Spectroradiometer: Working Principle & Protocol of Spectral Signature Collection

Transcription

1 Spectroradiometer: Working Principle & Protocol of Spectral Signature Collection Division of Agricultural Physics, Indian Agricultural Research Institute New Delhi Hyperspectral Remote Sensing : Hyperspectral remote sensing deals with large number of narrow spectral bands over a contiguous spectral range Because of its ability to detect narrow absorption features hyperspectral data are related to specific vegetation physico-chemical characteristics, ocean biological constituents, soil physical and chemical properties, mineral composition and snow characteristics Because of presence of large number of bands, hyperspectral data needs different analysis approach 1

2 B G R NIR SWIR MWIR LWIR 400 nm nm Panchromatic: one very wide band Multispectral: several to tens of bands Hyperspectral: hundreds of narrow bands Ultraspectral: thousands of narrow bands Spectral Resolution Absorption features are typically very narrow (< 20 nm), so narrow band widths are necessary. Many important band widths are also fairly shallow, so high radiometric resolution and SNR is also necessary. 2

3 Vegetation :Multispectral Vs. Hyperspectral RS Multispectral imaging is insufficient to derive the required terrestrial surface compositional parameters accurately. Full range imaging spectroscopy is required to measure composition, chemistry, health and change of ecosystems 12/21/2016 1:55 PM Signature of Different Landcovers : Hyperspectral vs Multispectral 0.50 Reflectance Crops Habitation Plantation Soil Water Modipuram (U. P.) Hyperion EO-1 (As per IRS Bands) Wavelength(nm) ~10 nm bandwidth Crops Habitation Plantation Soil Water 3

4 After Elachi, JPL Hyperspectral Sensors Satellite-based Imaging Spectroscopy Data: EO-1 Hyperion HYSI Airborne Imaging Spectroscopy Data: HyMap Probe CASI AVIRIS AVIRIS-NG µm 2 16 µm Ground Based Spectroradiometer FTIR Imaging Spectrometer (optical and thermal) (also can be used airborne) Purpose of Spectro-radiometer Different Spectral Imaging Sensors (SIS) have different spectral bandwidths and positions. Ground referencing spectral data that can be convolved to i.e. Re-sampled to, all of these different sensor bands must be of high enough resolution and wide enough spectral region to do so. Spectroradiometer : - Ground based Remote Sensing - Helps understanding basics of target and EM interaction - Point/single pixel based study - Spectral ground truth of Satellite Remote Sensing - Act as a reference for upscale to satellite /airborne imaging sensors with required bands 4

5 Components of Spectroradiometer 1. A Light source 2. A dispersive unit (monochromator) 3. A detector 4. (Fibres) 5. Absorbance / reflectance standard A. Light Source Solar Energy Spectrum at Earth's surface Toungsten Quartz halogen lamp Room light spectrum 5

6 B. Dispersive Unit : POST DISPERSIVE vs. PRE DISPERSIVE Portable spectrometers are typically used outside the controlled laboratory environment They are exposed to much higher levels of ambient light. In almost all cases, some of this ambient light will stray into the sample being measured. The errors produced by this ambient stray light are much greater for a pre-dispersive spectrometer than they are for a spectrometer that is post-dispersive. POST DISPERSIVE vs. PRE DISPERSIVE In a Pre-dispersive spectrometer the ambient stray light signal can represent a large fraction of the total light signal measured by the detector, thus, it can be a major source of error. In a Post-Dispersive spectrometer the ambient stray light scattered from the sample is also collected, but with the post-dispersive instrument only ambient stray light of the same wavelength as that being measured by the detector is added to the signal. Thus, the stray light signal represents a much smaller fraction of the total light signal measured by the detector resulting in an increase in instrument baseline stability. Post-Dispersive spectrometers provide the greatest amount of sampling, or sample interface flexibility. Post-Dispersive systems allow for non-contact and remote measurements where Pre- Dispersive systems generally have problems with ambient stray light. 6

7 C. Detector Region Names wavelength nanometers (nm) or microns (µ) Cosmic Ray Gamma Ray X Ray *Optical Ultra- Far Ultra-Violet nm Violet Ultra Violet C (UVC) nm (UV) Ultra Violet B (UVB) nm Ultra Violet A (UVA) nm Visible Photosynthetically Blue Light nm (VIS) Active Green Light nm Infrared (IR) Radiation Yellow Light nm (PAR) Red Light nm Near-Infrared (Near-IR) Far Red Short Wave Near Infrared (SW-NIR) Typical 1st NIR region detector (NIR1) or (SWIR1) Typical 2nd NIR region detector (NIR2) or (SWIR2) 'Conventional' Near Infrared (NIR) nm nm nm nm nm or µm Mid Infrared µm (Mid-IR) Thermal (emitted) 8-15 µm Far Infrared (Far-IR) µm C. Detector :Sampling Interval and Spectral Resolution Spectral sampling interval is the spacing between sample points in the spectrum. Sampling is independent of resolution and in ASD spectroradiometers is between 2 and 5 times per FWHM. The sampling interval here is 1.4 nm for the region nm and 2 nm for the region nm. Spectral resolution is defined as the full-width-half-maximum (FWHM) of the instrument response to a monochromatic source. This is in fact the definition ASD uses when stating spectral resolution specifications.. These spectral resolution values have been measured by calculating the FWHM of a near monochromatic peak in a spectrum acquired when viewing the output of a monochromator with the FieldSpec3 spectroradiometer. 7

8 C. Detector : MODULAR CONCEPT nm nm nm nm UV/VNIR SWIR1 SWIR2 UV/VNIR Grating with 512 Element Silicon Photodiode array detector Scanning SWIR1 Grating with Single Tempt controlled Indium Galinium Arsenide detector Scanning SWIR2 Grating with Single Tempt controlled Indium Galinium Arsenide detector WORKING MECHANISM VNIR Spectral Range 350 to 1000 nm Grating Splits the wavelength every 1.4 nm These wavelengths hits individual element in the SiP diode array. SWIR1 and SWIR2 Spectral Range 1000 to 1800 nm & 1000 to 2500nm SWIR1 Grating is mounted on a motor which scans continuously Wavelength is split every 2 NM and hits one by one on the single InGaS detector. 8

9 Field-of-View and Coverage Area D == effective diameter of foreoptic lens A == foreoptic's angular field-of-view X == distance to viewed surface Y == diameter of field-of-view Near Field (less than 1 meter): Y = D + 2 * X * Tan( A/2 ) Far Field (greater than 1 meter): Y = 2 * X * Tan( A/2 ) FOV 9

10 FOV Field-of-View The small size of the FieldSpec's foreoptics allows positioning the foreoptics at a greater distance from the surface under observation. A field spectroradiometer with a larger field-of-view means that fewer measurements are needed to approximate the spatial resolution of the imaging sensor, because the pixel size of most imaging sensor systems is several meters or more. The small size of the pistol grip and foreoptics greatly reduce errors associated with instrument self-shadowing. Even when the area viewed by the foreoptic is outside the direct shadow of the spectroradiometer, the instrument still blocks some of the illumination that would normally be striking the surface under observation, either diffuse skylight or light scattered off surrounding objects. Thus, the instrument as well as other objects including the user should be placed as far as possible from the surface under observation. This orientation requirement also applies to white reference measurements. 10

11 Fibreoptic Cable ASD spectroradiometers are designed with a permanent mount fiberoptic cable which feeds directly into the spectrometer. There are two advantages to this arrangement. 1. The fiber optic input allows the user to quickly move and aim the probe from point to point without having to move the entire spectroradiometer. 2. Since the fiber optic cable is connected directly into the spectroradiometer there is none of the signal losses associated with detachable couplings and connectors. It is well documented and proved that the detachable fiberoptic cable result in as high as 50 % signal loss. Fiber Optics Cables The fiber optic cable is made up of fifty-seven (57) randomly distributed glass fibers. Nineteen (19) of these fibers are 100- micron and are distributed to the VNIR region. The remaining thirty-eight (38) fibers are 200-micron and are evenly divided between the two SWIR regions. All fifty-seven fibers are housed in a plastic sheath. A standard cable has a diameter of 0.19 inches and is an f2 cable. The fibers are protected by a metal spiral inside the black cable casing. If there are kinks in the cable, the fibers are not necessarily damaged. If, however, the cable has been crunched so severely that the protective metal spiral can be seen, the chances are high that the fibers have also been damaged. Each broken fiber results in a ~5% loss of response. WARNING!The fiber optic cable should never be stored with a bend of less than a 5" diameter for long periods of time. 11

12 Field Portability - Portability - Battery Life Other Accessories Variety of Foreoptic Lenses for Different Field of View Remote Cosine Receptor, Sun Tracker, etc Spectralon White Reference Panels. Tripods Special sampling Accessories for Lab and Field Applications Different accessories Irradiance attachment, spectral needle, contact probe, leaf clip, Mug 12

13 LED Check for Fiber Optic Cable The FieldSpec spectroradiometer allows you to perform a visual verification of the fiber optic cable using built-in LEDs, a magnifier placed on the fiber optic cable, and the Fiber Check software. White Reference A material with 95-99% reflectance across the entire spectrum is called a white reference panel or white reference standard. Spectralon from Labsphere is the white reference standard that is very suitable for the VNIR and SWIR spectral ranges of ASD instruments Spectralon is made of polytetraflouroethylene (PTFE) and cintered halon. It has the characteristic of being nearly 100% reflective within the wavelength range of 350 nm to 2500 nm. A Spectralon white reference scatters light uniformly in all directions within that wavelength range. 13

14 Speed and Sensitivity can record a complete nm spectrum in 0.1 seconds Fast speed in combination with extremely low Noise-equivalent- Radiance (NeDL) are what make the optimal spectroradiometer Optical Fibre length and SNR 1meter 10 meter 14

15 Accuracy and Precision Precision: Precision is a measure of how close repeated measurements are around a particular value. Accuracy: Accuracy is a measure of how close repeated measurements are around the True Value. STRAY LIGHT REJECTION Stray light is one form of constant systematic noise. Stray light results in computed reflectance values that are different from the actual values. The appearance of reflectance signal in spectral regions of zero illumination energy and the appearance of biased signals in spectral regions of low illumination energy are indicators of stray light problems. Solar Energy Spectrum 15

16 STRAY LIGHT REJECTION Measured signal = true signal + dark current + stray light + random noise dark current can be easily recorded and subtracted so that it is a negligible contributor. Therefore, assuming that the dark current is pre-subtracted we can re-write the above formula as follows: Dark corrected measured signal = true signal + stray light + random noise Using dark corrected measured signal, computed reflectance is written as follows: Computed reflectance = (dark corrected measured signal from target) / (dark corrected measured signal from reference) = (true target signal + stray light + random noise) / (true reference signal + stray light + random noise) If stray light is negligible, then for regions of near zero illumination energy, i.e., signal less than or equal to the instrument random noise at the time the signals are recorded, computed reflectance is written as: Computed reflectance = (random noise at time of target measurement) / (random noise at time of reference measurement) STRAY LIGHT REJECTION Now, going back to the formula for computed reflectance: computed reflectance = (true target signal + stray light + random noise) / (true reference signal + stray light + random noise) Again, we consider the water bands where illumination energy is zero, but this time we consider the case of significant stray light. That is, stray light significantly greater than the lowest level random noise. Computed reflectance is then written as follows: computed reflectance = (stray light + random noise) / (stray light + random noise) In this case, the computed reflectance is no longer a simple ratio of random noise signals but the ratio of two significant signals. This anomaly results in what appears to be real reflectance signals where there should only be random noise 16

17 STRAY LIGHT REJECTION Figure above is reflectance spectrum for a 50% reflectance target computed by dividing the reflected solar radiance spectra for a 50% reflectance target by that for a 100% reflectance reference panel. The figure above is reflectance spectrum for a 50% reflectance target with the addition of a 0.5% stray light component. STRAY LIGHT REJECTION Collected reflected solar radiance spectra for both a polyester fabric and a 100% reflectance reference panel. The computed reflectance spectrum shows the effects of the deep atmosphere water vapor absorption band centered near 1900 nm. When the same material is measured in the laboratory using a tungsten filament illumination source, the reflectance spectrum shows absorption features in the 1900 nm region that were obscured in the spectrum collected using solar illumination. 17

18 RELIABILITY OF INSTRUMENT Choosing a Spectroradiometer.. - Purpose of use - Required sensor specification - Spectral range? 350 nm 1050nm 350 nm-1800nm 350nm nm 2000nm nm (FTIR) - Spectral Resolution - Sampling Interval - FOV 3 to 25 degree - Length of Optical fibre - Portability - Battery Life - Signal to Noise ratio - Additional functionalities - GPS enable - WiFi / Bluetooth connectivity - Post processing software - Accessories for other observations A. Spectra Vista, USA GER 1500, 3700 HR 1024 B. Analytical Spectral Devices Inc, USA Field Spec JR, FR, FS3, FS4 C. Spectral Evolution 18

19 Cautions while Field Measurements Field spectrometry is the quantitative measurement of radiance, irradiance, reflectance or transmission in the field. It involves the collection of accurate spectra and requires an awareness of the influences of: Sources of illumination. Atmospheric characteristics and stability. Winds. Instrument field-of-view. Sample viewing and illumination geometry. Instrument scanning time. Spatial and temporal variability of the sample characteristics. Initial Warming time, Optimization and DC removal Post Processing of spectral data White Reference Procedures White references should be collected approximately every ten (10) minutes and can be varied depending on whether conditions are changing rapidly or not changing very much. The Spectralon puck should be used when optimizing and taking a white reference measurement. Note:For best results, the Spectralon puck should be at the same distance and angle from the foreoptic as the sample will be during measurements. When saving reflectance data, point the probe at the Spectralon with the same viewing geometry for a minute or two every few measurements. If the relative reflectance of the Spectralon is less than or greater than one, a new white reference may be needed. If the relative reflectance of the Spectralon is greater than one, re-optimization is recommended. The FieldSpec spectroradiometer should be re-optimized for: Light changes. Any atmospheric changes. Temperature changes. Whenever accessory probes are changed. Note:Environmental conditions can change rapidly or slowly. It all depends on clouds, wind (affecting temperature), instrument warm up time, etc. 19

20 Maintaining Spectralon References Spectralon is an optical standard and should be handled in much the same way as other optical standards. Although the material is very durable, care should be taken to prevent contaminants such as finger oils from contacting the material s surface. Always wear clean gloves when handling Spectralon. To clean a lightly soiled Spectralon white reference If the material is lightly soiled, it may be air brushed with a jet of clean dry air or nitrogen. WARNING!DO NOT use Freon. Caring of Optical Fibre The fibers can be damaged by coiling the cable up too tightly. If left in a tight coil for longer than a week, the fibers are likely to develop longitudinal fractures that will not be detectable. These fractures in the fiber will cause light leakage, resulting in a weaker signal. The fiber optic cables should be stored by placing them loosely within the netting compartment on the instrument or on the fiber optic spool. Tips on care for the fiber optic cable: Do not pull or hang the spectroradiometer by the fiber optic cable. Do not use wires, ties, or clamps to tightly attach the fiber optic cable to objects, this may pinch or penetrate the protective jacket thereby damaging the fibers inside. Avoid whipping the fiber optic cable, dropping it, or slamming it into objects, this can cause fractures to the glass fibers. Avoid twisting the fiber optic cable, such forces may cause fractures to fibers. While the tip of the fiber optic cable is not particularly susceptible to damage, a tip cover is recommended to protect against abrasion and exposure to contamination. Replacement covers can be made by cutting pieces of eighth-inch shrink tubing to about 1.5" lengths and shrinking them onto the fiber cable tip. They will slide on and off the cable easily. The loss in response is further amplified when jumper cables are utilized. Jumper cables are fiber optic cables that can be attached to the fixed cable via an SMA adapter. The loss on the jumpers is largely due to the attenuation at the junction between the built-in cable and the jumper. 20

21 Battery Care Proper care of the NiMH rechargeable batteries assures long life. NiMH batteries can store 50% more power than NiCad and do not suffer from memory effects. Unlike NiCad batteries, NiMH batteries do not use heavy metals that may have toxic effects. Still, never drop NiMH batteries on a hard surface. New NiMH rechargeable batteries can last from 500 to 1000 charges. When you receive a new battery pack, charge it fully before using it the first time regardless of it showing full voltage and power. It can take from three (3) to five (5) charge-and-discharge cycles before the battery pack reaches its peak performance. At normal room temperature, NiMH batteries generally lose about 1% of their charge per day. Higher temperatures increase this loss, while lower temperatures (40-60F) can reduce this daily loss. The daily charge loss is not a significant factor if the batteries are used within the first five to ten days after they have been charged. One of the best ways to ensure that the NiMH batteries last as long as possible is to use them often. The more they are charged and used, the longer the NiMH batteries last. Whether discharged or not, the NiMH batteries should be charged at least every sixty (60) days. WARNING!Do not use batteries other than those supplied with Instrument, Do not use batteries in a manner unauthorized by supplies. Using improper batteries or improper use of batteries could result in bodily injury or damage to the instrument. RS3 Battery Indicator 21

22 Product FieldSpec 3 FieldSpec 3 JR Spectral Range nm nm Spectral Resolution Sampling Interval nm / 2100 nm nm nm nm /2100 nm nm nm Scanning Time 100 milliseconds 100 milliseconds Detectors 512 element Si photodiode array nm TE cooled, graded index InGaAs for nm Second TE cooled graded index InGaAS for nm One 512 elementsi photodiode array nm Two separate, TE cooled, graded index InGaAs photodiodes for nm and nm Input 1.4 m fiber optic (25 field of view) Optional foreoptics available 1.4 m fiber optic (25 field of view) Optional foreoptics available Noise Equivalent Radiance (NEdL) UV/VNIR 1.4 x 10-9 W/cm 2 700nm NIR 2.4 x 10-9 W/cm nm NIR 8.8 x 10-9 W/cm nm Weight 12 lbs or 5.2 kg 12 lbs or 5.2 kg Calibration Notebook Computer UV/VNIR 2.8 x 10-9 W/cm 2 700nm NIR 2.4 x 10-9 W/cm nm NIR 8.8 x 10-9 W/cm nm Wavelength, reflectance, radiance*, irradiance*. All calibrations are NIST traceable (*radiometric calibrations are optional) 1 GHz processor, 256 MB Ram, 20GB Hard Disk Drive, 1024x768 graphics resolution, 24 bit color, bi-directional parallel port, AC/DC adapter/charger, 64 MB USB Flash memory drive Remote Sensing Application Agriculture, Vegetation, and Forestry Ground Truthing Minerals, Mining, and Soils Airborne Measurements Snow and Ice Water Bodies Many more.. 22

23 THANK YOU Potential Applications.. Life on Earth Photosynthesis Revealed via Spectroscopy Source : Robert Green, JPL 23