Designing a Remote Sensing Project. Many factors to consider: here lumped into 12 sections hold on!! first some basic concepts

Transcription

1 Designing a Remote Sensing Project Many factors to consider: here lumped into 12 sections hold on!! first some basic concepts DVC Geography 160 Introduction to Remote Sensing J. Ellis DigitalGlobe (2006) Ellis GeoSpatial 2013

2 Spatial Resolution

3 IKONOS, Quickbird, Worldview-2, GeoEYE, Pleiades Geog 160 J. Ellis Ellis GeoSpatial 2013 Ellis GeoSpatial

4 MORE WAVELENGTHS BANDS AND/OR SMALLER PIXELS = $$$ Choose the optimum sensor for the job and the budget Maximum Spatial x, y, z Accuracy: Orthorectification of image with DEM & ground control points Maximum Spectral: Hyperspectral Thermal Sensors High-Resolution Satellites Multispectral Landsat TM Multispectral

5 Airborne Sensors

6 Satellite Sensors DVC Geography 160 Introduction to Remote Sensing J. Ellis DigitalGlobe (2006) Ellis GeoSpatial 2013

7 You must have a clear understanding of at least the following...: 1. What are the project's objectives? 2. What is the mapping scale of the project? 3. What are the cartographic requirements? 4. What is the required level of spatial resolution? 5. What is the portion of the wavelength spectrum needed and the required spectral resolution? 6. How big is the area?

8 Get a clear understanding before starting. 7. Who is the customer? 8. What is budget for the project? 9. How much time is there for the project? 10. What are your computer capabilities?? 11. What products are required for deliverables? 12. Who's been doing similar work (for assistance)?

9 1. What are the project's objectives? Each project has its own objectives. Take time to discuss. Show other examples case histories to help demonstrate expected results Be realistic & honest often remote sensing (by itself) cannot achieve the objectives! Is the client adverse to risk? Not wanting to try out new technology? Can other groups benefit and share costs? Short-term project-oriented or Long-term institutional (Fed, State, County, City )?

10 2. What is the mapping scale of the project? The mapping scale required by the project is often the most important factor determining the costs. Enlarge scale by 2X, increase area to map & plot by 4X - Going from 1:24,000 to 1:2400 results in 100X increase in plot size Limitations apparent when working with paper copies that are at their original mapping scale are often ignored in the Digital GIS world. Digital zooming leads to inappropriate use of original data Costs skyrocket if heads-up mapping Example: Project Scale 1:100,000 therefore, a 100-meter feature on earth = 1 mm on plot Let s say we are mapping 30-m pixels from Landsat TM Interpreter zooms-in to ~1:30,000 on screen to carefully digitize features 1 cm across (representing 300 m on ground). When plotted at 1:100,000, features only 3 mm across Interpretation may little value

11 3. What are the cartographic requirements? It is essential to understand the project's cartography requirements. It can be very confusing! Datum: For the U.S.A., the North American Datum (NAD-27 and NAD-83 and WGS-84 are standards... Projection: State Plane or UTM? For state-wide California images & maps, Albers Equal Area (Conic projection) used. GIS software may not be able to convert on-the-fly while displaying Images & maps of different datums & projections check the manual Conversion to one datum/projection and defining datum/projection for each GIS layer helps ensure your GIS stack will work with different software. How accurate do your GIS layers have to be? Horizontal & Vertical?

12 3. What are the cartographic requirements? There are different methods for spatially registering images (including scanned aerial photographs) to map projections (see Manzer, 1995). It is essential that the remote sensing team state the method used and the level of accuracy achieved. Scaling. The simplest and least accurate is termed the "scaled" image or photograph. With this procedure, images are "adjusted" to only 2 control points. There is no correction for terrain distortions, lens distortion, or aircraft tip and tilt (or satellite off-nadir viewing). image map

13 3. What are the cartographic requirements? Georeferencing This is better than scaling, because several control points are utilized and the image is "adjusted" or "warped" to fit these points. Adjustment is typically 1 st,, 2 nd, or 3 rd order. The computer program provides feedback to the operator on the relative error for each order of fit. This can be documented in metadata (rms & # GCPs). A polynomial transformation equation is generated that warps the image pixels to the control points. Georeferencing is the most common method of rectifying scanned maps and imagery to a map projection. There is no correction for terrain distortion.

14 3. What are the cartographic requirements? Digital orthorectification This provides the most accurate method of rectifying images to a map projection. The digital imagery are rectified to an orthographic projection (a dense network of elevation points or a digital terrain model - DTM) on a pixel-by-pixel basis (Manzer, 1995). A DEM/DTM is required (can be from different sources). A consistent scale and accuracy is achieved with digital orthorectification. Photogrammetric mapping technology is employed with this method used for both satellite & airborne imagery. The costs and resulting accuracies associated with orthorectification can be much greater than with georeferencing.

15 4. What is the required level of spatial resolution? Interview clients to determine what size objects and what type of features that they want to see on the imagery and want interpreted. Spatial resolution can be defined as the ability to distinguish between two closely spaced objects on an image or, more specifically, it is the minimum distance between two objects at which the images of the objects appear distinct and separate (Sabins, 1997). Sabins notes that spatial resolution is different for objects of different shape, size, arrangement, and contrast ratio. In addition, he notes that resolution alone does not determine whether an image is suitable for a particular application. Examples bright roads (narrower than width of pixels) against dark background, low contrast issues There are standard photogrammetric targets that are used for measuring a sensor s spatial resolution.

16 5. What is the portion of the wavelength spectrum needed and the required spectral resolution? Remote sensing is accomplished by sensors recording energy levels received from portions of the electromagnetic spectrum (Colwell, 1983; Sabins, 1997). Common names for those portions used by remote sensing are reflected visible light, reflected infrared (composed of near-infrared light (near IR), short wavelength IR (SWIR)), emitted thermal, and microwave (radar). Visible-near IR sensors are VNIR. What s VNIR-SWIR? In order to penetrate clear water and interpret bathymetry, a sensor that records visible light in the shorter wavelengths (blues and greens) is optimum (Harris and Kowalik, 1995). If the objective is to map surface deposits characterized by iron oxide stains (red soils or a reddish marker bed), then a sensor that records visible red light would be most effective.

17 To best delineate the boundary between water and dry land (for mapping reservoirs, flood extent, swamps, etc.), a sensor that records light in the reflected IR is excellent because water totally absorbs this IR energy, resulting in no energy being reflected back to the sensor. A very dark signature is recorded as compared with the adjacent dry land. To detect offshore oil seeps a combination of visible/reflected IR imagery and radar may be most applicable. For onshore oil detection, SWIR hyperspectral is required. For mapping hot spots associated with wildfires, Thermal band (3-5 microns) is optimum it can record heat through the smoke cover. Visible light, especially blue & green, is scattered by atmospheric haze, water droplets, & dust, resulting in poor imagery. Near IR and SWIR sensors record much sharper images in these conditions.

18 Where the project area is perpetually covered by clouds, dust, and/or smoke, or nighttime acquisitions are needed, then microwave sensors (radar) are required. For passive satellites (such as Landsat) and aircraft sensors, imagery acquired during the winter months when the sun is lowest are optimum to accentuate natural shadowing, improving detection of subtle topography, lineations (fracture studies) and geologic structure. For many engineering projects, high spatial resolution, black and white (B&W) imagery is preferred as it is sharper. To minimize shadows, aerial imagery is collected close to the local Noon Hour (+/- 2 hours) often specified in contract. Sun-synchronous, multispectral satellites collect ~10AM in the morning.

19 LANDSAT TM SPECTRAL BANDS* BAND WAVELENGTH CHARACTERISTICS (micrometers) TO 0.52 Visible blue-green. Optimum for bathymetric mapping in clear water. Sensitive to haze, water vapor, etc to 0.60 Visible green. Displayed in blue on 7,4,2 color images. Matches green reflectance peak of vegetation to 0.69 Visible red. Useful for identifying iron oxide in rocks and soils to 0.90 Reflected Near-Infrared. Displayed in green on 7,4,2 color images. Coincides with reflectance maximum of healthy vegetation to 1.75 Reflected Shortwave Infrared (SWIR). Indicates moisture content of soil and vegetation. Penetrates thin clouds. Useful for mineral identification to Thermal Infrared. Records radiant temperature. Has 120 m pixels to 2.35 Reflected Shortwave Infrared (SWIR). Displayed in red on 7,4,2 color images. Important rockforming minerals (clays, carbonates, and others) have distinctive absorption features in this band *taken from Sabins, 1997

20 6. How big is the area? Cost driver size of pixels or ground sampling distance (gsd) critical across the area. For remote sensing, costs = # of pixels to crunch. Determine if there are areas of higher priority or areas where more detail is needed. Use regional satellite sensors to provide uniform coverage of the entire study area and higher-resolution satellite or airborne sensors for the priority areas. Using images from different dates to cover a large area causes marked increase in costs, time required for color balancing, and issues with vegetation changes & mapping. Geomorphology, topography, and vegetation cover can significantly change across a large area, making color balancing difficult.

21 7. Who is the customer? Expectations vary considerably depending on who your customer is. Is it a consulting firm that needs images and maps yesterday, a county government with strict guidelines for metadata, a lawyer concerned with chain of custody and documenting processing steps, your researchoriented Professor, or a non-profit (NGO)? Digital remote sensing is an accepted planning, monitoring, environmental baseline, logistics tool - not a parcel, legal, taxation, or engineering tool. yet For some projects, the image is all important. For others, the derived product is critical. Project-driven customers tend to have looser specification & are not too interested in paying for documentation (metadata, attributes, etc.).

22 8. What is budget for the project? Projects that are able to utilize old data cost less. Government data are the least expensive while commercial data are the most expensive. Often data costs can exceed labor costs for new imagery. The cost of software, hardware, and plotting can be minor items in the budget. Try to document different products and services as separate line items in the budget. - Protects you, - Makes changes easier to manage, and - Helps you determine which tasks are profitable and which ones are losers

23 9. How much time is there for the project? New acquisitions can take time. Cost uplift for priority service. Internet downloads of archived data really fast! jumpstarts projects. Many phases to a project they all take TIME (and money ) Scoping Data searches Ordering Delivery of Data Loading & QC, formatting Processing Attributes & metadata? Plots? Presentation Products? Delivery & Billing Archiving

24 10. What are your computer capabilities?? (hardware, software, and operator skill level)...be honest! Be honest Partner with others to expand your capabilities Building a solid foundation for remote sensing Image processing software GIS software Middle-range XP workstation Maximize RAM, good graphics card, dual monitors removable drives, High-speed Internet Access Desktop scanner Microsoft Office for Small Business (need Powerpoint) Adobe Acrobat (to make your own.pdfs) Digital camera for field work

25 11. What products are required for deliverables? Can be budget-breaker, if not planned well At the beginning of a project, it is important to understand what the requirements (really expectations) are for the final products. Are deliverable to be digital and/or hardcopy? How many copies? In what formats, scales, and media are these products to be delivered? CAD and/or GIS format with attributes and metadata? Set expectations upfront with examples

26 12. Who's been doing similar work (for assistance)? #1 Get Involved with remote sensing & GIS professionals (people!) Attend local association meetings Attend local software user groups Subscribe to trade journals & online news Attend a local college and take a course Diablo Valley College has a GIS-GPS Certificate Program Six core courses Intro & Advanced GIS Intro & Advanced GPS Remote Sensing Cartography Attend conferences & workshops Join Professional Societies

27 Overlap on next slide

28 Overlap on previous slide