Nirmal Chaudhary

25/08/2011

Satellite Imagery and Digital Image Processing

Course content: (V) introduction, satellite navigation system, types of RS satellite, introduction to photo interpretation and photogrammetry, satellite imagery and digital image processing. (Vi) radiometric and geometric corrections, image enhancement and interpretation, spatial filtering, feature extraction, classification methods, maximum likelihood classifier. (vii) GIS/RS applications: for assessment, monitoring and conservation of biodiversity. Practical assignment will follow after each lecture.

Nirmal Chaudhary, M.Sc.

Satellite imagery and digital image processing: Introduction

The first photograph: taken in 1839

Nirmal Chaudhary

25/08/2011

Remote Sensing: The use of Electronic Radiation Sensors to record images of the environment, which can be interpreted to yield useful information (Curran, 1985)

Processing

Sensor

Object Physical Components of RS: 1. Source 2. Atmospheric interaction 3. Interaction with object 4. Sensor 5. Processor

1. The Source: Target illumination

1. The Source: Radiation and Emissivity

All energy above 0 0K emits radiation due to molecular agitation. Not only the sun but the Earth too, emits energy.

nm= 10-9 m m= 10-6 m cm= 10-2 m Electro-Magnetic Radiation = (Wave model of light ) Q= h = h (c/ ) (Particle model of light) C= velocity of light ( 3 X 10 8 m/s) = Wavelength (m) = frequency (cycles per second; Hz) h= Plancks constant (6.6262 X 10-34 Js Energy radiated by any body depends upon its absolute temperature and emissivity and is function of wavelengths: Stefan-Boltzmann s law

Nirmal Chaudhary

25/08/2011

1. The Source: EM Spectrum

Primary Colours

2. Atmospheric Interaction
Particles and Gases of Atmosphere affects the radiation via absorption and scattering. Scattering by molecules of size smaller than the wavelength of EM waves cause Rayleigh scattering. O2, NO2, CO2 etc are the cause. These cause smaller wavelengths of light to scatter more than longer wavelengths. Sky is blue in a clear midday. Scattering by aerosols with size larger than EM waves is Mie Scattering. Water drops, haze, dusts, Pollens etc are the cause. Non-selective scattering are caused by much larger particles than wavelength. This distribution of all the wavelength is nearly equally scattered. E.g. clouds. Ozone, Water Vapours, CO2 absorb radiation of particular wavelengths.

2. Atmospheric Interaction

3. Interaction with the Target

1. Radiation that is left after atmospheric interaction reaches the Earth skin. Surface roughness determines the reflection: Specular (smooth surface) and Diffuse (Irregular surface) Reflection. 2. Incident energy on target interact in one or more ways: Absorption, Transmission and Reflection: Reflection from targets are of particular interest, in Remote sensing. 3. Reflectance is ratio of the incident energy flux to the reflected energy flux of a surface. It range from 0 - 1. Spectrometer measures the Reflectance. 4. Reflectance with respect to wavelength are called spectral reflectance.

Atmospheric Transmission windows are the region of the spectrum which are not influenced by the atmospheric absorbents. For remote sensing the windows in range of 0.35 to 2.5 m in visible and reflected IR region (Optical region) , narrow windows in range of 3 to 5 m and a broad window from 8 to 14 are of interest.

Nirmal Chaudhary

25/08/2011

Question hour:
1. What are the advantage/disadvantage of air-borne (aerial) remote sensing and space-borne remote sensing? 2. What is electromagnetic spectrum? 3. Why clouds appear white? 4. Define atmospheric window.

Satellite Navigation Systems

5. What time of the year, remote sensing for forest area of Nepal will be most suitable and why?

Satellite Navigation Systems

1. Definition: System of artificial satellites for positioning, navigation and timing (PNT technology). 2. Features: a. High position accuracy from m to mm b. Capability of determining velocity and time in relation to the accuracy of position (PVT). c. Position location is in 3D: Latitude, Longitude and Altitude. d. Available 24 hours a day and all weather system e. Wide application: Environmental, Cadastral, Engineering, GIS, Planning, Navigation-land, sea, air 3. Global Navigation Satellite System (GNSS) is an umbrella term for all the present and future global, satellite based, radio-navigation system. a. Global Positioning System (GPS) of US Department of Defence b. GLONASS of Russian Federation c. GALILEO of European Union:2013 d. COMPASS /Beidou of China: 2020 e. Indian Regional Navigation Satellite System (IRNSS): 2014

Components of GPS technology

1. Space segment: Satellites and transmitted signals 2. Control Segment: ground facility controlling satellite tracking, orbit computation, satellite clock behaviour, system monitoring etc. 3. User segment: application, equipment and computational techniques for users

1. At an altitude of 20,200 km, 24 satellites located in 6 orbital planes inclined at 630 to the equator is sufficient to ensure that there will be at least 4 satellites visible, anytime anywhere on the earth 2. At present there are 31 GPS satellites since 1978. 3. The orbital period is 11 h 58 min so that each satellite make 2 revolution in one sidereal day. In solar day, the satellites will be in same position in sky about 4 min earlier each day.

Nirmal Chaudhary

25/08/2011

Basic of GPS
GPS satellite signals have following components:
Carrier waves: L1 (1575,42 MHz) and L2(1227.60MHz) Ranging codes in the carrier waves using Code division multiple access (CDMA) scheme: determines the transit time of signal. Navigation message: orbital information (ephemeris), satellite clock error parameter etc (Almanac)

Remote Sensing Satellites

Satellites may be Polar or Geo-stationary. Polar satellites travel north eastward at angle between 800 to 1000. They are placed at 600-1000 km from earth and most of them are sun-synchronous-meaning they pass overhead at the same time. Most of them pass the equator at 10:30 hour local solar time and may take 2 weeks to repeat scan of the same location. Landsat, SPOT and IRS are examples. Sun-Synchronous satellites have polar orbit with inclination angle between 980-990. Geo-stationary satellites are fixed at 3600 km above earth and are always at same position on earth. They are generally meteorological and communication purpose. Based on sensors on satellite, it could be Active or Passive. Passive sensor depends on energy from sun or from earth itself. After reflection from the target, this energy is recorded by sensors on platform. E.g. Photographic camera. Active sensors emit their own energy to the target and receives the reflected energy. Radio detection and Ranging (radar) and light detection and ranging (Lidar), Sound navigation ranging (SONAR) are the passive sensors.

The transit time when multiplied by speed of light, gives the distance of satellite and the receiver. With 4 satellites, the distance of the receiver is in 3D on Trilateration basis, because the accuracy will be high with at least 4 satellites. Errors: Ionosphere and Troposphere delay Orbital and Transit time error Less satellites and position Signal multipath

Fundamental sensor types are Across-track scanning sensor (Whiskbroom scanner) and Along-track scanning sensor (Pushbroom scanner). Whiskbroom scans the earth surface perpendicular to the direction of satellite movement. Eg. NOAA/AVHRR, Landsat/TM. Pushbroom scans parallel to the direction of motion. Eg. SPOT-1, IKONOS

Remote Sensing Satellites:

Meteosat-8: Owned by: European organization: Eumetsat. Sensor: SEVIRI (Spinning Enhanced VIS and IR Imager) Orbit: Geo-stationary, 00 longitude Swath: Full earth disc (FOV=180) Revisit time: 15 mins Spectral band (m): o.5-09 (PAN), 0.6, 0.8 (VIS), 1.6, 3.9 (IR), 6.2,7.3 (IR for WV), 8.7,9.7,10.8,12.0,13.4 (TIR) Ground pixel: 1 km (PAN), 3 km (all other band) Data Archive: www.eumetsat.de NOAA -17: National Oceanic and Atmospheric Administration
(When the sun acts up, NOAA knows why.)

Owned by: US Department of Commerce. Sensor: Advanced Very High Resolution Radiometer (AVHRR-3) Orbit: 812 km, Sun-Synchronous, 98.70 inclination Swath: 2800 km (FOV=1100) Revisit time; 2-14 times /day Spectral bands (m): 0.58-0.68 (1), 0.73-1.00(2),1.58-1.64(3A day),3.553.93 (3B night), 10.3-11.3 (4), 11.5-12.5 (5)

Spatial resolution: 1 km (at Nadir) , 6 kmX2 km (at limb) Data Archive: www.sa.noaa.gov

Nirmal Chaudhary

25/08/2011

Landsat-5 and 7: NASA and US Geological Survey (USGS) Sensor: TM and ETM+ Orbit: 705 km, Sun-Synchronous, 98.20 inclination Swath: 185 km (FOV=150) Revisit time; 16 days Spectral bands (m): 0.45-0.52 (1), 0.52-0.6 (2),0.63-0.69 (3),0.760.90 (4), 1.55-1.75 (5), 10.4-12.5 (6), 2.08-2.34(7),0.50-0.90 (PAN) Spatial resolution: 15m (PAN), 30m (band 1-5,7), 60m (band 6) Data Archive: http://landsat.gsfc.nasa.gov/images/ Envisat: Environmental Satellite; European Space Agency Sensors: Advanced SAR(ASAR), Medium Resolution Imaging Spectrometer

Resourcesat-1: Indian Space Research Agency Sensor: Linear Imaging Self-Scanning Sensor (LISS-4) Orbit: 817 km, Sun-Synchronous, 98.80 inclination Swath: 185 km (FOV=150) Revisit time: 24 days Spectral bands (m): 0.52- 0.59, 0.62- 0.68 , 0.77- 0.86 Spatial resolution: 6m Data Archive: http://landsat.gsfc.nasa.gov/images/ TerraSAR-X: German Aerospace Center and EADS Astrium GmbH Source: 965GHz. Sensor: X-band SAR antenna Orbit: 514 km sun synchronous, 97.440 Swath: width (10km 100km) length (5 km to 1650 km) Revisit time: 2.5 days at maximum. CALIPSO: Cloud-Aerosol Lidar with Orthogonal Polarization Owner: NASA (US) and Centre National dEtudes Spatiales (CNES, France) Source: 532 nm and 1064 nm with pulse repetition of 20.16 Hz.
Spatial Resolution: 125m, 333m,1km,5km, Swath: 1km X 70 km

(MERIS), GOMOS (Global Ozone Monitoring by Occultation of Stars), RA -2 (Radar Altimeter),MWR (Microwave Radiometer), MIPAS(Michelson Interferometer for Passive Atmospheric Sounding) along with other complementary instruments.

Orbit: 800 km, Sun-Synchronous, 98.60 inclination Swath: 56 to 405 km (ASAR), 1150 km (MERIS) Revisit time; 35 days Spectral bands (m): C-band (ASAR), 15 bands in 0.39-1.04 m EM Spectrum (MERIS; programmable) Spatial resolution: 30m or 150m (ASAR), 300m land and 1200m Ocean (MERIS) Data Archive: http://earth.esa.int/dataproducts/

Introduction to Photo interpretation and photogrammetry

Photo Interpretation involves: Photo Reading : to recognize features such as trees, rivers in a photo Photo measuring (Photo-grammetry) :crown diameter? Altitude? Distance? Information Extraction: Status of tree/forest/land/soil ? What is Photo-Interpretation? It is an act of examining photographic images for the purpose of identifying objects and judging their significance. ( Colwell, 1997) Photogrammetry is the science or art of obtaining reliable measurement by means of photographs (McGlone et al., 2004) Photogrammetry and photo-interpretation are more common words for Aerial photographs than Satellite imageries. Nowadays, the advances of satellite remote sensing has blurred the distinction among them.

Nirmal Chaudhary

25/08/2011

Image or Photo Interpretation is meant for information extraction. Information extraction can be Human method (Image interpretation) and Computer method (Digital Image processing)

Photogrammetry is concentrated to modelling of 3D images from 2D images using Stereoscopic viewing.

Aerial survey planning:

Scale: is the ratio of distance between two points in photo/image to the distance between same points on ground. This ratio is also called Representative Fraction (RF). distance in photograph RF = distance on ground If the distance width of a road is 0.2 mm on map and its width on ground is 5m, calculate the scale of the map. Soln = RF = 0.2 /(5 X 1000 )= 1/25000 = 1:25000 Problem: If the distance between two trees on a map is 0.2 inch, which has scale of 1:2400 then the distance on ground will be ? 0.2 inch = 0.2X2.4= 0.48 cm = 0.0048 m distance on ground = distance in photograph /RF = 0.0048 X (1/2400) = 0.0048 X 2400 =11.52 m RF can be also calculated as ratio of focal length of camera to the height of the camera from ground. Specification: Scale: 1:2400 Size of Photograph: 23X23 cm (9 X 9 inch) Overlap along line of flight (endlap): 60% Overlap across adjacent lines of flight (sidelap) : 20% Length of Area along lines: 30 km Width of Area across lines of flight: 20 km

Nirmal Chaudhary

25/08/2011

Kinds of photographs: Vertical: camera axis vertical 900: Nadir view Oblique: Off Nadir view

For any point (pixel) in at least two images we calculate the 3rd dimension (terrain coordinates) and this is the principle task in Photogrammetry. We need all the geometric parameters in the situations of when we take photos/images. Using the geometric parameters, we set the equation of rays [P P] and [P P] and calculating their intersection. Once the 3D coordinates are known, we can digitize maps, calculate distance, volume, slopes etc in Photogrammetry.

Films and Filters Photogrammetry can be considered as a traditional approach to remote sensing. Because, all the calculation process achieved from photogrammetry can be easily developed as algorithm in computers and are available in remote sensing/GIS softwares.

Satellite imageries and digital image processing

Satellite images are digital images of electro-magnetic energy received by sensors on satellite. A digital image has individual cells of uniform value. These individual cells are called picture cell (Pixel) and are generally square in shape. Digital image has coordinates of pixel number (sample) counted from left to right and line number counted from top to bottom.

Satellite imageries record reflected rays form target across several wavelengths in UV-Radio waves, hence are called: Multi-Channel, Multi-band or Multi-Spectral data.

Nirmal Chaudhary

25/08/2011

Multi-band images are stored/represented by combination of spatial position (pixel number and line number) Satellite images may be in BSQ, BIL or BIP format. Band squential (BSQ) the spatial position for each band is separately arranged Band interleaved by line (BIL) data are arranged in the order of band number and repeated with respect to line number Band interleaved by pixel (BIP) stores data with first line of first band, first line of second band and so on.

Resolution of Image:
Spatial Resolution (objects on ground) Spectral Resolution (portion of EMR) Radiometric Resolution (level of signal) Temporal Resolution (Sensor re-visit time) Spatial Resolution is the smallest size of the object that can be picked from the ground. It is mainly determined by the Instantaneous field of view (IFOV) and the height of sensor/camera from ground. Diameter on ground (D) = Height (H) X IFOV (Radians; )

Auxiliary data are supplementary to image data that describe image file, sensor, processing method etc.

Spectral Resolution is the number and dimension of specific wavelength interval of EMR that a sensor is capable of measuring. Smaller the width of band, higher is the spectral resolution. Resolution A < Res. C < Res. B Panchromatic image consist of large width band often entire Visual Spectrum. Multispectral image have relatively large narrower several bands. Hyperspectral image has large number of narrower bands.

Radiometric Resolution is the number of unique values the sensor can record for sensed EM waves. It is expressed in bits. Bits are expressed as power of 2 and number starts from 0. 1 bit = 21= 2: sensor can record 0 and 1 2 bit =22 = 4: sensor can record 0,1,2 and 3 8 bit = 28 = 256: sensor records 0,1,2,.255

Landsat TM Spectral resolutions

Nirmal Chaudhary

25/08/2011

Temporal resolution: Shortest period of time the sensor will pass over a same spot on the earth surface. Landsat makes 14 orbital revolution in Landsat revisits the same spot in 16 days, a day to cover whole earth. thus has temporal resolution of 16 days.

Digital Image processing

AVHRR on NOAA satellite has 12 hrs of temporal resolution. You cannot have all the resolutions higher, in same image. High Spatial Resolution= Low IFOV= Low energy sensed= Less value= Low Radiometry

Radiometric calibration/correction
There are chances the sensors recording of reflected EM energy from target does not coincide with the energy reflected form the object due to path disturbances (atmospheric, sun position and elevation, fog, haze etc). This gives radiometric distortion in image. To obtain real irradiance or reflectance, radiometric calibration is needed. Out of several methods for radiometric correction, Dark Object subtraction is one of them. It assumes that dark objects on earth surface does not reflect energy thus any value of the pixel from sensor that contains the dark object is due to atmospheric scattering. This value should be subtracted from the entire pixels of the image to get radiometrically correct image. Spectral radiance at sensor (L) for 8 bit image is obtained as:

Geometric calibration/correction
Geometric correction or Georeferencing: The process of transforming the x-y dimension of a image so that it has the same scale and project properties of a selected map projection. Ground Control Point (GCP) are needed to process the matching of points on an image with corresponding map (image) co-ordnates from which GCP is obtained. A good number of GCPs is 20 at minimum, distributed homogeneously over the image extent.

The two data sets are then used for coordinate transformation matrix, one using linear transformation. The result of Georeferencing is to produce new output grids aligned with northing and easting of reference map/image. The values of the newly acquired pixels of the output image is determined by sampling: Nearest Neighbour, Bi-linear, Cubic convolution.

10