline, and probably more importantly to better manage the decline and speed up its recovery, one needs first to collect information about these potential factors. The overall goal of this study is to derive seasonal and long term changes in foliage biomass to the Bathurst caribou herd from spring to fall using remote sensing time series and field measurements. Specifically, we establish and test the relationships between site measurements of foliage biomass in the middle of growing season and 30-m spatial resolution vegetation indices derived from Landsat imagery, using the field measurements of vegetation conditions collected in 1999 and 2005. Landsat-based 30-m resolution map of foliage biomass in the middle of growing season will be produced for the Bathurst caribou habitat. From the 30-m map, we will then up scale the relationships to 1 km between foliage biomass in the middle of growing season and 1-km spatial resolution vegetation indices derived from AVHRR imagery. Seasonal and long term changes in foliage biomass for the Bathurst caribou habitat from 1985 to 2005 will then derived using the AVHRR time series. Because the impacts of forage availability changes on caribou individual and population growth are likely to be location specific (e.g., calving ground, summer range, and winter range) and time specific (e.g., calving period, early summer, late summer, and fall), we will investigate and present forage availability changes accordingly.

Ecosystem Classification of Regional Planning Areas in Northern Yukon

Marcus Waterreus, Habitat/Remote Sensing Technician, Habitat Management, Fish & Wildlife Management Branch, Government of Yukon, Box 2703, Whitehorse ,Yukon Y1A 2C6, phone 867-667-3739, email mbwater@gov.yk.ca

John Meikle, Habitat Protection Coordinator, Habitat Management, Fish & Wildlife Management Branch, Government of Yukon, Box 2703, Whitehorse ,Yukon Y1A 2C6, phone 867-667-3538, email john.meikle@gov.yk.ca

Shawn Francis, Biophysical Coordinator, Department of Environment, Government of Yukon, Box 2703, Whitehorse, Yukon Y1A 2C6

The Commissions from two adjacent Regional Land Use Plans in northern Yukon (i.e. North Yukon and Peel Watershed Regional Land Use Plans) required an ecosystem map to help achieve their mandate of producing a land use plan that develops a vision for the area, and to make spatially explicit land use recommendations.  The ecosystem classifications helped provide the framework for describing the diversity of landscapes within the planning areas, the base for wildlife habitat interpretation, and a means to help identify conservation values useful for land use planning.

Project scale and cost considerations led to the choice of a Predictive Ecosystem Mapping (PEM) approach.  This evolving method to ecosystem mapping involved bringing spatial biotic and abiotic data to bear on a set of pre-determined ecosystem classes through computer-based models.  Available and derivable data, for land cover, landscape position, and soil moisture were used in the model.

The Yukon portion of Canada’s Earth Observation for Sustainable Development of Forests (EOSD) circa 2002 was the chosen land cover product, given its near complete coverage of the planning areas.  It contains twenty classes of vegetation and non-vegetated types, such as Open Conifer, Tall Shrub, and Rock/Rubble.  These classes are interpreted from suitable, snow-free Landsat 7 imagery at 25 metre resolution.  Landscape position, consisting of 5 primary classes, was derived from bioterrain interpretations for northern Yukon.  In order to accommodate significant climatic and physiographic variability within the planning areas it was necessary to modify the use of bioterrain in creating masks used in the model.  Soil moisture classes (i.e. dry, moist, and wet) were predicted through a set of topographic curvature classes calculated from a digital elevation model (DEM).

Modeling these inputs for the Peel Watershed Land Use Plan resulted in the description and mapping of 31 Ecosystem Classes at 25 metre resolution, with 7 High Elevation, 14 Low to Medium Elevation, 5 Riparian, 3 Wetland, and 2 Open Water classes.  While regional concentrations of Ecosystem classes vary, 75% of the planning region consists of 3 High Elevation classes: Rock/Exposed (20.1%), Dryas/Dwarf Shrub (9.9%), and Sub-alpine shrub (7.9%); and 4 Low to Mid Elevation classes: Wet Shrub (11.0%), Dry Shrub (10.4%), Dry Coniferous Forest (8.1%), and Wet Coniferous Forest (7.0%).  The remaining 25% of the Peel Watershed is covered by 24 ecosystem classes.

A Ground-Based Classification Scheme for Interpreting Satellite Derived Urban Vegetation Characteristics

Thoreau Rory Tooke, MSc Candidate, University of British Columbia, Department of Forest Resources Management, 2424 Main Mall, Vancouver, BC, V6T 1Z4

Nicholas C. Coops, Associate Professor, University of British Columbia, Department of Forest Resources Management, 2424 Main Mall, Vancouver, BC, V6T 1Z4

James A. Voogt, Associate Professor, University of Western Ontario, Department of Geography, 1151 Richmond Street, London, ON, N6A 5C2

Proposed presenter: Thoreau Rory Tooke, phone (604)822-4148, fax (604)822-8645, email pbcountry@hotmail.com

As our understanding of urban systems evolves, researchers are becoming increasingly aware of the importance of urban vegetation cover to physical processes such as micrometeorology and hydrology.  Recent research reveals that detailed vegetation characteristics, such as the structure of plant canopies and their health exert a strong influence on urban wind flow and rates of transpiration in the city.  As a result, high resolution remote sensing technologies can enable a new generation of urban micrometeorological forecasting by extracting detailed vegetation parameters of urban environments.

High resolution satellite sensors such as Quickbird and IKONOS provide detailed imagery of urban land cover across the Earth at regular intervals and at less cost than more traditional aerially mounted sensors.  One of the most common approaches to classify urban land cover from satellite imagery is by deriving sub-pixel estimates of basic urban features by spectral mixture analysis (SMA).  While these techniques provide repeatable results across cities, little work has been developed to produce field-based classification programs to validate the results.  In this study we present a new ground-based classification system for application to the detection of vegetation characteristics across urban areas.  Vegetation characteristics are carefully selected to provide critical elements important to the research fields of urban ecology, urban meteorology, and hydrology.

An application of our ground classification scheme is provided to exemplify how this research can be used to validate Quickbird derived SMA tree and grass endmembers over the City of Vancouver, BC, and to explain the detailed vegetation characteristics that influence the spectral response of urban vegetation.  Results indicate that the most accurate representation of the satellite derived tree endmember is by coniferous evergreen (r = 0.925, p<0.01) and deciduous broadleaf (r = 0.817, p<0.01) trees and the grass endmember by manicured (r = 0.663, p<0.01) grasses.  This research demonstrates the importance of collecting detailed vegetation characteristics across cities in order to more accurately assess remotely sensed land cover classes.

Poster Presentations

Examining the Relationships between Thermokarst and Headwater Drainage Networks using Remote Sensing in the Upper Kuparuk Basin

Erin D. Trochim Graduate Student, ftedt@uaf.edu, University of Alaska Fairbanks, P.O. Box 752342, Fairbanks, AK 99775, phone (907) 474-7975, fax: (907) 474-7979

Douglas L. Kane, Director of WERC and Prof. of Water Resources and Civil Engineering, University of Alaska Fairbanks

Anupma Prakash, Associate Professor at Geophysical Institute, University of Alaska Fairbanks

Surface runoff fluxe in the Arctic fresh water hydrological cycle can be modified by the feedbacks between continuous permafrost and variations in soil moisture, surficial runoff and channel routing. Predicting and characterizing potential hydrological response is an important component for engineering infrastructure appropriate for the climatic conditions. The Upper Kuparuk and Imnavait basins north of the Brooks Range in Alaska are part of a long-term monitoring effort, and provide an opportunity to pair hydrological studies and high-resolution topography models with remotely sensed data, to create a quality spatial distributed perspective. Previous research hypothesized that immature drainage basins in the foothills of the Northern Brooks Range could be substantially altered by warming climatic conditions, through deepening of the active layer and thermokarst development. The primary features most likely to be affected are water tracks, areas of preferential flow which comprise the majority of the drainage network. Imagery from EO-1’s Advanced Land Imager (ALI) captured in August 2004 was used to contrast techniques for quantifying water tracks using a combination of existing and lab-derived spectra of vegetation, soil and land covers. The results are compared to aerial photographs from 2007 where are water-tracks were digitized using an object orientated classification. The distribution of water tracks is analyzed with respect to glacial geomorphology, vegetation and soil GIS layers to develop logistic regression relationships.

Remote Sensing Technologies in Support of Wildland Fire Management Operations and Planning for Yukon.

Jason Adams, Spatial Database Administrator, Wildland Fire Management Unit, Protective Services Branch, Community Services, Government of Yukon.  Box 2703, C-19, Whitehorse, Yukon, Y1A 2C6.

David Milne, Prevention Coordinator, Wildland Fire Management Unit, Protective Services Branch, Community Services, Government of Yukon.  Box 2703, C-19, Whitehorse, Yukon, Y1A 2C6.

Remote sensing plays a key role in detecting, monitoring and documenting wildland fires in Yukon.  It provides a cost effective and efficient basket of tools in support of fire operations and planning on a day-to-day basis.

The recent availability of MODIS (Moderate Resolution Imaging Spectroradiometer) hotspot data through the platforms of Aqua and Terra in near real time, has greatly enhanced the capability of Yukon Wildland Fire Management to detect fires in remote areas. During periods of intense and smoky fire activity, such as occurred in 2004, MODIS has become the primary detection source for new fire ignitions while detection aircraft are grounded.

The Yukon government prioritizes fire management actions based upon a geographic five zone response policy.  Fires near communities will elicit a quick response from fire crews with the goal of complete extinguishment, while those in remote wilderness areas may only be monitored through the fire season.  Remote sensing, through the MODIS hotspot and 250m multi-spectral data provides opportunities to monitor fire growth in lower priority areas on a daily basis without the overhead of expensive aircraft flights. Specialized algorithms and scripts developed in conjunction with other provincial Fire Management Agencies automatically create fire perimeter polygons based upon hot spot point data.  These data are then used in turn to create daily fire activity maps and can also be used to show fire progression on large incidents.

Once the fire season is complete, there is a requirement to record and report the extent of fire activity through annual burned area maps.  LandSAT Thematic Mapper (TM) 30m resolution data are acquired at the end of each season to enable mapping of large fires throughout the territory.  The previous methods entailed transferring hand-drawn sketches on topographic maps to the GIS or through the use of airborne GPS tracks.  Digitizing fire scars on a tablet PC using hyperspectral LandSAT TM data has proven to be much more accurate than either of the two previous methodologies.  It is also quite a bit less expensive since it does not include any air charter costs.

CARACTÉRISATION DES TYPES D’Occupation DU SOL EN MILIEU URBAIN À PARTIR DE  L’Imagerie  RADAR EN BANDE C

Claude Codjia¹*, François Cavayas², Robert Desjardins³

¹Département de Géographie, Université du Québec à  Montréal, C.P.8888, succursale Centre-ville, Montréal Québec H3C 3P8, phone (514) 9873000, poste 0867, Courriel codjia.claude@uqam.ca
²Département de Géographie, Université de Montréal, C.P.6128, succursale Centre-ville, Montréal Québec H3C 3J7, phone 514 343 8016, Courriel francois.cavayas@umontreal.ca
³Département de Géographie, Université du Québec à Montréal, C.P.8888, succursale Centre-ville, Montréal Québec H3C 3P8, Courriel Desjardins.robert@uqam.ca

Le but cette étude est de caractériser et discriminer les types d’occupation du sol en milieu urbain sur les images Radarsat-1. Nous avons analysé la rétrodiffusion  du bâti et des différentes occupations du  sol en nous  basant  sur une typologie de densité du bâti  et en étudiant les paramètres statistiques de premier ordre correspondants sur les images Radarsat-1. Une simulation radar préalable, faite à partir d’un modèle numérique de surface, a permis de circonscrire les aires de rétrodiffusion des bâtiments en fonction des paramètres du capteur de  Radarsat-1. Les réponses radar correspondant à ces aires ont été analysées au travers des paramètres statistiques. Il ressort  globalement de cette investigation, d’une part, que les différents types d’occupation du sol basés sur la densité du bâti se distinguent relativement bien entre eux et qu’il est possible de reconnaître automatiquement ceux-ci; d’autre part, les réponses des objets urbains étudiés varient selon les paramètres du capteur dont le plus prépondérant est l’angle d’incidence. En outre, les variations de l’azimut du bâti par rapport à la trace au sol du satellite réduisent l’exactitude du résultat. Ce problème a été réglé en mettant au point un algorithme de compensation radiométrique qui est fonction de l’azimut. Par ailleurs, l’état phénoménologique de la végétation, les variations temporelles relatives à l’humidité du sol  et l’état de ce dernier influencent sensiblement les résultats.  La présente étude a été effectuée à partir d’un échantillon de plus de dix images Radarsat-1 de type F1, F2, F3 et F4 dont certaines ont été prises dans des conditions environnementales et saisonnières bien contrastées.

Méthode empirique de correction des effets cardinaux sur les images radarsat-1 portant sur le milieu urbain

Claude Codjia¹*, François Cavayas², Robert Desjardins³

¹Département de Géographie, Université du Québec à  Montréal, C.P.8888, succursale Centre-ville, Montréal Québec H3C 3P8, phone (514) 9873000, poste 0867, Courriel codjia.claude@uqam.ca
²Département de Géographie, Université de Montréal, C.P.6128, succursale Centre-ville, Montréal Québec H3C 3J7, phone 514 343 8016, Courriel francois.cavayas@umontreal.ca
³Département de Géographie, Université du Québec à Montréal, C.P.8888, succursale Centre-ville, Montréal Québec H3C 3P8, Courriel Desjardins.robert@uqam.ca

L’une des raisons qui limite la capacité d’interprétation des images radar en milieu urbain est la variabilité de la rétrodiffusion pour des objets identiques vus sous différents angles . En effet, des variations manifestes sur la même image prêtent à confusion et compliquent la tâche de l’interprète et des algorithmes de classification. L’objet de cette étude est de réduire ces écarts entre les échos du radar en mettant au point un algorithme qui corrige l’image en fonction de l’orientation des objets urbains par rapport au plan d’illumination dudit radar.

La présente étude commence par la recherche des relations entre l’orientation des objets et la rétrodiffusion du signal radar. Ces relations ont été établies grâce à une analyse approfondie d’images en modes ascendant et descendant ainsi que l’étude des rétrodiffusion de bâtiments identiques -matériaux et tailles- vus sous des angles  différents par rapport au satellite. L’algorithme de correction de la rétrodiffusion a été empiriquement mis en œuvre en observant le comportement de l’écho radar du bâti. Une règle de compensation de la rétrodiffusion a été établie en prenant comme référence les secteurs dont l’orientation est optimale pour la rétrodiffusion. Les autres endroits voient leur rétrodiffusion rehaussée en fonction de l’angle azimutal qu’ils font avec la trace du satellite.

Des tests ont été effectués sur différents types d’occupation du sol, notamment les zones résidentielles et les zones industrielles. En général les résultats obtenus pour les secteurs résidentiels sont concluants. Seuls quelques vieux quartiers donnent des résultats moyens en raison de la présence de grands arbres. Par contre le résultat est parfois mitigé pour les secteurs industriels. Les grandes installations de formes rectangulaires répondent très bien à l’algorithme tandis que celles de formes circulaires donnent des résultats moins intéressants. Quant à la végétation, en raison de sa structure non directionnelle, elle souffre peu de cet effet cardinal. Des tests de classification et de détection appliqués aux images issues de cette correction montrent des gains substantiels aussi bien sur la qualité que sur la précision  des résultats. Ces conclusions démontrent la pertinence de la compensation radiométrique comme méthode de correction pour les images radar.

Operational crop acreage estimation on a national scale based on statistics and remote sensing

Xianfeng JIAO1, Heather McNAIRN2, Bangjie YANG1 Jiali SHANG2 Zhiyuan PEI1

¹ Chinese Academy of Agricultural Engineering, Ministry of Agriculture of China 41 Maizidian, Chaoyang District, Beijing China 100026

² Agriculture and Agri-Food Canada, Ottawa, Ontario, Canada K1A 0C6

Proposed presenter
Dr. Xianfeng Jiao
Agriculture and Agri-Food Canada
960 Carling Avenue
Ottawa, Ontario, Canada
jiaox@agr.gc.ca
613-759-7309

The Chinese Academy of Agricultural Engineering (CAAE), Ministry of Agriculture of China (MOA), is mandated to monitor China’s main crop acreages. Remote sensing is considered the key technology for estimating crop acreage. Initially, remote sensing monitoring over the important cropping regions within China was the primary approach. Since 1999 CAAE has delivered, operationally, crop area estimates over these regions using inputs of EO data. Using this approach, annual crop acreage estimates are provided across China for wheat, corn, soybean, cotton and rice. However, such an approach poses many challenges. Crop acreages estimated over these selected regions must be extrapolated to provide crop acreage information for the entire country. This paper introduces a new sampling framework based on statistics and remote sensing and applies this method to paddy rice acreage estimation in the northeast of China.

An innovative sampling framework was developed by incorporating coarse and medium resolution satellite data in a statistical Double Sampling Method (DSM). The DSM was developed to estimate paddy rice acreage in the northeast of China. In phase I of the DSM total annual paddy rice planting area was derived from mid-resolution satellite imagery (MODIS). In this phase the annual paddy rice distribution and the sampling population were defined. The resultant distribution was used to define the phase II sampling. At a 95% level of confidence, the population was divided into 6 strata according to the method of accumulated square root. The sample was then taken from each of these strata using proportionate sampling. During the phase II sampling, standard 1:50,000 topographic maps were chosen as the sampling unit. This new unit design has significant advantages for sampling surveys within a geo-spatial context, when compared with sampling unit defined by county boundaries and satellite scenes.

Following this Double Sampling Method, a paddy rice inventory was conducted using remote sensing imagery. Crop acreage was estimated at a 95% confidence level.

This sampling scheme was applied to the national crop area estimation system in 2005 for paddy rice acreage estimation in the northeast of China. TM imagery was used for the paddy rice inventory. Humid and rainy conditions during the growing season make the acquisition of optical remote sensing imagery problematic. The availability of satellite data must be assured in order to meet acreage reporting deadlines. Consequently, CAAE and Agriculture and Agri-Food Canada (AAFC) are collaborating on methods to integrate SAR and optical imagery for operational crop acreage estimation, within the developed sampling framework.

This paper will introduce this new sampling approach and will discuss progress to date on integrated SAR and optical imagery for crop classification and acreage estimation.