Chapter 1 - Remote Sensing Flashcards
Advantages of Remote Sensing
unobtrusive
systematic data collection - helps with sampling bias that is inherent in lots of GIS
provides fundamental information, unlike lots of GIScience
Limitations of Remote Sensing
Oversold sometimes
Human error introduced in design specifications
Active remote sensing that emits EMR (LIDAR, RADAR, SONAR) can be obtrusive
Can be expensive (greatest expense is a well trained analyst)
Machines can become uncalibrated
Remote sensing process
- Hypothesis to be tested is defined using a specific type of logic and appropriate processing model
- In situ data & collateral data necessary to calibrate remote sensor is collected
- Remote sensor data are collected (ideally at the same time as calibration data)
- In situ and remote sensor data are processed using analog, digital image processing, modelling, and n-dimensional visualization
- Metadata, processing lineage, and accuracy are provided & results communicated
Biophysical variables
Some can be measured with remote sensing such as:
water vapor in air
soil moisture
normalized difference vegetation index (nvdi)
etc.
Hybrid variables
modeled by combining other variables together
Active remote sensing (and what it’s input is a function of)
emit EMR, and reabsorb as L = f(lamba, sxyz, t, theta, P, omega)
where lamba = wavelength
sxyz = location of pixel and its size
t = temporal info
theta = angle between radiation source & terrain target of interest
P = polarization of back-scattered energy
omega = radiometric resolution
Spectral resolution
number and dimension of specific wavelength intervals (bands)
Nominal Spectral Resolution
What we call certain bands. Bands follow a gaussian distribution curve and are nominally defined/ separated at Full Width at Half Maximum of intensity
Hyperspectral
records hundreds of bands
Ultraspectral
records many hundreds of bands
Spatial resolution
smallest angular or linear separation between two objects that can be resolved by a remote sensing system.
Can be calibrated by placing black and white tarps on the ground, obtaining areal photography, computing # of resolvable line pairs per millimeter
Nominal Spatial Resolution
D = beta* H
diameter = instantaneous field of view * height above the ground
smaller = greater spatial resolving power
- historical landsat MSS data not useful for urban applications because of large spatial resolution
- rule of thumb: to detect a feature, resolution should be smaller than 1/2 of the smallest part of the feature
Sampling density
Some systems such as LIDAR use pulses of EMR at various time intervals - the spatial resolution applies to the pulse and reception of the pulse, but the sampling density is more important for analysis & has to do with interval/ frequency of observations
Temporal Resolution
How often samples of 1 location are taken
Nadir
The point directly below a satellite. Some satellites can sample off-nadir, which introduces Bidirectional Reflectance Distribution Function issues
Pulse length
length of time required to emit a pulse -> short = more precise distance measurement
Radiometric resolution
Ability of a sensor to discern between greyscale levels, defined in bits
Polarization
when waves aren’t distributed uniformly across all 360 degrees perpendicular to direction of propagation. Sunlight is weakly polarized, but becomes depolarized when it hits nonmetal objects.
geniometer
documents changes in sensor radiance, L, caused by changing position of sensor & or source of illumination
Pros/ Cons of suborbital viewing
Available on demand, higher resolution, available on demand
atmospheric turbulence can distort images & be difficult to correct for, expensive
Analogue image processing pros/ cons
humans are great at compiling lots of info when looking at an image
precise measurement isn’t as easy, analysis isn’t repeatable, can’t store data,
Radiometric correction of remote sensor data
noise or error introduced by the sensor may be able to be corrected for
Geometric correction of remote sensor data
processes information so images are in proper place for GIS applications
Hard vs. Fuzzy classification
hard = discrete mutually exclusive classes fuzzy = blend together
Scene modeling components
1) scene model -> form & nature of energy & matter within scene & their spatial and temporal order
2) Atmospheric model -> relationship between atmosphere and energy entering & leaving
3) Sensor model -> behavior of sensor in response to energy influxes
