Robert H. Fraser

Hotspot and NDVI differencing synergy (HANDS): A new technique for burned area mapping over boreal forest

Biomass burning releases significant amounts of trace gases and smoke aerosol into the atmosphere. This has an impact on the Earths radiation budget, the magnitude of which has not yet been well quantified. Satellite remote sensing is well suited to assessing the area of biomass burning, a prerequisite for estimating emissions at regional and global scales. Commonly used satellite-based techniques for measuring burned areas include thermal hotspot detection and multitemporal NDVI analysis, each having several limitations. Here we present a new, hybrid approach for boreal burned area mapping called HANDS, or hotspot and NDVI differencing synergy. The automated technique was tested using satellite data covering Canada for the 1995 and 1996 fire seasons, and comparing results with official burned area statistics and conventional fire surveys. HANDS computed a national burned forest area of 6.8 million ha in 1995 and 2.0 million ha in 1996, corresponding favorably to Canadian Forest Service estimates of 7.1 million ha and 1.9 million ha, respectively. Moreover, in most cases, the technique accurately delineated the boundaries of individual burns and identified some burns that were missed with conventional mapping. When employed in conjunction with NOAA-AVHRR imagery, HANDS provides a consistent means of mapping large burns (>10 km2), which are characteristic for the boreal forest. New generation sensors (e.g., SPOT VEGETATION, Terra MODIS) should enable its successful application to a wider range of environments.

Estimating fire-related parameters in boreal forest using SPOT VEGETATION

The majority of burning in the boreal forest zone consists of stand replacement fires larger than 10 km 2 occurring in remote, sparsely populated regions. Satellite remote sensing using coarse resolution (c1 km) sensors is thus well suited in documenting the spatial and temporal distribution of fires in this zone. The purpose of this study was to investigate the utility of the SPOT VEGETATION (VGT) sensor for estimating three key parameters related to boreal forest fire: burned area, postfire regeneration age, and aboveground biomass. Based on a sample of fires across Canada, the best overall discrimination of burned forest was provided by a normalized short-wave-based vegetation index (SWVI) that combines near-infrared (NIR) and short-wave infrared (SWIR) channels from VGT. Multitemporal differencing of this index from anniversary date VGTcomposites was combined synergistically with active fire locations from NOAA/AVHRR to map Canadian forest that burned during 1998 and 1999. National burned area estimates for both years were within 15% of those compiled by the Canadian Interagency Forest Fire Centre. The normalized index also was correlated (R=.68) with the age of regenerating forests in Saskatchewan and Manitoba that burned between 1949 and 1998. An artificial neural network (ANN) model developed using temporal metrics computed from VGT could predict the age of these forests with an RMS error of 7 years (R=.83). By contrast, forest biomass based on Canada’s Forest Inventory (CanFI) was estimated with relatively poor accuracy (RMS=32 tons/ha) from VGT reflectance and terrestrial ecozone using a network model. We conclude that the VGT instrument is effective for mapping large boreal burns at the end of a fire season and approximating the age of regenerating burns less than about 30 years old. This information can be useful to supplement conventional groundbased data sets in remote areas where coverage may be incomplete. D 2002 Published by Elsevier Science Inc. The boreal biome covers 17% of Earth’s land area and comprises about 25% of its forestland. The major ecosystems within the boreal zone (forests, peatlands, and tundra) contain more than 30% of terrestrial carbon stores, thus representing a major component of the global carbon budget (Kasischke, 2000). Wildfires are a dominant factor controlling ecological succession and carbon storage in boreal forests, burning on average nearly 1% of the total forest area annually. Fire has an immediate direct impact on the carbon balance of boreal forests resulting from the conversion of living biomass and soil carbon into atmospheric carbon (CO2, CO, CH4). Amiro et al. (2001) estimated that

Detecting long-term changes to vegetation in northern Canada using the Landsat satellite image archive

Analysis of coarse resolution (~1?km) satellite imagery has provided evidence of vegetation changes in arctic regions since the mid-1980s that may be attributable to climate warming. Here we investigate finer-scale changes to northern vegetation over the same period using stacks of 30?m resolution Landsat TM and ETM + satellite images. Linear trends in the normalized difference vegetation index (NDVI) and tasseled cap indices are derived for four widely spaced national parks in northern Canada. The trends are related to predicted changes in fractional shrub and other vegetation covers using regression tree classifiers trained with plot measurements and high resolution imagery. We find a consistent pattern of greening (6.1?25.5% of areas increasing) and predicted increases in vascular vegetation in all four parks that is associated with positive temperature trends. Coarse resolution (3?km) NDVI trends were not detected in two of the parks that had less intense greening. A range of independent studies and observations corroborate many of the major changes observed.

Estimating direct carbon emissions from Canadian wildland fires 1

In support of Canadas National Forest Carbon Monitoring, Accounting and Reporting System, a project was initiated to develop and test procedures for estimating direct carbon emissions from fires. The Canadian Wildland Fire Information System (CWFIS) provides the infrastructure for these procedures. Area burned and daily fire spread estimates are derived from satellite products. Spatially and temporally explicit indices of burning conditions for each fire are calculated by CWFIS using fire weather data. The Carbon Budget Model of the Canadian Forest Sector (CBM- CFS3) provides detailed forest type and leading species information, as well as pre-fire fuel load data. The Boreal Fire Effects Model calculates fuel consumption for different live biomass and dead organic matter pools in each burned cell according to fuel type, fuel load, burning conditions, and resulting fire behaviour. Carbon emissions are calculated from fuel consumption. CWFIS summarises the data in the form of disturbance matrices and provides spatially explicit estimates of area burned for national reporting. CBM-CFS3 integrates, at the national scale, these fire data with data on forest management and other disturbances. The methodology for estimating fire emissions was tested using a large-fire pilot study. A framework to implement the procedures at the national scale is described.

Mapping insect‐induced tree defoliation and mortality using coarse spatial resolution satellite imagery

Insect‐induced defoliation causes significant timber and carbon losses in many forested countries. The purpose of this investigation was to examine the potential use of coarse spatial resolution satellite imagery for mapping tree defoliation and mortality caused by a large insect infestation. We examined 1 km multi‐temporal SPOT Vegetation (VGT) data over a coniferous forest region in Quebec, Canada that was severely defoliated during 1998–2000 by the eastern hemlock looper. A logistic regression model based on satellite change metrics was developed to map defoliation and mortality. The results suggest that coarse imagery is effective for mapping large‐scale conifer forest mortality caused by insects, and could also be useful for near real‐time monitoring of severe defoliation, although with 2–3 times greater errors of commission.

Comparative analysis of daytime fire detection algorithms using AVHRR data for the 1995 fire season in Canada: Perspective for MODIS

Two fixed-threshold (CCRS and ESA) and three contextual (GIGLIO, IGBP, and MODIS) algorithms were used for fire detection with Advanced Very High Resolution Radiometer (AVHRR) data acquired over Canada during the 1995 fire season. The CCRS algorithm was developed for the boreal ecosystem, while the other four are for global application. The MODIS algorithm, although developed specifically for use with the MODIS sensor data, was applied to AVHRR in this study for comparative purposes. Fire detection accuracy assessment for the algorithms was based on comparisons with available 1995 burned area ground survey maps covering five Canadian provinces. Overall accuracy estimations in terms of omission (CCRS=46%, ESA=81%, GIGLIO=75%, IGBP=51%, MODIS=81%) and commission (CCRS=0.35%, ESA=0.08%, GIGLIO=0.56%, IGBP=0.75%, MODIS=0.08%) errors over forested areas revealed large differences in performance between the algorithms, with no relevance to type (fixed-threshold or contextual). CCRS performed best in detecting real forest fires, with the least omission error, while ESA and MODIS produced the highest omission error, probably because of their relatively high threshold values designed for global application. The commission error values appear small because the area of pixels falsely identified by each algorithm was expressed as a ratio of the vast unburned forest area. More detailed study shows that most commission errors in all the algorithms were incurred in non-forest agricultural areas, especially on days with very high surface temperatures. The advantage of the high thresholds in ESA and MODIS was that they incurred the least commission errors. The poor performance of the algorithms (in terms of omission errors) is not only due to their quality but also to cloud cover, low satellite overpass frequency, and the saturation of AVHRR channel 3 at about 321 K. Great improvement in global fire detection can probably be achieved by exploring the use of a wide variety of channel combinations from the data-rich MODIS instruments. More sophisticated algorithms should be designed to accomplish this.

Automatic detection of fire smoke using artificial neural networks and threshold approaches applied to AVHRR imagery

Satellite-based remote sensing techniques were developed for identifying smoke from forest fires. Both artificial neural networks (NN) and multithreshold techniques were explored for application with imagery from the Advanced Very High Resolution Radiometer (AVHRR) aboard NOAA satellites. The NN was designed such that it does not only classify a scene into smoke, cloud, or clear background, but also generates continuous outputs representing the mixture portions of these objects. While the NN approach offers many advantages, it is time consuming for application over large areas. A multithreshold algorithm was thus developed as well. The two approaches may be employed separately or in combination depending on the size of an image and smoke conditions. The methods were evaluated in terms of Euclidean distance between the outputs of the NN classification, using error matrices, visual inspection, and comparisons of classified smoke images with fire hot spots. They were applied to process daily AVHRR images acquired across Canada. The results obtained in the 1998 fire season were analyzed and compared with fire hot spots and TOMS-based aerosol index data. Reasonable correspondence was found, but the signals of smoke detected by TOMS and AVHRR are quite different but complementary to each other. In general, AVHRR is most sensitive to low dense smoke plumes located near fires, whereas smoke detected by TOMS is dispersed, thin, elevated, and further away from fires.

SPOT VEGETATION for characterizing boreal forest fires

The potential of the recent SPOT VEGETATION (VGT) sensor for characterizing boreal forest fires was investigated. Its capability for hotspot detection and burned area mapping was assessed by analysing a series of VGT, NOAA/AVHRR, and Landsat TM images over a 1541 km2

A semi-automatic segmentation procedure for feature extraction in remotely sensed imagery

This paper presents a semi-automatic procedure that integrates thresholding, region growing, and edge detection techniques for feature extraction in remotely sensed imagery. An interface has been developed to provide an interactive platform of the procedure. Thresholding technique is employed to sample object of interest. Estimated properties (i.e., mean and variance) of the sample are applied for feature extraction using region growing. Since the derived object is subject to the sample and initial conditions, edge detection is incorporated to calibrate initial parameters by examining how the derived object matches the local edges inherent in the imagery. The program is loosely linked to PCI (PCI Geomatics, Richmond Hill, Ontario, Canada), a widely distributed image processing software. We demonstrate applications of this procedure by deriving burned scars using SPOT VGT and NOAA AVHRR imagery.

Multi‐temporal Mapping of Burned Forest over Canada Using Satellite‐based Change Metrics

Abstract A procedure for continental‐scale mapping of burned boreal forest at 10‐day intervals was developed for application to coarse resolution satellite imagery. The basis of the technique is a multiple logistic regression model parameterized using 1998 SPOT‐4 VEGETATION clear‐sky composites and training sites selected across Canada. Predictor features consisted of multi‐temporal change metrics based on reflectance and two vegetation indices, which were normalized to the trajectory of background vegetation to account for phenological variation. Spatial‐contextual tests applied to the logistic model output were developed to remove noise and increase the sensitivity of detection. The procedure was applied over Canada for the 1998‐2000 fire seasons and validated using fire surveys and burned area statistics from forest fire management agencies. The area of falsely mapped burns was found to be small (3.5% commission error over Canada), and most burns larger than 10 km2 were accurately detected and mapped (R2 = 0.90, P<0.005, n = 91 for burns in two provinces). Canada‐wide satellite burned area was similar, but consistently smaller by comparison to statistics compiled by the Canadian Interagency Forest Fire Centre (by 17% in 1998, 16% in 1999, and 3% in 2000).