Lecture 2_Remote Sensing Fundamentals.pdf

LSGI536 Remote Sensing Image Processing Lecture 2 Remote Sensing Fundamentals: Platforms, Sensors, and Image Characteristics Dr. Zhiwei Li Research Assistant Professor Department of Land Surveying and Geo-Informatics The Hong Kong Polytechnic University Email: [email protected] Outline 1. Basic...

LSGI536 Remote Sensing Image Processing Lecture 2 Remote Sensing Fundamentals: Platforms, Sensors, and Image Characteristics Dr. Zhiwei Li Research Assistant Professor Department of Land Surveying and Geo-Informatics The Hong Kong Polytechnic University Email: [email protected] Outline 1. Basic Specifications of Satellite and Sensor 2. Common Satellite Platforms and Sensors 3. Characteristics of Remote Sensing Images 2 Section 1 Basic Specifications of Satellite and Sensor 3 Basic Specifications of Satellite and Sensor Sensors and Imaging Scanners Satellite Orbit Swath Width Repeat Cycle and Revisit Time FOV and IFOV 4 Measurement Techniques 5 Sensors Sensors are instruments that collect data about Earth processes or atmospheric components. Along with being carried aboard satellites or aircraft, sensors also can be installed on the ground (in situ). There are two types of sensors: o Active sensors provide their own source of energy to illuminate the objects they observe. o Passive sensors detect energy emitted or reflected from the environment. How do sensors work? [link1, link2] 6 Multispectral Scanners The imaging technologies utilized in satellite programs have ranged from traditional cameras to mechanical scanners that record images of the earth’s surface by moving the instantaneous field of view of the instrument across the earth’s surface to record the upwelling energy. The system that can collect data in many spectral bands and over a wider range of EM spectrum is called multispectral scanner. Multispectral scanners are either ✓ Across-track (whisk broom) or ✓ Along-track (push broom) 7 Multispectral Scanners Across-track scanner (whisk broom) Along-track scanner (push broom) 8 Multispectral Scanners Across-track scanner (whisk broom) A. Oscillating mirror B. Detectors C. IFOV D. GSD E. Angular field of view (θ) F. Swath (2H*tan(θ/2)) Along-track scanner (push broom) A. Linear array of detectors B. Focal plane of the image C. Lens D. GSD 9 Satellite Orbit Types of Satellite Orbits https://www.youtube.com/watch?v=n70zjMvm8L0 5 min (The 6th and 7th chapters) 10 Satellite Orbit Geostationary orbit Polar orbit Sun-synchronous orbit 11 Geostationary orbits (GEO) About 35,780 km above ground Geostationary orbits ▪ Period of rotation equal to that of Earth (24 hours) so the satellite always stays over the same location on Earth. ▪ Constant spatial coverage. ▪ Ideal orbit for telecommunications or for monitoring continent-wide weather patterns and environmental conditions. ▪ Multiple observations per day. Kepler’s Third Law 12 Geostationary orbit example - metrological satellite Real-time image of Himawari-9 https://www.data.jma.go.jp/mscweb/data/himawari/sat_img.php?area=fd_ 13 Polar orbits Pass over the Earth’s polar regions from north to south. The orbital track of the satellite does not have to cross the poles exactly for an orbit to be called polar, an orbit that passes within 20 to 30 degrees of the poles is still classed as a polar orbit. Global coverage Larger swath size means higher temporal resolution. Sometimes orbital gaps. 14 Sun Synchronous orbits Near-polar orbits Cover each area of the world at a constant local time of day called local sun time. Since there are 365 days in a year and 360 degrees in a circle, it means that the satellite has to shift its orbit by approximately one degree per day. This ensures consistent illumination conditions when acquiring images. 15 Swath Width As a satellite revolves around the Earth, the sensor “sees” a certain portion of the Earth’s surface. The area imaged on the surface, is referred to as the swath. Imaging swaths for spaceborne sensors generally vary between tens and hundreds of kilometres wide. As the satellite orbits the Earth from pole to pole, its east-west position wouldn’t change if the Earth didn't rotate. However, as seen from the Earth, it seems that the satellite is shifting westward because the Earth is rotating (from west to east) beneath it. This apparent movement allows the satellite swath to cover a new area with each consecutive pass. The satellite’s orbit and the rotation of the Earth work together to allow complete coverage of the Earth’s surface, after it has completed one complete cycle of orbits. 17 Swath Width Landsat 8 Swath Animation https://youtu.be/xBhorGs8uy8 ~ 1 min 18 Repeat Cycle A satellite generally follows a path around the Earth. The time taken to complete one revolution of the orbit is called the orbital period. The satellite traces out a path on the earth surface, called its ground track, as it moves across the sky. As the Earth is also rotating, the satellite traces out a different path on the ground in each subsequent cycle. The time interval in which nadir point of the satellite passes over the same point on the Earth’s surface for a second time (when the satellite retraces its path) is called the repeat cycle of the satellite. For instance, Radarsat-1 was designed in a Sunsynchronous orbit with a 343/24 repeat cycle that implies after 343 revolutions and 24 nodal days, the satellite shall, within an error, return to the same spot over the Earth. 19 Revisit Time i.e. temporal resolution The satellite revisit time is the time elapsed between observations of the same point on earth by a satellite. Different from the repeat cycle which only depends on the orbit, the revisit time is relevant to the payload of the satellite. It depends on the satellite's orbit, target location, and swath of the sensor. Repeat cycle and revisit time https://hsat.space/wp-content/uploads/2020/09/SSO.mp4?_=1 20 FOV and IFOV FOV, or Field of View, is the whole area that your sensor can see at a set distance. IFOV, or Instantaneous Field of View is the smallest detail within the FOV that can be detected or seen in an instant. The IFOV is normally expressed as the cone angle (β) within which incident energy is focused on the detector. FOV and IFOV are related to spatial resolution. 21 Size of IFOV Most airborne and satellite systems IFOV=0.5-5 mRad Small IFOV good for high spatial detail (i.e., high spatial resolution). Large IFOV means large amount of energy focused on the detector. ✓ more sensitive to scene radiance; ✓ better radiometric resolution; ✓ can distinguish very slight energy differences; Thus, there is a trade-off between high spatial resolution and high radiometric resolution in the design of multispectral scanner systems. 22 IFOV and signal-to noise ratio For large IFOV, signal much greater than background electronic noise - thus higher S/N ratio than one with a small IFOV. Thus, spatial resolution is sacrificed for these higher signal levels. System noise in Landsat image band 2 23 Questions Which type of orbit is commonly used for weather forecasting? List and describe the four types of image resolutions. Explain the difference between FOV and IFOV. 24 Section 2 Common Satellite Platforms and Sensors 25 Common Satellite Platforms and Sensors Medium Resolution Satellites - Landsat Moderate Resolution Satellites - Terra (MODIS) Copernicus Satellites constellation - Sentinels High Resolution Satellites o IKONOS o QuickBird o WorldView Small Satellites and Sensors 26 Medium Resolution Sensors Landsat A joint program of National Aeronautics and Space Administration (NASA) and United States Geological Survey (USGS) www.nasa.gov/landsat landsat.usgs.gov 27 Landsat History Landsat: Celebrating 50 Years (5 mins) https://www.youtube.com/watch?v=7XKVSTX1vdE 28 Landsat Missions Overview Landsat Missions Timeline 29 Landsat Missions Overview Spectral Bandpasses for all Landsat Sensors 30 Landsat 7 Landsat 7 carries the Enhanced Thematic Mapper Plus (ETM+) sensor, an improved version of the Thematic Mapper instruments that were onboard Landsat 4 and Landsat 5. Eight spectral bands, including a pan and thermal band: 1. 2. 3. 4. 5. 6. 7. 8. Band 1 Blue (0.45 - 0.52 µm) 30 m Band 2 Green (0.52 - 0.60 µm) 30 m Band 3 Red (0.63 - 0.69 µm) 30 m Band 4 NIR (0.77 - 0.90 µm) 30 m Band 5 SWIR-1 (1.55 - 1.75 µm) 30 m Band 6 Thermal (10.40 - 12.50 µm) 60 m Low Gain / High Gain Band 7 SWIR-2 (2.08 - 2.35 µm) 30 m Band 8 Panchromatic (PAN) (0.52 - 0.90 µm) 15 m 31 Landsat 7 On May 31, 2003 The Scan Line Corrector (SLC) that compensates for the forward motion of the satellite was failed The failure is permanent The sensor’s line of sight traces a zig-zag pattern Repeat cycle 16 days 32 Landsat 7 Daily Landsat-scale evapotranspiration estimation over a forested landscape in North Carolina, USA [link] 33 Landsat 8 Operational Land Imager (OLI) - Built by Ball Aerospace & Technologies Corporation Nine spectral bands, including a pan band: 1. 2. 3. 4. 5. 6. 7. 8. 9. Band 1 Coastal/Aerosol (0.43 - 0.45 µm) 30 m Band 2 Blue (0.450 - 0.51 µm) 30 m Band 3 Green (0.53 - 0.59 µm) 30 m Band 4 Red (0.64 - 0.67 µm) 30 m Band 5 NIR (0.85 - 0.88 µm) 30 m Band 6 SWIR-1 (1.57 - 1.65 µm) 30 m Band 7 SWIR-2 (2.11 - 2.29 µm) 30 m Band 8 Panchromatic (PAN) (0.50 - 0.68 µm) 15 m Band 9 Cirrus (1.36 - 1.38 µm) 30 m Thermal Infrared Sensor (TIRS) - Built by NASA Goddard Space Flight Center Two spectral bands: 1. 2. Band 10 TIRS 1 (10.6 - 11.19 µm) 100 m Band 11 TIRS 2 (11.5 - 12.51 µm) 100 m Repeat cycle 16 days 34 Landsat 7 vs Landsat 8 36 Landsat 9 OLI-2 sensor TIRS-2 sensor Illustration of Landsat 9 Observatory Source: Landsat 9 Data Users Handbook 37 Landsat 9 ❑ The OLI–2 improves radiometric precision (14-bit quantization increased from 12 bits for Landsat 8). Nine spectral bands: 1. 2. 3. 4. 5. 6. 7. 8. 9. Band 1 Coastal/Aerosol (0.43 - 0.45 µm) 30 m Band 2 Blue (0.450 - 0.51 µm) 30 m Band 3 Green (0.53 - 0.59 µm) 30 m Band 4 Red (0.64 - 0.67 µm) 30 m Band 5 Near-Infrared (0.85 - 0.88 µm) 30 m Band 6 SWIR 1(1.57 - 1.65 µm) 30 m Band 7 SWIR 2 (2.11 - 2.29 µm) 30 m Band 8 Panchromatic (PAN) (0.50 - 0.68 µm) 15 m Band 9 Cirrus (1.36 - 1.38 µm) 30 m ❑ Thermal Infrared Sensor 2 (TIRS-2) 1. Band 10 TIRS 1 (10.6 - 11.19 µm) 100 m 2. Band 11 TIRS 2 (11.5 - 12.51 µm) 100 m 38 Landsat 8 + Landsat 9 Landsat 9 will replace Landsat 7 (launched in 1999), taking its place in orbit (8 days out of phase with Landsat 8). The combined Landsat 8 and Landsat 9 revisit time for data collection with be every 8 days. 39 Worldwide Reference System (WRS) The Worldwide Reference System (WRS) is a global notation system for Landsat data. It enables a user to inquire about satellite imagery over any portion of the world by specifying a nominal scene center designated by PATH and ROW numbers. Landsat satellites 1, 2 and 3 follow WRS-1, and Landsat satellites 4, 5, 6, 7, 8, and 9 follow WRS-2. [link] 40 Worldwide Reference System (WRS) A map of the Worldwide Reference System-2 [link] 41 Worldwide Reference System (WRS) [link] 42 Worldwide Reference System (WRS) A B 2 1 3 C Landsat WRS Path Row that covers HK 43 Download Landsat imagery https://earthexplorer.usgs.gov/ 44 Terra Instruments Since 1999, the Terra satellite has been continually observing Earth. Terra is an international mission carrying five instruments that observe Earth’s atmosphere, ocean, land, snow and ice, and energy budget. United States: MODIS (Moderate-resolution Imaging Spectroradiometer) CERES (Clouds and Earth’s Radiant Energy System) MISR (Multi-angle Imaging SpectroRadiometer) Japan: ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) Canada: MOPITT (Measurements of Pollution in the Troposphere) 45 Terra Instruments Twenty Years of Terra in Our Lives 3 min 46 Moderate Resolution Imaging Spectroradiometer (MODIS) MODIS is a key instrument aboard the Terra (originally known as EOS AM-1) and Aqua (originally known as EOS PM-1) satellites. Terra MODIS and Aqua MODIS are viewing the entire Earth's surface every 1 to 2 days, acquiring data in 36 spectral bands, or groups of wavelengths. Orbit: 705 km, 10:30 a.m. descending node (Terra) or 1:30 p.m. ascending node (Aqua), sun-synchronous, near-polar, circular. Swath Dimensions: 2330 km (cross track) by 10 km (along track at nadir). Design Life: 6 years These data will improve our understanding of global dynamics and processes occurring on the land, in the oceans, and in the lower atmosphere. https://modis.gsfc.nasa.gov/about/specifications.php 47 Moderate Resolution Imaging Spectroradiometer (MODIS) Spatial Resolution: 1. 250m (bands 1-2) 2. 500m (bands 3-7) 3. 1000m (bands 8-36) Wavelengths of 36 bands: 1. 1-19 from 405 nm to 2155 nm 2. 20-36 from 3.66 µm to 14.28 µm Four different categories of bands for earth observation Bands 1 to 7: Land Bands and Cloud Bands, Bands 8 to 16: Ocean Colour Bands, Bands 17 to 36: Atmosphere and Cloud Bands, Bands 20 to 36: Thermal Bands (collect measurements in night mode). 48 Moderate Resolution Imaging Spectroradiometer (MODIS) 36 channels in MODIS image [link] 49 Application examples of MODIS imagery “Massive sandstorm forms over Mongolia, hits China's capital Beijing as the worst in a decade.” Early Season Dust Storm Hits Beijing (March 15, 2021, Link) 50 ESA Sentinels for Copernicus ESA (European Space Agency) is developing a new family of missions called Sentinels specifically for the operational needs of the Copernicus programme. Each Sentinel mission is based on a constellation of two satellites to fulfil revisit and coverage requirements, providing robust datasets for Copernicus Services. These missions carry a range of technologies, such as radar and multispectral imaging instruments for land, ocean and atmospheric monitoring. Sentinel family 51 ESA Sentinels for Copernicus Sentinel-1 is a polar-orbiting, all-weather, day-and-night synthetic aperture radar for land and ocean services. Sentinel-1A was launched on 3 April 2014 and Sentinel-1B on 25 April 2016. Sentinel-2 is a polar-orbiting, multispectral high-resolution imaging mission for land monitoring to provide, for example, imagery of vegetation, soil and water cover, inland waterways and coastal areas. Sentinel-2A was launched on 23 June 2015 and Sentinel-2B followed on 7 March 2017. https://rus-copernicus.eu/portal/ 52 ESA Sentinels for Copernicus Sentinel-2 Data Products 53 ESA Sentinels for Copernicus Sentinels for Copernicus (Optional) https://youtu.be/xcflQZJ5n88 ~ 5 mins Know all about the Sentinel Mission https://youtu.be/W3fv7TUmqf8 ~ 3 mins 55 DigitalGlobe Commercial satellites DigitalGlobe Satellite Imagery: Worldview, GeoEye and IKONOS 56 IKONOS Commercial satellite Panchromatic: 0.45-0.90 μm, 0.82 m. Multispectral (3.2 m): #1: Blue 0.45–0.52 μm, #2: Green 0.52–0.60 μm, #3: Red 0.63–0.69 μm, #4: Near IR 0.76–0.90 μm Data quantization: 11-bit (2 bytes per pixel) pixel values 0-2047 Swath width: 11.3 km Areas of interest: single image at 13 km x 13 km Revisit time: less than 3 days Decommissioned in March 2015 57 QuickBird Panchromatic resolution 0.6m Multispectral resolution 2.4m Mission duration: 13 years and 2 months De-orbited on January 27, 2015 58 WorldView-4 Specification Details Orbit Sun-synchronous Altitude 617km Mission Lifetime Decommissioned Spatial Resolution Panchromatic 31 cm at nadir (GSD)* Multispectral 1.24 m at nadir (GSD)* *Ground Sample Distance Accuracy < 4 m CE90 Spectral Bands Panchromatic: 450 – 800 nm Blue: 450 – 510 nm Green: 510 – 580 nm Red: 655 – 690 nm Near infrared: 780 – 920 nm Stereo Available Yes Largest Scale 1:1000 Dynamic Range 11 bit Coverage: Up to 680,000km² per day Pan sharpened WorldView-4 image at 0.3m resolution 59 PROBA-1 (Project for On-Board Autonomy) Compact high-resolution imaging spectrometer (CHRIS) weight 14kg This small (60×60×80 cm; 95 kg) boxlike system, with solar panel collectors on its surface, has remarkable image-making qualities. It hosts two instruments including a hyperspectral system (200 narrow bands) that images at 17 m resolution, and a monochromatic camera that images visible light at 5 m resolution. 60 CubeSat | Mini cube satellites What is a CubeSat? https://youtu.be/nsdMcqiBmvY (2 mins) Cubesats | Mini cube satellites https://youtu.be/-BGXRGoEnAc (15 mins) Educational CubeSat project: https://github.com/oresat/getting-started 61 Unmanned Aerial Vehicle (UAV) – Micro-drones [link] How are drones helping farmers keep an eye on crops? [link] (2:10 - 3:15) 62 Questions Describe the strengths and limitations of the satellites introduced 63 Section 3 Characteristics of Remote Sensing Images 64 Characteristics of Remote Sensing Images Raster image Binary data Color Image Image resolutions 65 Digital data A data model is a way of defining and representing reality in a system, different models are used for different systems. In particular, GIS or geospatial data models are used to describe geographic features in the reality. In general, we have the following two models Vector data model represents features in term of discrete points, lines, and polygons Raster data model represents features in term of a matrix of cells In raster, each cell carries value representing the geographic characteristic. 66 Raster Image 67 Raster Image 68 Raster Image 2 20 35 50 43 35 50 43 35 50 50 73 96 119 119 134 28 43 58 50 43 58 50 43 58 58 81 103 126 126 141 50 66 81 73 66 81 73 66 81 81 103 126 149 149 149 66 81 96 88 81 96 88 81 96 96 119 141 164 164 157 50 66 81 73 66 81 73 66 81 81 103 126 149 149 164 58 73 88 81 73 88 81 73 88 88 111 134 157 157 172 66 81 96 88 81 96 88 81 96 96 119 141 164 164 179 73 88 103 96 88 103 96 88 103 103 126 149 172 172 187 66 81 96 88 81 96 88 81 96 96 119 141 164 164 194 66 81 96 88 81 96 88 81 96 96 119 141 164 164 202 73 88 103 96 88 103 96 88 103 103 126 149 172 172 210 88 103 119 111 103 119 111 103 119 119 141 164 187 187 217 111 126 141 134 126 141 134 126 141 141 164 187 210 210 225 119 134 149 141 134 149 141 134 149 149 172 194 217 217 232 126 141 157 149 141 157 149 141 157 157 179 202 225 225 240 69 Comparisons between remote sensing images and natural images Remote Sensing Images Landsat-8 OLI true color images illustrating the coverage over the coastal waters of French Guiana (4 scenes) Natural Images Examples in the ImageNet dataset Image Source: Zorrilla et al., 2019, Optics Express; Deng et al., 2009, CVPR 70 Comparisons between remote sensing images and natural images Image resolution (size) Remote Sensing Images The width/height of a Landsat-8 image is 6000-8000 Large image size Natural Images The average size of an ImageNet image is 469x387 Relatively small image size Image Source: Zorrilla et al., 2019, Optics Express; Deng et al., 2009, CVPR 71 Comparisons between remote sensing images and natural images Image channel Remote Sensing Images 11bands (Coastal aerosol, B, G, R, NIR, SWIR1, SWIR2, Panchromatic, Cirrus, TIRS1, TIRS2) Multiple channels/bands Natural Images 3 R-G-B bands RGB channels Image Source: Zorrilla et al., 2019, Optics Express; Deng et al., 2009, CVPR 72 Comparisons between remote sensing images and natural images Geo-location information Remote Sensing Images With geo-location information Natural Images Without geo-location information (might contain geo-tags) Image Source: Zorrilla et al., 2019, Optics Express; Deng et al., 2009, CVPR 73 Comparisons between remote sensing images and natural images Remote Sensing Images Natural Images Large image size Relatively small image size Multi-channel RGB channel Large data value range (e.g. 0-65535) Small data value range (0-255) With geo-location (georeferenced) information Without geo-location information 74 Spectral characteristics: bits and bytes All instructions carried out within a computer are in Binary Code which consists of the digits 0 & 1. These are called binary digits or bits. This machine code is executed by a series of electrical pulses which send signals off (0) or on (1). The number 2 is the base of the binary number system, just as 10 is the base of the decimal number system Any binary number comprises a string of bits that are worked from right to left in increasing powers of 2. Thus the binary number 101 is interpreted in decimal (our number system) as 5 (1 × 20=1, 0 × 2¹=0, 1 × 2²=4→5). 75 Examples of binary data MODIS Data product Quality Assessment flags Landsat-8 Quality Assessment Band Bits Bit 0 = 0 = not fill Bit 1 = 0 = not a dropped frame Bit 2 = 0 = not terrain occluded Bit 3 = 0 = not determined Bit 4-5 = 01 = not water Bit 6-7 = 00 = not determined Bit 8-9 = 00 = not determined Bit 10-11 = 01 = not snow / ice Bit 12-13 = 10 = could be cirrus cloud Bit 14-15 = 11 = cloudy Source: MODIS Surface Reflectance User’s Guide (Collection 6) Landsat 8 Data Users Handbook 76 Conversion between binary digits and decimal numbers (10111)₂ = (1 × 2⁰) + (1 × 2¹) + (1 × 2²) + (0 × 2³) + (1 × 2⁴) = (23)₁₀ Mod(x, y): function return remainder of two numbers after division Mod(23,2) =1 Mod(11,2) =1 Mod(5,2) =1 Mod(2,2)=0 Mod(1,2)=1 77 Conversion between binary digits and decimal numbers (1010101)₂ = (?)₁₀ (99)₁₀ = (?)₂ 78 Conversion between binary digits and decimal numbers (1010101)₂ = (85)₁₀ (99)₁₀ = (1100011)₂ 79 Counting in Binary With 6 bits the maximum number that can be represented is 63. What is the maximum number with 7 bits? What is the maximum number with 8 bits? (8-bit scale) 80 Counting in Binary With 6 bits the maximum number that can be represented is 63. What is the maximum number with 7 bits? 63 + 1×26= 127 What is the maximum number with 8 bits? 127 + 1×27= 255 The number of bits needed to represent most remote sensing data is 8 This is convenient since the standard storage unit in computers is 8 bits This corresponds to 1 byte, which takes up only 1 unit of computer storage space (memory or disk) 81 Bytes Thus 1 byte = 1 unit, 1Kb = 1024 bytes (210), 1Mb = 1024 × 1024 bytes = 1,048,576 bytes Using this system, some numbers require more storage space than 1 byte, e.g., integers above 256, and real numbers, examples are 270, 3.456, 0.24 In remote sensing (and also GIS) we recognize that data will be stored in order of preference as 1. BYTE data (least storage space, smaller file sizes) 2. INTEGER 3. REAL (largest file sizes) Some remotely sensed data are measured on a scale of 01023 and the storage requires 10 bits Most computer systems allow Integers to be stored as 8-, 16-, or 32-bit quantities 82 Data type Data type (name) Typical bit length Value range byte / int8 1 byte = 8 bit -128 to 127 (signed); 0 to 255 (unsigned) int16 2 byte = 16 bit -32768 to 32767 (signed); 0 to 65535 (unsigned) float 4 byte = 32 bit -3,4e^38 to 3,4e^38, ‘single precision’; floating point number according to IEEE 754 double 8 byte = 64 bit -Inf to +Inf, ‘double precision’ Define the type of value and its range. A variable can be operated and transformed in the computer program. 83 Color Image Color image is produced by using three raster arrays, i.e. RGB Each array element holds pixel values that represent the levels of one the three primary Colors of light. Normally, each pixel value has 0-255 levels or 8-bit data but could be 10-bit, e.g. AVHRR 84 Color allocation for Color composite images True Color – 3 bands False Color – 3 bands Pseudo Color – single band True Color image False Color NIR image 85 Color allocation for Color composite images True Color image Flase Color image 86 Color allocation for Color composite images Flase Color images 87 Color allocation for Color composite images Pseudo Color images Use 1 band of data Image data values 0-255 Color LUT can be greyscale or any Color palette Used mainly for classified images 88 Image Resolutions The size the sensor can resolve The ability of the sensor to differentiate variations in brightness The ability to detect changes over time The ability of the sensor to define fine wavelength intervals The four Remote Sensing resolutions that define the image data 89 Spatial resolution The fineness of spatial detail visible in an image. “Fine detail” means that small objects can be identified on an image. System’s spatial resolution is expressed in meters (or feet) of the ground-projected instantaneous field of view (IFOV) Finer spatial resolution → greater the resolving power of the sensor system 91 Spatial resolution 92 Spatial resolution and pixel size The spatial resolution and pixel size are often used interchangeably. In reality, they may not be equivalent. An image sampled at a small pixel size does not necessarily have a high resolution. Why? 240px by 240px 10 m resolution, 10 m pixel size 80px by 80px 30 m resolution, 10 m pixel size 30px by 30px 80 m resolution, 10 m pixel size 93 Spatial resolution vs extent However, there is trade-off between spatial resolution and spatial extent. The higher the spatial resolution, the smaller area it can be covered by one single image. Landsat-7 ETM+ 30m MODIS 250m to 1km 94 Temporal resolution How frequently a satellite observe the same area on the Earth. It is also known as Repeat Cycle. Depends on a variety of factors, including the satellite/sensor capabilities (as some satellites can point its sensor to specified area), orbits, the swath overlap, and latitude (higher latitude gives increasing overlap in adjacent swaths). high temporal resolution = better source for change detection. 95 Temporal resolution 96 Spectral resolution Spectral resolution describes the ability of a sensor to define fine wavelength intervals. Coarse – sensitive to large portion of EM spectrum contained in a small number of wide bands Fine – sensitive to same portion of EM spectrum but have many narrow bands Goal – Recording very fine spectral details to distinguish between scene objects and features Landsat-7 image contains 7 channels/bands. Landsat-8 image contains 11 channels/bands. Hyperion can resolve 220 spectral bands (from 0.4 to 2.5 µm) Airborne Visible / Infrared Imaging Spectrometer (AVIRIS): 224 contiguous spectral channels (bands) with wavelengths from 400 to 2500 nano-meters. 97 Multispectral vs Hyperspectral The main difference between multispectral and hyperspectral imaging is the number of wavebands being imaged and how narrow the bands are. Multispectral imagery generally refers to 3 to 10 discrete “broader” bands. Hyperspectral imagery consists of much narrower bands (10-20 nm). A hyperspectral image could have hundreds of thousands of bands. 98 Spatial and spectral resolution between Panchromatic and visible light bands Panchromatic image (B&W) records all of the visible portion of the electromagnetic spectrum which can achieve high Signal to Noise Ratio (SNR). Its spectral resolution is fairly coarse so that adequate signals can be acquired to use smaller detectors giving higher spatial resolution. Color film is also sensitive to the reflected energy over the visible portion of the spectrum, and has higher spectral resolution. It is individually sensitive to the reflected energy at the blue, green, and red wavelengths of the spectrum. However, it has lower spatial resolution. 99 Spectral reflectance An example showing the spectral reflectance of typical ground objects 100 Spectral reflectance Image Source: Sun et al., 2017 101 Spectral reflectance Comparison of spectra for alunite from four sensors with different spectral resolutions (Image Source: USGS) 102 Materials spectra collection Collection of field and laboratory measurements for the USGS Spectral Library (Image Source: USGS) 103 Materials spectra collection 𝑅e𝑓𝑙𝑒𝑐𝑡𝑎𝑛𝑐𝑒 = 𝑅𝑒𝑓𝑙𝑒𝑐𝑡𝑒𝑑 𝑟𝑎𝑑𝑖𝑎𝑛𝑐𝑒 𝐼𝑛𝑐𝑜𝑚𝑖𝑛𝑔 𝑟𝑎𝑑𝑖𝑎𝑛𝑐𝑒 White reference (left), measured reflected radiance of a white reference and target surface (middle), and calculated reflectance of the target surface (right) 104 Radiometric resolution The ability of the sensor to differentiate very slight variations in brightness/energy. The maximum number of brightness levels available depends on the number of bits used in representing the energy recorded. The finer the radiometric resolution of a sensor, the more sensitive it is to detect small differences in reflected or emitted energy. Coarse radiometric resolution: record a landscape using only a few brightness levels or few bits (i.e. at very high contrast) Fine radiometric resolution: record the same landscape using many levels of brightness or bits and thus more sensitive to detect small differences in reflected energy. 105 Radiometric resolution a b 1 bit – 2 levels 2 bits – 4 levels c d 3 bits – 8 levels 4 bits – 16 levels 106 Radiometric resolution 3 bits – 8 levels 2n-1 4 bits – 16 levels b c d 2 bits – 4 levels … a 2n bright levels 1 bit – 2 levels 0 107 Radiometric resolution (Landsat MSS) Multispectral Scanner System (Landsat TM) AVHRR 108 Radiometric resolution However, there is trade-off between radiometric resolution and spatial resolution/spectral resolution. Finer radiometric resolution requires sufficient signal strength and desirably high signal-to-noise ratio to give correct signal allowing discrimination of very slight energy differences. In general, high flux per unit area and signal correction is necessary. To give finer radiometric resolution, it can be done by reducing spatial resolution (larger GSD size) or by broadening the band incident upon a sensor. SNR is ratio of desired signal power to noise power, i.e. 109 Trade offs between image resolutions It is very difficult to obtain extremely high spectral, spatial, temporal, and radiometric resolution at the same time. 110 Trade offs – a short summary It is very difficult to obtain extremely high spectral, spatial, temporal, and radiometric resolution at the same time. Several sensors can obtain global coverage every one-two days because of their wide swath width meaning lower spatial resolution. Higher spatial resolution polar/non-polar may take 8-16 days to attain global coverage. Geostationary satellites obtain much more frequent observations but at lower spatial resolution due to the much greater orbital distance and only over a fraction of the earth. SNR can also be increased by broadening the wavebands and thus increasing radiometric quality. However, this also scarifies spectral resolution, i.e. the ability to discriminate fine spectral differences. 111 Questions Why it is very difficult to obtain extremely high spectral, spatial, temporal, and radiometric resolution at the same time? Try to explain in your own words. 112 Summary 1. Basic Specifications of Satellite and Sensor Sensors and Imaging Scanners Satellite Orbit Swath Width Repeat cycle and Revisit time FOV and IFOV 2. Introduction to the Satellites/Sensors Medium Resolution Satellites - Landsat Moderate Resolution Satellites - Terra (MODIS) Copernicus Satellites constellation - Sentinels High Resolution Satellites Small Satellites and Sensors 3. Characteristics of Remote Sensing Images Raster image Binary data Color Image Image resolutions 113 Discussion 114 Group Project Members and topic for the group project 115 Which kind of satellite images should I select for my research? Data availability (accessibility, time range, etc.) Scale issue (spatial/temporal resolutions) Others o Are there any existing products that meet the requirements? o What image processing methods/techniques should be used? o… 116 Homework Read the following satellite data user handbooks before the next lecture. ❑ Landsat 8 Data Users Handbook (Recommended) https://d9-wret.s3.us-west-2.amazonaws.com/assets/palladium/production/s3fspublic/atoms/files/LSDS-1574_L8_Data_Users_Handbook-v5.0.pdf ❑ MODIS Surface Reflectance User’s Guide https://modis-land.gsfc.nasa.gov/pdf/MOD09_UserGuide_v1.4.pdf ❑ Sentinel-2 User Handbook https://sentinels.copernicus.eu/documents/247904/685211/Sentinel2_User_Handbook.pdf/8869acdf-fd84-43ec-ae8c-3e80a436a16c?t=1438278087000 117 End of Lesson 2 118