1 Availability and Potential Use of Low Resolution Satellite Imagery for Peacekeeping and Disaster Management Mryka Hall-Beyer
2 Spatial resolution: The ability to see detail Expectations people have: to be able to see as our eyes do, changing focus To be able to lift the binoculars and zoom in further To make the clouds go away
3 Tradeoffs: 1. Higher resolution = smaller area Constraint: file size, compression, download rate and storage / retrieval 2. Higher resolution = longer return time Constraint: orbital altitude, sun synchronicity, pointability 3. Higher resolution = more difficult rectification to subpixel accuracy Constraint: control points, precision of instrument positioning Summary: more detail = more cost
4 Low resolution: 250m to 16 km pixels Much low resolution imagery is free Gain in return for less spatial detail: Overview of large area Rapid processing turnaround, Rapid coverage return (daily or twice daily)
5 What is low and what is high resolution? Midrange is Landsat 30 m (TM, since 1984) or 80 m (MSS, older) $C900; repeat time 16 days
6 Pixel size Ha/pixel Pixel/ha 15x45 m lots/px 250 m m m m
7 Image use: At low resolution, What do we want to know? What can be seen on the surface that will give us this information? What are those objects : Size (Multispectral) colour Temporal variability? Are there any surrogates to help identify the object? Do these require manual interpreters or can they be automated?
8 What do we want to know at this resolution? 1. An overview device to situate and focus on sites for acquiring higher resolution imagery 2. Monitoring suspect areas: change detection 3. Mapping extent of natural disasters and population displacement
9 What is seen on the surface? Vegetation density (change must be distinguished from seasonal variability)
11 What is seen on the surface? Vegetation density (change must be distinguished from seasonal variability) Vegetation extent
12 MODIS image gallery
13 What is seen on the surface? Vegetation density (change must be distinguished from seasonal variability) Vegetation extent Land/water boundaries and water properties
14 Flooding in Mozambique MODIS Image Gallery
15 MODIS image gallery
16 What is seen on the surface? Vegetation density (change must be distinguished from seasonal variability) Vegetation extent Land/water boundaries and water properties Thermal signatures
17 MODIS image gallery
18 Objects size: On 1000 m, can detect land clearing, flooding, fires On 250 m, can also detect some roads (but not vehicles), large building construction, parking lots, air strips, large ships and ship tracks. Netherlands, May 6, m rgb=143 MODIS: MODIS Image Gallery
19 Objects multispectral colour: All low-resolution sensors include visible, near infrared and thermal bands. It is unlikely that objects would fail to be visible because of their spectral response. Microwave is an entirely different subject!
20 Objects temporal variability: Temporal variation will often be the target to be detected Automation of task requires consistent radiometric and atmospheric correction, and good location Spurious variation must be eliminated (e.g. vegetation seasonality) Close timing of successive images minimizes some of these problems
21 Surrogate objects: Examples Dust, smoke or thermal signatures Ship tracks
22 Stated purposes of low-res imagery: 1. to provide global observations and scientific understanding of land cover change and global productivity including trends and patterns of change in regional land cover, biodiversity, and global primary productivity. Rapidly changing land cover can be a surrogate indicator for camp formation, scorched earth of some sort, materiel concentration, natural disasters. Possible application in forensics.
23 Stated purposes of low-res imagery: 2. Seasonal-to-interannual climate predictions that improve forecasts of the timing and geographical extent of transient climate anomalies; Probably not directly useful
24 Stated purposes of low-res imagery: 3. Natural hazards including disaster characterization and risk reduction from wildfires, volcanoes, floods, and droughts; Direct application to disaster relief, in some cases to operations planning
25 Stated purposes of low-res imagery: 4. Long-term climate variability, to help scientists identify the mechanisms and factors that determine long-term climate variation and trends, including human impacts; Not directly relevant to acute planning, but eventual results will aid in long-term planning
26 Stated purposes of low-res imagery: 5. Atmospheric ozone, to help scientists detect changes, causes, and consequences of changes in atmospheric ozone. Probably not much here for our purposes.
27 Low resolution imagery: what is available Weather images from geostationary satellites AVHRR: 1.1 km and derived products at 4, 8 and 16 km Various Terra-1 (EOS-AM) products: new in 2000, particularly MODIS
28 Note: the ASTER instrument aboard EOS-AM is 15 to 90 m resolution, pointable for stereo, visible, near, mid and thermal infrared. At the moment, data is free. Tradeoff: not continuously operated so coverage is incomplete.
29 ASTER example image: details not given in source
30 Note 2: Other EOS-AM instruments are designed to characterize atmospheric aerosol, gas composition and particulate matter. These are at various x,y resolution and atmospheric depth. This information could be useful in specific disaster and peacekeeping applications. Data is archived and freely available.
31 Examples: lowest spatial resolution GOES 9: 1 to 8 km Vis, mir, tir ½ hour lead time web On-line 21 day archive, to 1979 by request http;// Boundaries are post-processed
32 Examples: AVHRR 1 km 1.1 km Vis and nir 12 hours except for composites Web/ftp Archive from 1978 Downloaded from Unrectified: note distortion across track
33 MODIS properties: 250 m red and nir (2 channels) 500 m blue, green, mid ir (5 channels) 1000 m thermal, narrow bandwidths in vis, nir, mid ir 3 weeks, eventually 24 hours Web search, ftp after notification Archive starting June 9, 2000
34 Example: MODIS 1000, 500, 250 m 1000 m vis
35 500m vis
36 250 m vis
37 1000 m 500 m 250 m
38 MODIS Image Gallery Comparison of 30m vs 250 m resolution Fires in Montana, August, 2000
39 How data acquisition works: NOAA and DAAC: different websites, same basic procedure 1. Register as a user (name, address, , phone). No restrictions nor fee.
40 2. Specify sensor, dataset, location, date, other criteria (e.g. cloud cover)
41 3. Choose from returned compatible datasets (images). Detailed metadata and often a browse product may be viewed onscreen
42 Submit order Receive confirmation number and download instructions via Usually 5 minutes to confirmation, 2 hours to instructions, including file size
43 Typical file sizes Alberta, AVHRR 1 km, compressed, 1 band: 5 MB; uncompressed, 5 bands: 18 MB MODIS scene (1354 x 3600 pixels), 250 m, bands: 26 MB MODIS scene, 1 km, all 36 bands: 125 MB ASTER scene, 60 x 60 km, all bands: 112 MB
44 File formats and usefulness: EOS-AM: EOS-HDF format (also used for Landsat) AVHRR: Level 1B (HRPT, LAC, GAC) Free downloadable software for viewing and exporting as.txt format: somewhat user-unfriendly (command line or Javabased) Most major image processing software systems will read these formats directly, but will not necessarily retain metadata such as channel labels. Georeferencing usually retained.
45 MODIS Direct Download system: Requires >=3m antenna, free software Presently at 5 university sites in US, one in Australia, UK, Russia, Spain Experimental, may become more common; equipped sites may make data available in near real-time Downloads all 36 bands when satellite in line of sight of receiving station Calculate overpasses (Space Station too!) at
46 Direct download example: test image 6 March
47 Direct download near real time: Image acquired at download station Nov 14 at 17:33 UCT, on web at 18:00 UCT. Note, however, lack of metadata.
48 Summary: 1. Low spatial resolution imagery is easily available, in near real time, at no cost for imagery. 2. No extraordinary software or hardware is required (except for direct reception) 3. Technician with moderate technical training could process images and make them available to the field with rapid turnaround
49 Summary 2: 1. Main requirement is good communications, including reasonably wide bandwidth for file transfer 2. Visible objects include some of direct interest. 3. Images may serve as convenient overview for resource allocation to high resolution images.
50 Some useful websites: MODIS image gallery: _GALLERY/modimgview/allimages.html ASTER information: Data gateway: Direct Broadcast documentation: AVHRR recent and archived images: Satellite overpass data: Weather satellite data:
51 -Oh, right We never did get rid of those clouds