CHARACTERISTICS OF REMOTELY SENSED IMAGERY. Radiometric Resolution

Transcription

1 CHARACTERISTICS OF REMOTELY SENSED IMAGERY Radiometric Resolution

2 There are a number of ways in which images can differ. One set of important differences relate to the various resolutions that images express. Resolution: the fineness of detail that can be distinguished in an image. Detail, however, can refer to several different dimensions. RESOLUTION?

3 Spatial Information Imagers Altimeters Sounders Temporal Information (Platformdependent) Spectral Information Spectrometers Polarimeters Scatterometers Radiometers SpectroRadiometers Intensity Information ELEMENTS OF RESOLUTION

4 HOW CAN IMAGES DIFFER?

5 Air photos: characteristics of emulsion of film (related to grain size). Digital images: number of data storage units, or bits, that the sensor has available to capture the range of energy or brightness levels. RADIOMETRIC RESOLUTION DEFINED

6 At the heart of any computer program or digital remote sensing system is the binary system (bits) that represents numeric values using two symbols, 0 or 1. In computing, word is the term for the natural unit of data used by a particular computer design. A word is simply a fixed-sized group of bits that are handled as a unit by the system. Common word sizes are 8 bits, 16, 32 or 64 bits. Landsat satellites Quickbird BINARY? WORD?

7 SPECTRAL REFLECTANCE CURVES

8 RR is a measurement of how sensitive a sensor is to differences in energy (photons), or radiance values (reflectance). > the sensitivity = the finer/higher the radiometric resolution. Increasing radiometric resolution can improve the discrimination of features that have similar spectral responses, but slightly different reflectance levels. Resolution is a function of the analog-to-digital conversion. We also need to consider the sensor s Signal to Noise ratio. RADIOMETRIC RESOLUTION

9 A satellite sensor measures radiance at the sensor itself, not surface reflectance from the target. In other words, the sensor is only measuring the photon intensity when they hit the detector surface. If satellite instruments could measure surface reflectance, the task of interpreting images would be a lot easier, and classified land cover products would be much more accurate. SPECTRAL REFLECTANCE?

10 Surface reflectance is the ratio of the intensity of light reflected from a surface over the intensity of incident light. Unfortunately, the orientation of the surface feature (usually due to slope and aspect), bidirectional reflectance, and atmospheric scattering and absorption complicate our ability to accurately measure surface reflectance. SPECTRAL REFLECTANCE?

11 TOA VERSUS ATMOSPHERICALLY- CORRECTED SPECTRUMS.

12 So, although we refer to spectral reflectance curves, they are primarily pedagogical tools since the satellite sensor isn t actually recording the spectral reflectance of the object per se. We hope / anticipate that the radiance values recorded by the sensor act as useful surrogates for the actual reflectance values. SPECTRAL REFLECTANCE CURVES

13 Output [0 255] Output [0,1,2,3] Output [0,1] % Reflectance Input [0 100%] Sensors convert (quantize) the continuous values into integer values, the range of which depends on the number of bits being used by the sensor. RADIANCE MAPPING }

14 4 8 bit 16 bit BITS PER PIXEL

15 How many different shades of gray can we distinguish? Typically it is stated that we can only distinguish 16 levels of grey, although some say that we can distinguish up to 30, but we also know that unless a (black & white) photo is scanned using 256 shades of gray, it doesn t look right. A computer, of course, can distinguish every unique colour or shade of gray. Colour images: How many colour values are possible in a 3 colour, 3 band Landsat image? 2 8 x 2 8 x 2 8 = 256 x 256 x 256 = >16 million THE VALUE OF NUMBERS

16 A DEM represented using 2, 4, 8 and over 256 colours THE NUMBER OF VALUES

17 8-bit radiometric resolution 2 8 = 256 levels per colour 3-bit radiometric resolution 2 3 = 8 levels per colour RADIOMETRIC RESOLUTION

18 Landsat 3: 6 bits Spot, Landsat (after 3): 8 bits Ikonos, Quickbird: 11 bits ASTER VNIR and SWIR: 8 bits; TIR: 12 bits Often the quantization values aren t reported or easily determined (e.g.). TYPICAL QUANTIZATION VALUES

19 Standards for storing and processing digital images can be different than the radiometric resolution of data Example Quickbird satellite: 11 bit radiometric resolution provided as 8 bit or 16 bit 8 bit: shadows may appear black 16 bit: detail is visible in shadows due to small differences recorded in wavelength reflectance 8 bit 16 bit DIGITAL STORAGE AND DISPLAY

20 Thermal resolution is often described in terms of the NETD (Noise Equivalent Temperature Difference), or the MRTD (Minimum Resolvable Temperature Difference). Both of these criteria refer to the ability of the system to detect a small change in temperature of an object in the field of view. For the GOES 1-M sensor, the resolution is about 1K For MODIS, it varies by band NASA Terra satellite image of San Diego wildfires. Red pixels indicate fire activity WHAT OF THERMAL (IR) SENSORS RADIOMETRIC RESOLUTION?

21 Radiometric resolution determines the range of radiance values that can be discerned from an image. The higher the quantization rate, the finer the distinctions that can be made. With the hyper radiometric resolutions now available from some sensors, we are able to discern details in areas that previously were too dark (or too bright). SUMMARY URL

22