notes for remote sensing of the cryosphere · (portions of the text are taken from: remote sensing...

15
Notes for Remote Sensing of the Cryosphere Leigh Stearns University of Kansas, Department of Geology (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school in 2014. For more information about sensor data, see Allen Pope’s 2014 Remote Sensing article: Open Access Data in Polar and Cryospheric Remote Sensing.) 1. The Electromagnetic Spectrum: Remote sensing of earth processes relies on the interaction between electromagnetic waves and matter. The interaction between materials and electromagnetic waves depends on both the characteristics of the electromagnetic radiation (e.g., frequency) and on the chemical and physical properties of the targets. In many cases (passive sensors), the source of the electromagnetic radiation is the sun, which can be approximated as a black body (an idealized body that absorbs all incident electromagnetic radiation, regardless of frequency) at a temperature of ~5800 K. Though a large number of remote sensing applications deal with the visible portion (400–700 nm) of the electromagnetic spectrum, visible light occupies only a fraction of it. Indeed, a considerable portion of the incoming solar radiation is in form of ultraviolet and infrared radiation, and only a small portion is in form of microwave radiation (Tedesco, 2015). Leigh Stearns: Remote Sensing of the Cryosphere Lecture 1: Remote Sensing Basics Objectives: - Be able to describe how the electromagnetic spectrum is used in remote sensing; - Be able to name the different types of sensors and their main differences; - Be familiar with basic remote sensing processing steps and corrections; - Know how to find imagery! 1

Upload: others

Post on 24-Jun-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

Notes for Remote Sensing of the CryosphereLeigh Stearns

University of Kansas, Department of Geology

(Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school in 2014. For more information about sensor data, see Allen Pope’s 2014 Remote Sensing article: Open Access Data in Polar and Cryospheric Remote Sensing.)

1. The Electromagnetic Spectrum:Remote sensing of earth processes relies on the interaction between electromagnetic waves and matter. The interaction between materials and electromagnetic waves depends on both the characteristics of the electromagnetic radiation (e.g., frequency) andon the chemical and physical properties of the targets. In many cases (passive sensors), the source of the electromagnetic radiation is the sun, which can be approximated as a black body (an idealized body that absorbs all incident electromagnetic radiation, regardless of frequency) at a temperature of ~5800 K. Though a large number of remote sensing applications deal with the visible portion (400–700 nm) of the electromagnetic spectrum, visible light occupies only a fraction of it. Indeed, a considerable portion of theincoming solar radiation is in form of ultraviolet and infrared radiation, and only a small portion is in form of microwave radiation (Tedesco, 2015).

Leigh Stearns: Remote Sensing of the Cryosphere

Lecture 1: Remote Sensing BasicsObjectives:

- Be able to describe how the electromagnetic spectrum is used in remote sensing;- Be able to name the different types of sensors and their main differences;- Be familiar with basic remote sensing processing steps and corrections;- Know how to find imagery!

1

Page 2: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

Figure 1: The electromagnetic spectrum and types of sensors. The top panel shows the wavelengths of energy that are radiated by the sun and Earth. The middle panel shows the wavelengths that are transmitted through the atmosphere. The bottom panel shows the wavelengths relevantfor different remote sensing platforms(Lillesand and Kiefer, 1999).

Before the Sun’s energy reaches the Earth’s surface, three interactions that are relevant for remote sensing occur in the atmosphere: absorption, transmission, and scattering. The energy transmitted is then either absorbed by the surface material or reflected. As Figure 1 (middle panel) shows many wavelengths in the electromagnetic spectrum are not transmitted through the atmosphere. These wavelengths are therefore not useful for remote sensing of the Earth’s surface, simply because none of the corresponding energy can penetrate the atmosphere. The useful ranges are referred to as the atmospheric transmission windows and include:

- The window from 0.4 to 2 μm. The radiation in this range (visible, NIR, SWIR) is mainly reflected energy. Because this type of radiation follows the laws of optics, remote sensors operating in this range are often referred to as optical ones.

- Three windows in the thermal range, namely two narrow windows around 3 and 5 μm, and a third, relatively broad window extending from approximately 8 to 14 μm.

Because of the presence of atmospheric moisture, strong absorption is occurring at longer wavelengths. There is hardly any transmission of energy in the range from 22 μm to 1 mm. The more or less “transparent” range beyond 1 mm is the microwave range.

2. Sensor Types:

Leigh Stearns: Remote Sensing of the Cryosphere 2

Page 3: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

A remote sensor measures reflected or emitted energy. An active sensor has its own energy source, while a passive sensor utilizes energy from the sun. As a result, passive sensors operate predominantly in the shortwave part of the electromagnetic spectrum (except for passive microwave sensors). Active sensors include radar, LiDAR, and active microwave sensors and generally operate over longer wavelengths.

Image files consist of a digital image for each spectral band of the sensor. A digital image is simply a 2-dimensional array of pixels. Each pixel has an intensity value (digital number) and a location address (referenced by its row and column number). Imagery can be classified based on four types of resolution: spectral, spatial, radiometric, and temporal.

- Spectral resolution describes how a particular sensor divides up the electromagnetic spectrum into specific bands.

- Spatial resolution is the size of each individual pixel.- Radiometric resolution refers to the smallest change in intensity level that can

be detected by the sensor. - Temporal resolution is how many times a specific area is imaged by the

sensor.

Figure 2. Resolutions for sensors and platforms operating in the visible (and near visible) parts of the electromagnetic spectrum

3. Processing steps and corrections:

Leigh Stearns: Remote Sensing of the Cryosphere 3

Page 4: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

All imagery requires some sort of processing. Occasionally, corrections are applied to the imagery ibefore you download it (e.g. Level-0 imagery is typically the uncorrected image, Level-1 imagery has sensor and geometry corrections applied, and Level-2/Level-3, etc are higher-order products). Typical processing schemes include atmospheric corrections, sun illumination geometry, surface-induced geometric distortions, spacecraft locations, effects of Earth’s rotation and curvature, abnormalities of the instrument platform, etc. Processing that you will likely need to do, include geometric corrections and map projections.

4. Where to find imagery:Figure 3 shows a list of current and planned satellite missions that relate to cryosphere applications (no guarantees that it’s all inclusive!). These missions are funded by a rangeof different agencies, each with different methods of distributing, storing and archiving imagery. Below are a few tips that hopefully will help you navigate this system.

- ESA is launching a new constellation of satellites, Sentinel. Sentinel 1 and 2 (both are satellite pairs) are particularly useful for cryosphere applications because of their high spatial and temporal resolution. Sentinel 1 is a radar; Sentinel 2 is a multispectral sensor (similar to Landsat 8).

- NASA imagery is managed by the Earth Observing System Data and Information System (EOSDIS). EOSDIS funds various Distributed Active Archive Centers (DAACs) to process, archive, document, and distribute data from NASA’s part and current Earth-observing satellites and field measurement programs. Most DAAC holdings are also available through REVERB (but not all). In addition, individual DAACs often have application-specific programs that might help subset, download, or process the imagery. They also have very helpful employees who are well-versed in using their specific imagery. The following DAACs may be useful to you:

o National Snow and Ice Data Center (NSIDC DAAC): cryosphere products

o Alaska Satellite Facility (ASF DAAC): SAR products from polar regionso Physical Oceanography (PO DAAC): ocean productso Land Processes (LP DAAC): ASTER, MODIS, and other land sensors

Leigh Stearns: Remote Sensing of the Cryosphere 4

Page 5: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

Figure 3: Cryosphere satellite missions (http://globalcryospherewatch.org/satellites/overview.html).

Most imagery that you’ll download (especially the Level 1 scenes) will be in HDF or NetCDF format. If you haven’t worked with these file formats before, don’t freak out. There are good plugins or code snippets for most processing platforms that will help you efficiently work with this imagery.

Leigh Stearns: Remote Sensing of the Cryosphere 5

Page 6: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

1. Optical Imagery:a. Aerial PhotographyAerial photography is perhaps the most traditional and longest used system in remote sensing. A photochemical reaction occurs when a silver halide crystal ‘grains’ is exposedto light, transforming into metallic silver. Chemical development then results in a photonegative from which further processing is possible. In this instance, the film is the detector and the film and filters determine the spectral response.

Aerial (and ground-based) photographs can be used to map glacier extent, derive glacier velocities, produce digital elevation maps through stereo-imaging, and map structures.

b. Multispectral imageryIn multispectral imagery, the radiance is observed for a ‘spectral band’, not for a single wavelength (Figure 4). A spectral band or wavelength band is an interval of the electromagneitc spectrum for which the average radiance is measured. Sensors like a panchromatic camera, a radar sensor, or a laser scanner only measure in one specific band while a multispectral scanner or a digital camera measures in several spectral bands at the same time. Multispectral sensors have several ‘channels’, one for each spectral band.

Figure 4: Band spectra for Landsat 7 and Landsat 8.

Leigh Stearns: Remote Sensing of the Cryosphere

Lecture 2: Passive Sensors and ApplicationsObjectives:

- Be able to differentiate between different types of optical imagery;- Be able to describe several cryosphere applications that utilize optical, thermal

and passive microwave sensors.

6

Page 7: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

An individual band will ‘appear’ black and white, but when the bands are stacked and assigned a R,G,B, color, composite images are made. Composite images are useful for classifying the extent of different surface materials – dependent on their reflectance signature. For example, a false-color composite (band 4 = red, band 3 = green, band 2 = blue) is useful for highlighting changes in vegetation type.

Additional band combinations are useful for differentiating between snow, clouds and land. Since snow is highly reflective in the visible part of the spectrum and highly absorptive in the NIR and SWIR (Figure 5), a band combination that isolates these two spectral properties will help classify snow in imagery. With the Normalized-Difference Snow Index (NDSI), the normalized difference of two bands (1 in visible, 1 in NIR or short-wave IR) is used to map snow.

Figure 5: Spectral properties of snow and ice.

Hyperspectral imagery consists of hundreds of bands (nearly continuous along the electromagnetic spectrum) enabling researchers to differentiate between different types of rocks and minerals.

Multispectral imagery can be used to make glacier extent, classify snow/ice facies, determine ice velocity, produce DEMs, and map structures.

2. Thermal Imagery:There are three basic ways in which energy can be transferred: conduction, convection, and electromagnetic radiation. Radiation is of primary interest to remote sensing

Leigh Stearns: Remote Sensing of the Cryosphere 7

Page 8: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

because it is the only form of energy transfer that can take place in a vacuum such as theregion between the Sun and the Earth.

An object’s internal kinetic heat is also converted to radiant energy (often called externalor apparent energy). The electromagnetic radiation exiting an object is called radiant flux (Φ) and is measured in watts. The concentration of the amount of radiant flux exiting (emitted from) an object is its radiant temperature (TR). There is usually a high positive correlation between the true kinetic temperature of an object (TK) and the amount of radiant flux radiated from the object (TR). Therefore, we can utilize radiometers placed some distance from the object to measure its radiant temperature, which (hopefully) correlates well with the object’s true kinetic temperature. This is the basis of thermal infrared remote sensing.

Thermal imaging can be used to measure glacier/iceberg/plume extents and variabilityin their temperature over time.

3. Passive Microwave:Passive microwave data, as the name implies, involve the detection and measurement of thermal radiation in the microwave region of the electromagnetic spectrum (typically 5–90 GHz). The fundamental variable that is measured is the brightness temperature. The absorption and emission of radiation by the Earth’s atmosphere can be significant, even dominant, in the microwave spectrum, and some instruments are optimized for measuring properties of the atmosphere itself. These include the measurement of sea iceextent and concentration and various parameters describing snow cover, both on land and on sea ice. A major advantage of passive microwave data is the fact that it can be collected at night and (except at the highest frequencies) through cloud. A major disadvantage is that the spatial resolution of the data (5 km to 50 km) is poor compared with visible-near infrared systems or SAR.

Passive microwave data is used to measure sea ice extent, snow melt, accumulation patterns, snow water equivalent, snow pack temperature and snow depth.

Leigh Stearns: Remote Sensing of the Cryosphere 8

Page 9: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

1. Radar:Radar systems emit a radar pulse, which undergoes backscatter at the surface depending on the interaction of the radar frequency and the material. The amplitude and phase echo is recorded back at the satellite (or airplane). The phase is a function of the distance from the satellite (airplane) to the ground.

Figure 6: A geometric model for a SAR system. Slant range is the length between the antenna and ground pixel and ground range is the distance between the ground track and the ground pixel.

Depending on the radar system variable look angles can be deployed. The image formation process is done following the order of reception of the return pulses. Hence, the two-way travel time, t, from the sensor to the target and back to the antenna determines the location of the points in the image in range direction. The respective geometry is referred to as slant range (Figure 6). Since the travel time of the electromagnetic wave in the air is approximately the velocity of light c, the slant range distance of the antenna (SR) to an object on the ground is given by

Leigh Stearns: Remote Sensing of the Cryosphere

Lecture 3: Active Sensors and ApplicationsObjectives:

- Know the basic terminology of radar imagery and processing steps;- Be able to describe how radar echograms and interferograms are used in

glaciology;- Be able to describe what LiDAR is, and the basic corrections that are needed in

order to apply it to glaciology;

9

Page 10: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

SR = ct/2.

The ground resolution cell size of a side-looking radar is controlled by two independentsensing system parameters: pulse length (range) and antenna beamwidth (azimuth). In order to achieve an appropriate azimuth resolution, Doppler shift along track is used to synthesize a virtually long antenna.

While some of the geometric effects can be corrected using a sensor model and DEM, the backscatter signal is influenced by the local incident angle (Figure 7). As a rule-of-thumb, slopes oriented towards the sensor appear brighter and slopes reflecting the signal away from the antenna appear darker. However, the degree to which this occurs depends strongly on the surface type and related backscattering mechanisms. Hence, it is difficult to correct these radiometric effects although generally, SAR data calibration ismuch easier and accurate than for optical imagery

Figure 7: Illustration of the generalimpact of the local incident angle (&) on the backscatter intensity

The image formation according to the travel time of the signal has very specific implications in regard to image distortions. While there are no image artifacts due to topography on a plane surface, Figure 8 illustrates the various geometric distortions in SAR images in mountainous terrain, namely foreshortening, layover and shadowing.

Leigh Stearns: Remote Sensing of the Cryosphere 10

Page 11: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

Leigh Stearns: Remote Sensing of the Cryosphere 11

Page 12: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

Figure 8: Geometric distortions introduced by the image formation process of the side-looking active SAR system. a) foreshortening in areas oriented towards the sensor, b) layover effects in very steep terrain and c) shadowing in areas where the radar signal does not reach the ground.

Synthetic Aperture Radar (SAR) is a method of active microwave imaging that uses the echo returns from many radar pulses to create a single image. Because it is an active microwave system, the image shows radar backscatter rather than optical reflectance or emitted radiation; a simplistic way to think of the image is it being related to surface roughness. SAR technology has many research and monitoring applications in the polarregions (e.g., iceberg tracking, glacier and ice sheet velocity measurements, sea ice mapping, lake ice mapping, etc.), and especially valuable is the capability toimage through cloud cover and during long periods of darkness during polar winters. As such, polar regions have been historically well served by SAR satellites.

Interferometric SAR exploits the difference in phase between two positions. The phase difference is shown as fringe patterns (interferograms), which are used to determine surface topography or motion (Figure 9).

Leigh Stearns: Remote Sensing of the Cryosphere 12

Page 13: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

Figure 9: A cartoon representation of the basic principles of radar interferometry. As the satellitemakes its first pass over a ground surface (‘Initial ground surface’), it collects radar wavesreflected off of the ground surface (solid wave, ‘pass 1’). During a subsequent orbit (oftenmonths to years later), when the satellite again passes over the same ground surface, anothercollection is made from very nearly the same orbital location (dashed wave, ‘pass 2’). If theground surface deformed during the time between data collections (e.g., ‘Subsided groundsurface’), then the collected radar waves of the second pass will be out of phase compared tothose collected during the first pass (example waves A-E, at right). The phase difference of thewaves is then converted into the component of ground motion along the line-of-sight of thesatellite (either towards or away from the satellite), and is represented by a color as part of a fullcolor cycle. Since the technique is based on the phase difference of multiple waves, the accuracyis constrained by detectable fractions of the radar wave’s wavelength (http://volcano.si.edu).

2. LasersLIght Detection And Ranging (LiDAR) uses the same principle as RADAR. The LiDAR instrument transmits laser out to a target. The transmitted light interacts with atmosphere and target. Some of this light is reflected / scattered back to the instrument

Leigh Stearns: Remote Sensing of the Cryosphere 13

Page 14: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

where it is analyzed. There are two main types of LiDARs used to laser altimetry – a single pulse and a photon counting system. Among other advantages, the newer photoncounting system, which will be deployed on ICESat-2, has a larger footprint allowing slope corrections and cross-over points to be more easily tracked.

Several corrections are necessary to convert elevation changes to mass change. First, the density of the surface layer must be known. The density will be a function of precipitation patterns, air temperature, snow melt, and is therefore often poorly constrained. In addition, estimates of firn compaction (which can be essential in parts of the ice sheet) need to be quantified. Finally, most airborne or satellite LiDARs have a small footprint, so corrections for surface slope or cross-over offsets need to be made (Figure 10).

Figure 10: Cartoon showing how laser scanner operates on an aircraft.

3. Active Microwave Using the same wavelengths as passive microwave and synthetic aperture radar instruments, active scatterometry sends out microwave pulses and measures the

Leigh Stearns: Remote Sensing of the Cryosphere 14

Page 15: Notes for Remote Sensing of the Cryosphere · (Portions of the text are taken from: Remote Sensing of the Cryosphere (Tedesco, 2015) and Nick Barrand’s notes for the summer school

returned backscatter at multiple angles. As with other microwave sensors, the returned signal depends on two properties: surface roughness and electrical properties. Originally, scatterometers were designed to measure winds over the open ocean, as derived from surface roughness. However, because scatterometer signals can penetrate some surfaces, they give bulk, as well as surficial information, which is a function of land/sea ice cover.

Leigh Stearns: Remote Sensing of the Cryosphere 15