ENGRG Introduction to GIS

Transcription

1 ENGRG 9910 Introduction to GIS Michael Piasecki March 27, 2014 Lecture 09: Introduction to Remote Sensing in GIS Slides by Dr. Jane Southworth, U of Florida & Dr. Tarendra Lakhankar, CCNY CREST What Is Remote Sensing? Minimal definition: remote sensing is the noncontact recording of information from the ultraviolet, visible, infrared, and microwave regions of the electromagnetic spectrum by means of instruments such as cameras, scanners, lasers, linear arrays, and/or area arrays located on platforms such as aircraft or spacecraft, and the analysis of acquired information by means of visual and digital image processing. 3/27/2014 ENGRG 9910 Intro to GIS 2 1

2 What Is Remote Sensing? Remote sensing is an applied science and is a tool to help understand other fields such as ecology, climatology, geology, soil science and hydrology. Why are we discussing RS within a GIS course?... because Remotely Sensed data has geospatial context and Its structure (raster) lends itself for geospatial computations and Interpretations. 3/27/2014 ENGRG 9910 Intro to GIS 3 What Is Not GIS? GIS (Geographic Information Systems) RS (Remote Sensing) GPS (Global Positioning System) RS and GPS data are often used within a GIS but are NOT a GIS 3/27/2014 ENGRG 9910 Intro to GIS 4 2

3 Advantages of remotely sensed images as spatial data sources Large area coverage Extended spectral range Geometric accuracy Permanent record 3/27/2014 ENGRG 9910 Intro to GIS Definition Remote sensing is the science and art of obtaining information about an object, area, or phenomenon through the analysis of data acquired by a device that is not in contact with the object, area, or phenomenon under investigation. 3/27/2014 ENGRG 9910 Intro to GIS 6 3

4 Acquisition 3/27/2014 ENGRG 9910 Intro to GIS 7 Radiation 3/27/2014 ENGRG 9910 Intro to GIS 8 4

5 Radiation Electromagnetic waves are radiated through space from some source. When the energy encounters an object, even a very tiny one like a molecule of air, one of three reactions occurs. The radiation will be: (1) reflected off the object, (2) absorbed by the object, or (3) transmitted through the object. 3/27/2014 ENGRG 9910 Intro to GIS 9 White Light and its Separation 3/27/2014 ENGRG 9910 Intro to GIS 10

6 Reflection & Transmission 3/27/2014 ENGRG 9910 Intro to GIS 11 Wavelength and Frequency 3/27/2014 ENGRG 9910 Intro to GIS 12 6

7 Bands Used for RS Microwaves are longest wavelengths used in remote sensing We are blind to everything except this narrow band UV are shortest wavelengths practical for remote sensing 3/27/2014 ENGRG 9910 Intro to GIS 13 Reflectivity and Albedo Albedo: reflective quality of a surface, expressed as percent of incident light reflected 3/27/2014 ENGRG 9910 Intro to GIS 14 7

8 Remote Sensing In remote sensing, we are largely concerned with REFLECTED RADIATION: we see colors, infrared film records vegetation, and radar images the earth. The source of a vast majority of this reflected radiation is the sun. 3/27/2014 ENGRG 9910 Intro to GIS 1 Sensors Different sensors measure different types of energy Human Eyes visible only (RGB) Photographic Systems visible, narrow regions of IR (near IR, R, G), and UV Radiometers / Scanners wide range, given sufficient energy 3/27/2014 ENGRG 9910 Intro to GIS 16 8

9 Different Remote Sensing Platform Aerial Photography Satellite Images 3/27/2014 ENGRG 9910 Intro to GIS 17 Aerial Photography The first known aerial photograph was obtained by Gaspard Felix Tournachon (Nadar) from a tethered balloon 1,700 ft. above Paris, France in /27/2014 ENGRG 9910 Intro to GIS 18 9

10 Aerial Photography In 1903, Julius Neubronner patented a breast mounted camera for carrier pigeons that weighed only 70 grams. A squadron of pigeons is equipped with light weight 70 mm aerial cameras. 3/27/2014 ENGRG 9910 Intro to GIS 19 Aerial Photography Oblique aerial photograph of a European castle obtained from a camera mounted on a carrier pigeon. The pigeon s wings are visible. 3/27/2014 ENGRG 9910 Intro to GIS 20 10

11 Aerial Photography Aerial photos form the basis of most large area maps and surveys in most countries, e.g., USGS 1:24,000 Must get the data into the GIS first: On screen digitizing Digitizing tablet Tracing features which are then digitized Scanning Buy digital data 3/27/2014 ENGRG 9910 Intro to GIS 21 Aerial Photography Why use aerial photos? Improved vantage point Defines situation at a point in time Permanent record Can record information beyond visible spectrum Increased spatial resolution and geometric fidelity Why use aerial photos instead of satellite images? Ability to choose your own dates Relatively inexpensive (?) Higher resolution (?) 3/27/2014 ENGRG 9910 Intro to GIS 22 11

12 Primary uses in GIS Surveying and topographic mapping: identify the horizontal and vertical positions of objects in a study area Categorize or assign attributes to surface features e.g., land cover maps Background (basemap) for other features e.g., soil survey maps, vector objects 3/27/2014 ENGRG 9910 Intro to GIS 23 Types of aerial photographs Orientation of photo Vertical Oblique high vs. low oblique Spectral characteristics of film Black and White (panchromatic) Black and White IR UV Color Color IR Scale of photo Large (e.g., 1:2000) or small (e.g., 1:24,000) Stereo pairs 3/27/2014 ENGRG 9910 Intro to GIS 24 12

13 Vertical Aerial Photography Camera film plane Vertical Aerial Photograph Over Level Terrain Altitude above-groundlevel (AGL) field of view Goosenecks of the San Juan River in Utah Optical axis Principal point (PP) 3/27/2014 ENGRG 9910 Intro to GIS 2 Low Oblique Aerial Photography Low-Oblique Aerial Photograph Over Flat Terrain field of view Optical axis Horizon is not shown in photograph Low-oblique photograph of a bridge on the Congaree River near Columbia, SC. 3/27/2014 ENGRG 9910 Intro to GIS 26 13

14 High Oblique Aerial Photography High-Oblique Aerial Photograph Over Flat Terrain High-oblique photograph of the grand Coulee Dam in Washington in 1940 field of view Optical axis Horizon is shown in the photograph 3/27/2014 ENGRG 9910 Intro to GIS 27 Panchromatic and Black & White Infrared 3/27/2014 ENGRG 9910 Intro to GIS 28 14

15 Stereoscopy Overlap about 1/3 3/27/2014 ENGRG 9910 Intro to GIS 29 Stereoscopic Two images overlapping (Stereo pair of prints) Using the stereoscope (simple instrument) Normal Vision 3/27/2014 ENGRG 9910 Intro to GIS 30 1

16 Stereoscopy Naked eye stereoscopic viewing: a learned skill, by controlling accommodation Cross your eyes until 4 images appear then allow images to converge to a set of three focusing at the center of the image 3/27/2014 ENGRG 9910 Intro to GIS 31 Photos can be taken using different aerial carriers! 3/27/2014 ENGRG 9910 Intro to GIS 32 16

17 Aerial Photo Interpretation Pre interpretation issues Classification System Anderson LULC (USGS): 9 classes, 92 next level Minimum Mapping Unit (MMU): needs & capability Digital vs. Analog system: Digital rules! Date and Type of Photo(s): IR, natural color, Features in photos Shape: straight lines, faults, floodplains Size: single vs. multiple family Pattern: row crops Shadow: length implies height Tone / Color: film type, chlorophyll Texture: tree size/age Association: riparian trees/streams 3/27/2014 ENGRG 9910 Intro to GIS 33 True color recording False color recording 3/27/2014 ENGRG 9910 Intro to GIS 34 17

18 Geometric Quality Many forms of distortion occur: Terrain Camera tilt Film deformation Camera lens Other camera errors Atmospheric refraction 3/27/2014 ENGRG 9910 Intro to GIS 3 Terrain and Tilt Distortion Relief displacement: real world on flat film plane Terrain distortions are radial (higher:outward ) Relief distortions affect angles and distances on an aerial photograph Scale is not constant on aerial photographs A vertical aerial photograph taken over varied terrain is not orthographic Different methods for geometric correction of aerial photographs (film/sensor, camera, lens ). Or fly higher, which leads us to 3/27/2014 ENGRG 9910 Intro to GIS 36 18

19 Satellite Images Advantages over Aerial Photos: very high perspective (>600 km) significantly reduces terrain distortion an extended spectral range very precise pointing direction can offer near perfect vertical orientation 3/27/2014 ENGRG 9910 Intro to GIS 37 Disadvantages over Aerial Photos Usually cover larger areas may pay for a lot of area you don t need (IKONOS, Quickbird changing this) Needs special image processing software Scheduled data collection and so less flexibility for dates (also not a long historical record) Effective resolution is better for many photographs than satellite images Aerial photos are often cheaper Differences are diminishing as more and better (higher resolution) satellite sensors are being launched. 3/27/2014 ENGRG 9910 Intro to GIS 38 19

20 Satellite Sensors Hundreds of different sensors are currently operational, providing many different image options. Selection of the sensor to use is a function of the questions being asked, necessary spectral, spatial and temporal resolution, budget issues, etc. Some of the major sensors are: Landsat TM SPOT GOES SeaWiFS ATLAS IKONOS Quickbir Landsat MSS NOAA AVHRR IRS Daedalus ASTER OrbView 3 MODIS Selection of the sensor to use is of MAJOR importance, and has implications for the research questions to be addressed and their results 3/27/2014 ENGRG 9910 Intro to GIS 39 Where is this? How did you know? IKONOS Pan band 1m resolution 3/27/2014 ENGRG 9910 Intro to GIS 40 20

21 Resolution & Scale In the past the availability of existing data at specific scales and resolutions has been limited. Now, increasing number of multi spectral datasets available. We need to select the correct data based on: spatial resolution spatial extent spectral resolution temporal resolution (return time) 3/27/2014 ENGRG 9910 Intro to GIS 41 SPOT Pan 10m Landsat TM 30m AVHRR 1.1km Entire State of Indiana IKONOS 4m Tree crowns 3/27/2014 ENGRG 9910 Intro to GIS 42 21

22 Nature of Satellite Image Data Raster data model Digital Numbers (DN), that are proportional to the reflectance, or Brightness Value (BV) in some subset (band) of the EM spectrum. Quantized grid of small areas on the Earth s surface. The energy of reflected electromagnetic radiation in each grid cell is a function of the characteristics of the objects in that cell. 7 Landsat Bands Individual Data Files /27/2014 ENGRG 9910 Intro to GIS 43 Data Availability and Acquisition Emphasis on Landsat TM, MSS; SPOT; AVHRR Landsat EOSAT (Space Imaging), USGS (EROS Data Center) SPOT SPOT Image Corp. AVHRR Advanced Very High Resolution Radiometer (Spectral, not Spatial Resolution); Satellite by National Oceanic and Atmospheric Administration (NOAA). Much data available from USGS EROS, other sources MODIS, ASTER, EOS Others (See Space Imaging Site: JERS Indian (IRS) etc. 3/27/2014 ENGRG 9910 Intro to GIS 44 22

23 3/27/2014 ENGRG 9910 Intro to GIS 4 3/27/2014 ENGRG 9910 Intro to GIS 46 23

24 Orbits and Swaths Geostationary orbits: Very high altitude satellites (approximately km) Focus on the same area of the Earth at all times Continual data collection over a specific area Eg. Weather and communications satellites Near polar orbits Satellite travels northward on one side of the Earth and then southwards during the second half of its orbit ( km) In sun synchronous orbits, ascending path can be on a shadowed side with the descending path on the sunlit side. Passive sensors would only record data during the descent. Path over same location at same time every day. 3/27/2014 ENGRG 9910 Intro to GIS 47 Finding Your Scene(s) Worldwide Reference Systems Path Row Notation The combination of a Path number and a Row number uniquely identifies a nominal scene center. The Path number is always given first, followed by the Row number. Ex. Florida The notation 17 39, for example, relates to Path number 17 and Row number 39. Orbits: Sun synchronous Polar 3/27/2014 ENGRG 9910 Intro to GIS 48 24

25 Scale/Resolution in Remote Sensing m 10 m 20 m 40 m 3/27/2014 ENGRG 9910 Intro to GIS 49 What Can Be Done with RS Data? Remote Sensing of Biophysical Variables. Location and Elevation/Bathymetry Temperature Soil Moisture Foliage Biochemistry Surface Roughness Vegetation Type e.g., land cover class Hydrological Variables e.g., evap., snow cover Atmospheric Variables e.g., wind speed, water vapor AND SO ON: What is your question about the surface of the Earth, or the atmosphere above it, or the geology beneath it? 3/27/2014 ENGRG 9910 Intro to GIS 0 2

26 Remote Sensing Platforms Have Been Designed to Sense Electromagnetic Energy in Various Bands in Order to Take Advantage of The Absorption /Transmission Properties of the Atmosphere Absorbance/Reflectance Spectrum 3/27/2014 ENGRG 9910 Intro to GIS 1 Transmission through Atmosphere 3/27/2014 ENGRG 9910 Intro to GIS 2 26

27 Landsat 4 and Thematic Mapper Band Applications: Color Band ( m) Application Blue Soil/vegetation discrimination, deciduous/coniferous forest differentiation, clear-water bathymetry Green Growth/vigor indication for vegetation, sediment estimation, turbid-water bathymetry Red Crop classification, ferric iron detection, ice and snow mapping Near Infrared (NIR) Biomass surveys, water-body delineation Short-wave IR Vegetation moisture, snow-cloud differentiation Short-wave IR Hydrothermal mapping, rock/soil type discrimination for mineral and petroleum geology Thermal IR Thermal mapping, plant stress, urban/non-urban land-use differentiation Source: Mika, A.M. Three decades of Landsat instruments. PE & RS. 43: /27/2014 ENGRG 9910 Intro to GIS 3 Reflectance spectra of vegetation Chlorophyll reflects higher Green and Near Infrared, but absorbs more Red Normalized Difference Vegetation Index NDVI is (IR R)/(IR+R); range is 1 to +1 NDVI of an actively photosynthesizing leaf is, e.g. (72 22)/(72+22) = 0.3 Colored lines approx. represent TM bands 1-4 Modified from Jensen, J Remote Sensing of the Environment. Prentice-Hall 3/27/2014 ENGRG 9910 Intro to GIS 4 27

28 Normalized Difference Vegetation Index, NDVI April September 3/27/2014 ENGRG 9910 Intro to GIS Spectra of Different Materials = 0. m = 0.8 m Fundamental Principle of Remote Sensing: For any given material, the amount of radiation that is reflected (absorbed, transmitted) varies with wavelength. 3/27/2014 ENGRG 9910 Intro to GIS 6 28

29 Spectra of Different Materials 3/27/2014 ENGRG 9910 Intro to GIS 7 Color Composites Landsat (R,G,B) /27/2014 ENGRG 9910 Intro to GIS 8 29

30 Pre Processing Essential Steps prior to analysis or Why you need a Remote Sensing class to do this stuff! Image rectification and restoration Rectification: correcting geometric distortions, putting data into real world coordinate system Restoration/Calibration: correcting radiometric errors, and eliminating noise in the data Image enhancement Applying algorithms to image data which allow the image to be more effectively displayed for visual interpretation. 3/27/2014 ENGRG 9910 Intro to GIS 9 Rectification (Geometric Correction) Source of distortions Earth curvature, earth rotation, atmospheric refraction, relief displacement, nonlinear sweep of sensor IFOV Two major types Systematic: well understood, easily corrected ex: Eastward rotation of earth as sensor sweeps (causes skew ) Typically corrected before users receive the data Random: Not well understood, use GCPs to correct GCPs are features which can be identified on both the image, and on a georeferenced data source (like a topographic map, for instance) Best features to use are road intersections Coordinate Transformation Least squares regression analysis is used to determine coefficients for transformation equations in X and Y directions 3/27/2014 ENGRG 9910 Intro to GIS 60 30

31 Georeferencing Image to Map Rectification 3/27/2014 ENGRG 9910 Intro to GIS 61 Radiometric Correction Factors that influence overall scene illumination: Atmospheric conditions Viewing geometry Instrument response characteristics Solar illumination corrections Sun elevation correction Accounts for the seasonal position of the sun relative to the earth Earth sun distance correction Accounts for the seasonal changes in the distance between the earth and sun Atmospheric corrections Haze compensation Minimize the influence of path radiance effects Platform specific corrections Analog to Digital response functions (rare to encounter) 3/27/2014 ENGRG 9910 Intro to GIS 62 31

32 Noise Removal Striping or banding Variations in responses of individual detectors used within each band Landsat MSS data has sixth line striping Solution: Create histograms separately for each detector in a given band. This can help you identify the faulty detector, and you can then adjust the grayscale of the problem lines. Line drop A number of adjacent pixels along a line may contain spurious DNs Solution: Average pixel values from above and below problem line(s) to fill in the problem areas Bit errors ( salt and pepper ) Random pixels can have non systematic variations Solution: Use moving window averaging scheme 3/27/2014 ENGRG 9910 Intro to GIS 63 SPOT image with striping (false color composite) Same image destriped 3/27/2014 ENGRG 9910 Intro to GIS 64 32

33 Classic Application: Land Cover Classification 3/27/2014 ENGRG 9910 Intro to GIS 6 Philosophy of Land Cover Classification When you look at the landscape, it looks heterogeneous. Humans want to categorize the heterogeneity, so they create classes. These classes may or may not have any biological reality in all cases they are convenient fictions. Biogeographers, botanists, zoologists, planners, and others have recognized different levels of organization of land cover. At all levels, land covers are defined by their dominant vegetation or human structures. 3/27/2014 ENGRG 9910 Intro to GIS 66 33

34 Two (3) Types of Classification 1. Supervised You define the land cover classes by indicating areas of particular land cover. 2. Unsupervised The computer defines the land cover classes by finding clusters of points in feature space. (3. Hybrid) 3/27/2014 ENGRG 9910 Intro to GIS 67 Correspondence of Classification System and Remote Sensing Data Scale of heterogeneity Resolution of Sensor Platform Ultimately judgement of analyst and ability of sensor platform to differentiate land cover types. Must match extent and grain of system to that of instrument, as appropriate for your questions. 3/27/2014 ENGRG 9910 Intro to GIS 68 34

35 Limitations & Potential Problems? Data storage issues will become more problematic as the volume of data for analysis continues to increase. Currently some of these data sources are prohibitively expensive e.g., AVIRIS, NASA plane to fly sensor = $60,000 + images = $90,000 (Booked for years in advance!). Researchers must ensure that they select a sensor with the appropriate scales: temporal, spatial, and spectral. 3/27/2014 ENGRG 9910 Intro to GIS 69 Remote Sensing and GIS For remote sensing to achieve all its promise, it must be fully integrated within GIS Ease of use Ease of analysis Increased manipulation power Ability to extract data simply and display it in an attractive and informative way. 3/27/2014 ENGRG 9910 Intro to GIS 70 3

36 Example: Environmental Modeling Satellite Image Coverages GIS Spatial Coverages 1:0, TM 1991 TM 1996 TM Slope Elevation Roads Towns Fieldwork 1990 s Change Grid Landscape Metrics of Change Spatially lagged NDVI Weighted accessibility index 2000 TM Initial Model Creation & Testing Change grids Model Validation & Calibration Fieldwork 2002 Model Prediction for 2000 Sensitivity Analysis & Hypothesis Testing Final Model & Predictions 3/27/2014 ENGRG 9910 Intro to GIS 71 Aerial Photos or Satellite Images: Which to use? Consider the following: Size of the study area Spectral range needed Digital format needed? Is there the necessary software and trained personnel for dealing with satellite images What level of spatial detail is needed: Photos very fine (centimeters) Images less so (Quickbird 0.64m at best) 3/27/2014 ENGRG 9910 Intro to GIS 72 36

37 Geology: Terrestrial Nevada surface materials map based on specific chemical bonds and AVIRIS hyperspectral imagery. Analysis performed using Tetracorder software (Trekkies feel free to laugh at this name). 3/27/2014 ENGRG 9910 Intro to GIS 73 Geology: Planetary Tharsis Volcanos on Mars; Sensors: MOLA and Viking 3/27/2014 ENGRG 9910 Intro to GIS 74 37

38 Atmospheric Science Hurricane Dennis, GOES-9 3/27/2014 ENGRG 9910 Intro to GIS 7 Hydrology Soil Equivalent Water Depth Snow Depth Daily Transpiration Surface Water Depth SPLASH hydrological data, LANDSAT inputs 3/27/2014 ENGRG 9910 Intro to GIS 76 38

39 Soil Science 3/27/2014 ENGRG 9910 Intro to GIS 77 Agriculture Gallo Vineyards, Sonoma County, ADAR-00 3/27/2014 ENGRG 9910 Intro to GIS 78 39

40 Ecology Theodore Roosevelt National Park, North Dakota, Leafy Spurge Mapping 3/27/2014 ENGRG 9910 Intro to GIS 79 Change Detection Yasuni National Park, Ecuador, LANDSAT /27/2014 ENGRG 9910 Intro to GIS 80 40

41 Interpret these 2 images. While I am interested in what you see, I am more interested in how you know what the different features are. What skills are you using etc. Source: Digital Globe 3/27/2014 ENGRG 9910 Intro to GIS 81 Let s interpret this one. Again not only what is it, but how do you know what it is? What skills do you employ? Also, what is different about this image? Source: Digital Globe 3/27/2014 ENGRG 9910 Intro to GIS 82 41

42 And finally.interpret this. Again, how does it differ? How do you interpret features etc. 3/27/2014 ENGRG 9910 Intro to GIS 83 So now you are an image analyst! If you are to use remotely sensed data aerial photography or satellite data you MUST take the necessary courses in it It s a great data source and lots of fun! ENGRG Intro to RS or EAS Environmental Remote Sensing and Image Analysis 3/27/2014 ENGRG 9910 Intro to GIS 84 42