Remote Sensing. Measuring an object from a distance. For GIS, that means using photographic or satellite images to gather spatial data

Transcription

1 Remote Sensing Measuring an object from a distance For GIS, that means using photographic or satellite images to gather spatial data

2 Remote Sensing measures electromagnetic energy reflected or emitted from objects airborne or satellite-based instruments

3

4 Imagery - A Rich Data Source NASA Aster and TM Imagery, San Diego, CA

5 NASA Aster and SRTM Image Data, the Grand Canyon, AZ

6 Broad Spectral Range NASA MISR Data, the Hawaiian Islands

7 The World Changes Images provide a permanent record 10/22/07

8 The World Changes Images provide a permanent record 10/23/07

9 Change: Deforestation, Rondonia, Brazil (NASA Landsat Image)

10 Detailed, geometrically accurate record

11 Broad Area Coverage

12 Broad Area Coverage

13

14 Background Light (300,000 km per second) A stream of energy packets called photons This stream of particles behaves like a wave or traveling energy Light waves consist of electric and magnetic fields Measured by wavelength (peak to peak) or frequency (how many waves pass a point in space per second) wavelength and frequency are related wavelength x frequency = Speed of Light λ x f = C

15 Electromagnetic energy is a mixture of waves with different frequencies Each wave represents energy that varies at a given frequency. Source:

16 For each wave, there is an electronic (E) component and a magnetic component (M). The Amplitude (A) reflects the level of the electromagnetic energy. Source:

17 Plotting amplitude against the wavelength you get an electromagnetic curve, or spectrum Source:

18 Source:

19 Radiation (electromagnetic energy) is emitted by the Sun, and attenuated by the atmosphere. Specific bands of wavelengths are used for remote sensing

20 xrays Blue.4 to.5 μm (micrometers) Green.5 to.6 μm (micrometers) Red.6 to.7 μm (micrometers)

21

22 Visible Light - How we see colors Emit/Reflect Absorb Addition of light waves of different frequencies Absorb some of the frequencies you see what s reflected (paint,pigments like chlorophyll)

23 Energy, Atmosphere, and Surface Interactions

24 Sources of Information Variations in electromagnetic fields that can be used to identify and characterize objects: Spectral-radiometric (color, temperature) Spatial (pattern, size, shape, texture,...) Temporal

25 Spectral Reflectance Curves % of energy reflected Wavelength (μm)

26 A B C Spectral Resolution and Spectral Sampling The three shaded bars A, B, and C represent three spectral bands. The width of each bar covers a spectral range within which no signal variation can be resolved. The width of each spectral band represents its spectral resolution. The resolution of A is coarser than the resolution of B. This is because spectral details within band A that cannot be discriminated may be partly discriminated with a spectral resolution as narrow as band B. The resolution relationships among the three bands are: Resolution of A < Resolution of C < Resolution of B Adapted from Source:

27 Concept of Spectral Bands A B C Adapted from Source:

28 Sampled at different Spatial resolutions Adapted from Source:

29 Simple example of spectral signature (visible light only) 9,9,0 (real sunlight is 9,9,9) 0,0,9 0,9,0 9,0,0 Red,Green,Blue 5,3,0 Source:

30 A = Absorbed Visible Light R = Reflected Object description Color reflected Red Gre en Blu e Ex. Red apple red R A A Yellow block yellow R R A Green block green A R A Pinkish colored block (magenta) magenta R A R Blue block blue A A R Turquoise colored block (cyan) cyan A R R Source:

31 Simple example of spectral signature (visible light only) 9,9,0 (real sunlight is 9,9,9) 0,0,9 0,9,0 9,0,0 Red,Green,Blue 5,3,0 Source:

32 Radiation (electromagnetic energy) is emitted by the Sun, and attenuated by the atmosphere. Specific bands of wavelengths are used for remote sensing

33 xrays Blue.4 to.5 μm (micrometers) Green.5 to.6 μm (micrometers) Red.6 to.7 μm (micrometers)

34 Spectral Reflectance Curves % of energy reflected Wavelength (μm)

35 What Information Can Be Remotely Sensed? Fundamental Variables Planimetric (x,y) location and dimensions Topographic (z) location Color (spectral reflectance) Surface Temperature Texture Surface Roughness Moisture Content Vegetation Biomass

36

37

38 Two Main Types of Films

39

40 Black and White Image

41

42 True Color Image

43

44 Color Infrared Color Infrared Image

45

46

47 Upland forest just beginning growing season Cool season grass (brome, fescue) False Color True Color April False Color IR and True Color

48 Upland forest with full canopy False Color True Color May False Color IR and True Color

49 Bauer, 2000

50

51 Time of Year and Film Type Black and white (Panchromatic) lower costs, wide sensitivity "True" color - Enhanced interpretation Color infrared film (also near-ir or color-ir) - Enhanced vegetation, more contrast

52 What Information Can Be Remotely Sensed? Fundamental Variables Planimetric (x,y) location and dimensions Topographic (z) location Color (spectral reflectance) Surface Temperature Texture Surface Roughness Moisture Content Vegetation Biomass

53

54

55

56 Photo Interpretation This is the process of identifying and mapping the features that appear on the photos

57

58 What Information Can Be Remotely Sensed? Fundamental Variables Planimetric (x,y) location and dimensions Topographic (z) location Color (spectral reflectance) Surface Temperature Texture Surface Roughness Moisture Content Vegetation Biomass

59 Aerial Photographs

60

61

62

63 Photos are usually scanned and converted to digital images for on-screen display and measurements

64 Scale most commonly controlled by 1. flying height 2. lens focal length Scale is approximately equal to f / H f = focal length H = flying height

65 Photo Scale Set by flying height, focal length Most mapping cameras use 6 lens Reduce scale by flying higher

66 Increasing flying height reduces scale (objects get smaller, area covered by each photo increases) Increasing focal length increases scale (objects get larger, are covered decreases)

67 Scale is NOT Constant Can be over flat terrain with perfectly vertical photographs - rarely occurs Terrain - some objects are closer to lens, hence larger scale Tilt - causes perspective distortion

68 Perspective vs. Orthographic Views

69 Terrain Variation Causes Relief Displacement Features are displaced radially from their planimetric position due to differences in relative elevation

70 Tilt Distortion

71 Tilt measured as the angle between a line perpendicular to the film and a line perpendicular to the datum. Typically specified to be less than 3 o on vertical aerial photos.

72 How Big Are Tilt and Terrain Errors? Larger errors with more tilt (even under 3 o ) Larger errors with more relief (proportional to elevation difference) Larger at smaller scale

73 80 60 Terrain Range = 0 m Tilt = 0 o Tilt < 1.4 o Tilt 1.4 to 2.8 o Tilt 2.8 to 4.2 o Error (m) Photo Scale (1000s) 0

74 Error (m) Terrain Range = 500m Tilt = 0 o Tilt < 1.4 o Tilt 1.4 to 2.8 o Tilt 2.8 to 4.2 o Photo Scale (1000s) 0

75 Parallax - Relative shift in position with a change in viewing location. Closer (taller) objects shift more. We measure parallax to estimate height.

76 Overlapping Stereophotographs to create parallax shifts

77 Source: Source:

78 Overlapping Stereophotographs to create parallax shifts

79 Relief Displacement Height affects horizontal position Displacement are radial higher outward shift Lower inward shift In a planimetric map, A should occur at the same location as B But is displace by d Question: how much is d? Use similar triangles to find out

80 Similar Triangles Strategy: Measure p (from photo) We know H (from flight record) We can get h (e.g., from DEM) We need to relate d to p, h, and H And solve for d

81 Terrain Variation Causes Relief Displacement Features are displaced radially from their planimetric position due to differences in relative elevation

82 Big triangle CNS Small triangle 1 - Cna Small triangle 2 - ABS We may see: D/P = d / p and D/P = h/h So d/p = h/h d=p*h/h

83

84 So We know H Measure p on photo Know h from DEM Apply equation d = p* h/h Apply for every spot (cell, pixel) on the photo, yielding an ortho-corrected image

85 What about tilt? Tilt plus terrain distortion complicates the equations, e.g., Photo x = x0 f [ m11(xp-xl) + m21(yp-yl) +m13(-zl). Where m11 = cosφ*cosκ, m21= -cosφ*cosκ, etc. The idea is the same photo measurement are used with knowledge of flying height, ground height, and now tilt (ω, φ, κ) to calculate and remove displacement.

86 Softcopy Photogrammetric Workstations Orthophoto production

87

88 An orthophotograph or orthoimage tilt/terrain distortion removed

89 An orthophotograph or orthoimage tilt/terrain distortion removed

90 Besides geometric fidelity, we are also interested in the photo information content How do we interpret the photographs? Select a photographic system appropriate to the task, i.e., scale, coverage, time of year, and film type which best renders the details of interests

91 Scale Without magnification, you are stuck at about 2-3 mm To identify individual trees -10 m across with magnification scale =.5 mm / 10,000 mm, or about 1:20,000 without magnification scale = 2 mm / 10,000 mm, or about 1:5,000

92 Coverage Scale and format determines area per photo, e.g. 9" 1:10,000 scale yields photos which contain 7,500 feet on edge 9 = 0.75 ft * 10,000 ground feet/photo feet = 7,500 ft

93 Photo Dimensions and Areas for Some Common Photo Scales Photo Scale Scaled edge 9 Photo (miles) Width 60 Road (in) Scaled Photo Area Side of 40 ac. on Photo (in, side) 1: ac : ac :15, ,240 ac 1 1:20, mi :24, mi :40, mi 2 0.4

94 Satellite Images Advantages - High view, little relief displacement - Ultra-stable satellites, little tilt distortion - Extended spectral range, from radar to far infra-red - Low cost per unit area (for large study areas) - Digital images, which may be easily enhanced, integrated into a GIS

95 Satellite Images Disadvantages Limited acquisition flexibility, fixed schedules Expensive for small areas, due to fixed frame size Limited scale/resolution Requires sophisticated, moderately expensive systems to take advantage of digital image

96

97 Systems Ikonos, Quickbird 0.6 (panchromatic) to 3.5 meter (3-band color) resolution Three to five day repeat visit Images 10 to 30 km on a side Spot Panchromatic, 2.5 to 10 m resolution 3-band color, 5 to 20 m resolution Returns from every 5 to 26 days, depending on requirements and latitude Image 60 km on a side Thematic Mapper (TM) 7 bands, 30 m resolution Return time of 16 days Image approximately 185 km on a side

98 High-Resolution Satellite Systems IKONOS (1999) High resolution system 680 km orbit Revisit times are typically 1-3 days 1 meter panchromatic 4 meter visible/infrared Scan width of 13 km pointable.

99 Quickbird (2000) 61 cm (2 foot) pan meter mulitpsectral Image side 16.5 km 1-3 day revisit

100 Geoeye (2008) Highest resolution at 0.5 meters (soon to be 0.25 m) Color and panchromatic images) This half-meter resolution satellite image features a portion of the Bonneville Dam, located 40 miles east of Portland, Oregon in the Columbia River Gorge. The dam provides electrical power generation, fish and wildlife protection, recreation and navigation. Bonneville Dam was built and is managed by the U.S. Army Corps of Engineers and was designated as a National Historical Landmark District in The image was collected on Aug. 2, 2010 by the GeoEye-1 satellite from 423 miles in space as it moved from north to south over Oregon at a speed of four miles per second.

101 Geoeye (2008) Highest resolution at 0.5 meters (soon to be 0.25 m) Color and panchromatic images) This half-meter resolution satellite image shows Tahrir Square and surrounding buildings in Cairo, Egypt. The image was collected by the GeoEye-1 satellite on January 29, 2011 at 10:29 a.m. local time from 423 miles in space as it moved from north to south over Egypt at a speed of four miles per second.

102 Primary Uses for High Resolution Satellite Data On-screen digitizing similar to aerial photographs Infrastructure mapping (roads, etc.) Detail landuse mapping Topographic mapping Disaster assessment (fire, hurricane, flooding) Habitat and other mapping

103 Medium Resolution Sensing Systems Landsat Thematic Mapper (TM, ETM+) 7 spectral bands 30 meter resolution multispectral, 15 meter panchromatic 16-day repeat cycle data since 1984 Historical importance, new collections difficult Multispectral Scanner (MSS) 4 spectral bands 80 meter resolution 16-day repeat cycle data since 1972 SPOT coarse modes 5 spectral bands 10 to 20 meter resolution 1-3 day repeat cycle

104 TM Data Good for regional, some local analyses typically can t distinguis h objects smaller than 25 meters wide

105 SPOT 20 meter More detail than Landsat, but smaller imaged area

106 COARSE RESOLUTION LAND SENSORS AVHRR (since the 1970s) Advanced Very High Resolution Radiometer 5 spectral bands 1.1 km resolution 12 hour repeat cycle MODIS Moderate Resolution Imaging Sensor 36 spectral bands 1 km, 500m and 250 m resolutions 1 to 2 day repeat cycle

107 NOAA AVHRR Imagery

108 MODIS Image

109 Most common useful applications Landcover mapping, large areas e.g., wetlands, urban, forest Disaster evaluation, management Crop monitoring Change detection (for example, deforestation) Snow monitoring, runoff estimation Geologic prospecting Vegetation health monitoring

110 Image Classification from Multispectral Data

111 Satellite vs. Photos Which to Use? Satellites Lower detail (barely) Expanded spectrum Inherently digital Stable platform Higher flight path Inexpensive for large areas Aerial Photos Higher detail Less expensive for small areas Flexible repeat time Fly under clouds Simple handling