S技术在水利工程中的应用遥感 33页

Remote Sensing Last Lecture Absorption features Vegetation Soils Rocks and minerals Water, ice and snow Conclusions Last Lecture EMR which has interacted with matter will be deficient in some wavelengths, due to absorption processes occurring at atomic or molecular scales These absorption features create distinctive “spectral signatures” for different Earth surface types These can be exploited by “multispectral” remote sensing methods for mapping and monitoring purposes Remote Sensing Spectral Signatures Remote Sensing Remote Sensing 地表任一点反射的电磁波被分解为若干个波段，每个波段有一个反射率 Remote Sensing 反射率的大小用灰度表示，并被转换成数字信号，通常是 2 8 （ 0-255 ） 50 40 30 20 10 0 0.4 0.6 0.7 0.8 1.3 R E F L E C T A N C E (%) Wavelength (micrometers) Visible 0.5 Near IR Grass GREEN BLUE GREEN RED BLUE GREEN RED NEAR IR Near IR Band Placement Band Display Band 1 Band 2 Band 3 Band 4 Band 5 Band 7 BLUE GREEN RED NEAR IR SHORT WAVE IR MID- WAVE IR NEAR IR Color Theory All colors created from additive primary colors: Red Green Blue Complementary colors: Magenta Yellow Cyan Red Green Blue M C Y W Black Multispectral Display BLUE GREEN RED NEAR IR SHORT WAVE IR MID- WAVE IR LONGWAVE IR 1 Landsat TM Band 2 3 4 5 7 6 Band Combination = 7 4 2 (LANDSAT) Color Guns = Band Composite Output = Band 3 Visible Red Band 2 Visible Green Band 1 Visible Blue Individual Landsat Bands Applied to Color Guns Resulting Image Band 4 Near Infrared Band 3 Visible Red Band 2 Visible Green Individual Landsat Bands Applied to Color Guns Resulting Image AVRIS Hyperspectral Cube Airborne Visible/Infrared Imaging Spectrometer 224 spectral channels 400 – 2500 nm spectral resolution, 20 meter ground resolution. Brine Shrimp pond Sensing Sensor array Lens 225 214 199 198 202 176 Each “cell” recorded as a “digital number” (DN) or “brightness value” Measures amount of EM radiation The brighter the signal, the higher the value. Pixels Each cell is called a “ picture element ” , or pixel Each pixel represents a single brightness value for a specific geographical area 225 204 188 146 214 198 169 152 202 200 178 162 i columns j rows i x j = 4 x 4 = 16 pixels 114 109 101 97 Sensor Properties Spatial resolution Spectral resolution/# bands Radiometric resolution Temporal resolution Source: NASA Spatial Resolution Measure of the smallest angular or linear separation between 2 objects that can be resolved by the sensor In practice, sensor system ’ s nominal spatial resolution is the dimension in meters (or feet) on the ground projected instantaneous field of view (IFOV) Generally, smaller spatial resolution  greater the resolving power of the sensor system Spatial Resolution IKONOS 4m Landsat 30m DOQ 0.5m © Space Imaging (cont.) Spatial Resolution Useful rule : T o detect a feature, the spatial resolution of the sensor system should be less than ½ the size of the feature measured in its smallest dimension. Spectral Resolution Number and size of the bands which can be recorded by the sensor – nominal spectral resolution Coarse – sensitive to large portion of ems contained in a small number of wide bands Fine – sensitive to same portion of ems but have many small bands Goal – finer spectral sampling to distinguish between scene objects and features More detailed information about how individual features reflect or emit em energy increase probability of finding unique characteristics that enable a feature to be distinguished from other features. Spectral Resolution the SPECTRAL resolution defines the range of light stored in the image A black and white photograph stores a visible light; it has one channel that stores the light for 0.4 to 0.7 micrometers A natural color image stores reflected red, blue and green light in different channels; e.g. 0.45 - 0.52 m m for blue, 0.52 - 0.60 m m for green and 0.63 - 0.69 m m for red A LANDSAT image contains 7 channnels as described above that store reflected light other than visible light. A HYPER-SPECTRAL image contains hundreds of channels. E.g. A hyperspectral image that collects visible light may divide the visible light range into 300 channels, each channel containing a narrow range of wavelengths. Spectral Resolution/# Bands 100s of Bands Hyper-spectral NIR SWIR LW IR SWIR Band 2 .53-.62 Band 3 .63-.69 Band 1 .45-.52 Visible Band 4 .79-.90 Band 5 1.55-1.75 Band 7 2.08-2.35 Band 6 10.4-12.4 Near IR SWIR LWIR 1000s of Bands Ultra- spectral Multi- spectral Spectral Resolution/# Bands Radiometric Resolution the RADIOMETRIC resolution defines the range of values that an individual pixel can have Refers to the sensitivity of the sensor to incoming radiance. Typical digital images have a range of values from 0 – 255 (a total of 256 possible values). An image that just has black or white pixels would only store 0 (black) or 1 (white). RADAR images have range from 0 to 4.3 x 10 9 . Radiometric Resolution Temporal Resolution How often the remote sensing system records imagery of a particular area. Examples – Landsat 18 days SPOT 26 days Temporal Resolution 2752 Km at the Equator 185 Km Orbit 1, Day 1 Orbit 2, Day 1 Orbit 1, Day 8 Orbit 2, Day 8 Landsat Trade-Offs Aerial Photo IKONOS Landsat Spatial Resolution ½ m 4m 30m # Bands 1 4 7 Radiometric Resolution 8 bit 11 bit 8 bit Temporal Resolution On demand 3-4 days 16 days © Space Imaging Spatial and temporal requirements for remote sensing applications in agriculture. Resolution requirements for different applications Summary of Resolution By increasing 1 or any combination of these resolutions, increase chance of obtaining remotely sensed data about a target that contains accurate, realistic, and useful information. Downside of increased resolution  need for increased storage space, more powerful processing tools, more highly trained individuals. The End Source: Space Imaging