On February 11, 2013 NASA launched the Landsat 8 Earth-observing satellite, and 100 days later transferred operational control to the US Geological Survey.
Landsat 8 is beaming 400 photos back to Earth per day, having reached its final altitude of 438 miles (705 kilometres) in April.
Enhanced Landsat 8 data have quickly found their way into a wide range of operational applications, including forest health monitoring by the US Forest Service, burn severity mapping by the USGS, NASA and the National Park Service, and cropland mapping by the National Agricultural Statistical Service.
Australian researchers are finding that improved Landsat 8 data have enhanced their ability identify and quantify areas of land degradation or improvement in the extensive Australian outback.
In Canada, generated from Landsat satellite data, the Earth Observation for Sustainable Development of Forests (EOSD) forest cover map consists of 610 segments, or tiles, each representing an area of about 15,000 square kilometres. The tiles detail 21 land cover classes as they existed in about 2000.
Multi-source Vegetation Inventory
The EOSD team used all of the Landsat images that intersect with Canada’s forested ecozones, which cover about 60 per cent of the country. About 80 per cent of the country was mapped, and all land cover types found in those images were classified, including forest cover.
At 25-metre resolution, the tiles represent the highest spatial resolution satellite-derived map data available for the total area of Canada covered.
The images taken by Landsat 8 are 19.5 per cent more accurate than those taken by any previous Landsat satellite and that difference can be very important to some applications. The monitoring of land degradation, forest health, and forest fire sensitivity and burn severity in Canada and the US have all been improved.
The 610 tiles can be downloaded from the Internet for a wide range of uses. The downloaded tiles can be pieced together without losing data integrity or information. Researchers can then use that information in their own landscape and land cover studies—either as is or as source information to generate their own value-added products.
In November, Joanne White and Michael Wulder, of the Canadian Forest Service (Pacific Forestry Centre), Natural Resources Canada, had published in the Canadian Journal of Remote Sensing their research, “The Landsat observation record of Canada: 1972-2012”.
“We report several spatial and temporal characteristics of the Landsat observation record for Canada (1972-2012), including image availability by year, growing season, sensor, ecozone, and provincial or territorial jurisdiction,” say the authors in the Abstract.
Hyperspectral sensors carried by satellites can document up to 490 different wavelengths of sunlight reflecting off the Earth’s surface. The sensors generate in-depth, spectrally layered data packages. Together, these stacks of spectral images reveal objects and data that cannot be picked up by multispectral sensors—sensors that record far fewer spectra or combine ranges of visible, near-infrared and infrared wavelengths into far fewer spectral bands.
Hyperspectral imagery has been used to map mineral deposits and geology in Canada’s north. Canadian Forest Service (NRCan-CFS) researchers are now developing ways to use it in forestry.
Each type of ground cover—often each species of ground cover—absorbs and reflects a specific combination of wavelengths. If these are identified, validated and made available, they could be used to improve forest inventory and health information, as well as increase information about biodiversity, natural disturbances and the effects of climate change in Canada’s forests.
For example, NRCan researchers and their colleagues have used imagery recorded by a satellite orbiting 700 kilometres above the Earth to map five individual tree species growing along British Columbia’s coast. The researchers also used similar airborne imagery to map chlorophyll, water content, and nitrogen levels within west coast forest canopies’ leaves.
Natural Resources Canada uses the information to estimate forest inventory from many sources. A project now underway in the Northwest Territories (NWT) is applying a novel approach to this challenge of estimating forest inventory attributes. Using a variety of new methods developed by NRCan-CFS researchers and several partner agencies, information is being collected from multiple data sources, including: field plots; LiDAR (Light Detection and Ranging) units mounted on aircraft and satellites; Landsat Thematic Mapper satellite images; and, existing forest inventory where available.
When Madison’s spoke to Wulder for the March 9, 2012, (please see “Vegetative Remote Sensing” Vol 62 No 10) issue of your Madison’s Lumber Reporter, the CFS was in the process of collecting and organizing data collected by LiDAR.
This volume of information will help guide Canada’s decision makers to new forest policy in view of total value across the landscape.
For the NWT project, these maps were subsequently generated from a spatial modelling and mapping exercise that scaled the LiDAR-estimated inventory attributes with Landsat Thematic Mapper and other biophysical data.
The satellite-derived data were translated into a format referred to as the Satellite Vegetation Inventory (SVI), to resemble a conventional forest inventory dataset. Over an area of interest, the Multi-source Vegetation Inventory (MVI) consists of the Forest Vegetation Inventory (FVI) and the SVI in locations where FVI data does not exist. The MVI data can be easily viewed and manipulated in a Geographic Information System (GIS) by forest technicians and managers.
“In the Northwest Territories, the MVI was completed over a pilot study area and is now being extended to much of the southern Taiga Plains Ecozone, encompassing an area about 200 000 km2 in size,” according to the Natural Resources Canada website, here https://www.nrcan.gc.ca/forests/remote-sensing/13441).
“Combining several remote sensing technologies in this way to estimate forest inventory attributes will greatly improve resource assessment and reporting by the NWT’s Department of Environment and Natural Resources as well as by CFS.
“The more up-to-date inventory information generated through the MVI method will also be invaluable to the National Forest Carbon Monitoring, Accounting and Reporting System and for use in updating the NWT’s part of the National Forest Inventory.”
According to the NWT Environment and Natural Resources website, ground sampling programs always accompany a management level inventory and may also be carried out on reconnaissance or operational inventories. They are conducted to provide additional information not available from aerial photography, most importantly volume. Temporary sample plots are established in the field and the data are used to provide information on a wide range of attributes including trees, ecology, site and soils.
Remote sensing can be used to characterize forest ecosystems across large areas. However, the effectiveness of using remotely sensed data for large-area forest inventories depends on the relationship between the scale of the object of interest and sensor-specific characteristics such as resolution and spatial extent, says the NRCan website.
Remote sensing systems that acquire images with large spatial extents will generally have a lower resolution, and thereby capture less detail, than images acquired at a higher resolution, which usually depict forest characteristics across smaller spatial extents. For example, trees are smaller than the pixel size of medium spatial resolution remotely sensed data (10 to 30 metres), and this prohibits measurement of specific properties, such as tree locations and crown dimensions. At higher spatial resolutions, however, trees become larger than the image pixel size, allowing for direct measurement of particular properties.
Remote sensing can, for example, create imagery to assess fire fuel hazard potential and pest and disease outbreaks in both native and planted forests. By integrating current and next-generation remote-sensing data with geographical and terrain information, scientists are able to provide solutions that are cost-effective and accurate.