exam Flashcards
what is the two passes of the chain method and what do they collect
1st pass spectral building- program reads through dataset and builds clusters.
the analysis provides 4 things radius, space dust par, number of pixels and max number of clusters.
what is remote sensing
remote sensing is the science of aQURINGING INFORMATION ABOUT THE EARTHS surface without actually being in contact with it. This is done by sensing and recording reflected or emitted energy and processing, analysing and applying that information.
what are the threee remote sensing detectors
RADAR radio detection and ranging
LIDAR light “ “
SONAR sound navigation “ “
often cheaper than in place sensing and collects data consistently and synoptically
what are some advantages of remote sensing
passive sensors are unobtrusive
no sampling bias and collect data systematically
only tech that can provide data over a very large area of geographic land
what are some limitations of remote sensing
often oversold
humans may select the wrong sensors
can become uncalibrated
often very expensive
what is ancillary/collateral data
additional information that is added to data and transported using the same form as the original. e.g. datasoil maps, dems, political boundaries
what is in-situ data
data not collected by the satellites
often necessary field techniques
calibrations accuracy with gps
often includes the use of transducers
what are sources of error for insitu data
intrusive error-shadow soil compactions, trampling
baised sampling design
improper operation of device
instrument not calibrated
what is the difference of passive and active sensors
active sensors emit energy passive receive reflected energy
what are the 4 different types of resolution
temporal
radiometric
spatial
spectral
what is spectral resolution
ability of a sensor to define fine wavelength intervals (bands)
multispectral- 8 bands
hyper spectral- 288 bands
ultra spectral- over hundreds
what is spatial resolution
size of area on the ground represented by one pixel worth of energy measurement.
defined in m
IFOV- instantaneous field of view
what is radiometric resolution
ability to discriminate very slight differences in energy
how many shades there are
2 vs 8 bit
what is temporal resolution
how often it records imagery of a particular area.
what are 4 examples of sensor platforms
insitu- cherry picker
airborne- AVIRIS
UAV- fly under clouds on demand but expensive, government restrictions
satelites
what is a analogue conversion
image- any pictorial representation
photograph- images recorded on film
pixel-smallest- 2-D element that is the smallest non-divisable element of digital image
what are digital numbers
the amount of radiance being measured at the sensor for that section of ground.
each pixel has its own digital number
8 bit os 255 range
aka brightness value
what is a scanning system
sensor with narrow swath for that sweeps over terrain to produce 2D image of surface
what are the two types of scanning systems and what are the main differences
across track/descrete detectors
rotating mirror to scan the Earth in lines perpendicular to the sensor’s motion.better quality but slower
along track/linear array
use forward motion of platform to record successive
what is an orbit
path followed by a satellite they can vary by altitude orientation and rotation relative to earth
what is a geospatial orbit
high altitude and matches earths orbit. collect data continuously over an area
mostly for camps and weather
what is a near polar orbit
north south in conjunction with earths orbit
northward has ascending shadows southward has sunlit side imaging
sun synchronis- cover each area at constant local time
what is a swath
the area imaged on the surface
what are the two main satellites
landsat and spot
spot is French and has a better spatial resolution
what are some univariate image stats
mode, median, mean, range, variance, std dev, skewness (asymmetry of histogram) kurtosis (weakness of curve)
what are some example of multivariate stats
covariance (joint variation of two variables)
Pearsons product moment correlation coefficient.
what is density slicing
conversion of the continuous tone of an image to a series of discrete class intervals or slices
can be either natural breaks, equal size or equal area intervals
what is a false colour composite
when the rib bands are replaced with different bands and looks different to irl
also colour infrared 4,3,2
what are the two main geoprocessing techniques
geometric correction
radiometric correction
what is geometric correction
placing the reflected, emitted or backscatter measurements in their proper map locationh
what is radiometric correction
improving the accuracy of surface specular reflectance, emittance or backscatter measurements obtained using a RS system
what are the two types of errors that come from imagery
internal- sensor itself predictable and constant
external- atmosphere and scene changes. use GCP
how does radiometric correction correct the sensor system detection error
random bad pixels (randomly distributed pixels with no value. to fix use mean of pixels around it
partial line or column dropouts (LANDSAT 7)
line start problems failed to correct pixel at beginning of the scan line
n-line striping sensor out of adjustment
radiometric correction atmospheric correction methods of correction
model atmosphere
model with in-situ data-local characteristics (absolute)
minimisation-using multiple looks at the same object (relative)
what is target and path radiance
path 1 very little attenuation
path 2 never reaches the surface
path 3 Rayleigh or mie scattering
path 4 reflected and scattered into IFOV
path 5 adjacent area reflected up and bounced back to study area
what is the equation for radiance
Ls(total radiance at sensor)=Lt(total radiance from target + Lp(path radiance 9TRY TO REMOVE THIS))
what are the three types of scattering
Rayleigh- gas molecules
mie- aerosols
non selective- closest to surface
what is the empirical approach to radiometric normalisation
using targets that do not change between dates
what is the main use feature of geometric correction
ground control points
what is spatial interpolation
geometric relationship between input pixel location (row/column) and associative map coordinate of same point (x,y)
then relocate every pixel in original (x,y) to proper position in output image
what are the three methods of intensity interpolation
nearest neighbour
bilinear interpolation (looks at 4 nearest pixels and gives BV based on weighted distances)
cubic convolution (same as above but uses 16 surrounding pixels
what is an enhancement
The alteration of the appearance of an image in a way that the info is easily interpreted by the viewer
what are the two different operations in relation to enhancements
point operations (modify each pixels BV independently
local operations (modify each pixel in context of the surrounding BV more fuzzy)
difference between image reductions and magnifications
reductions zoom out 36 square to 9 square
modifications zoom in multiple pixels form from the same original pixel
what is a contrast enhancement
alters the range of BV
shows greater difference in small BV differences
what is a linear contrast enhancement
only considers max and min
best for normal distribution
start of rise happens at start of bell curve then at end of bell curve all numbers are 255
what are the two non linear enhancements
histogram equalization
logarithmic (greatest impact on darker BV values)
what is spatial filtering
method for selectively emphasising or suppressing info at different spatial scales
what is low frequency filtering
local sharp fluctuations are removed. de-emphases the high frequency details.
what are the types of low frequency filtering
moving average filter (box the outliers)
median filter(create a box around number and pick median of that box
what is high frequency filtering
enhance high frequency local variations
what is band ratioing
reduces the effect of BV’s from identical surface material
what is a vegetation indexes
measure vegetation amount and condition
types of vegetation indices
SR NDVI
Sr is better than NDVI for separating high end biomass but not as good for low end biomass
what are the 5 types of classification
hard crisp classification (create a map with descrete categories)
Fuzzy set (soft) classification (considers heterogeneity and imprecise nature of the world)
Geographic object based image analysis (GEOBIA) not a per pixel classification
Numeric methods- Numeral networks
Hybrid with ancillary data
what is a training site
homogeneous area representing a specific land cover or land use class of interest
what is feature selection
select the bands that best asssist in discriminating classes
what is the difference between commission and omission error
commission- pixel assigned to a class to which it does not belong
omission error pixel not assigned to a class to its appropriate class
what is parallelepiped classification
boolean logic cream mean vectors for each class in each band then use std dev for class selection
what is minimum distance to means classification
euclidean distance to the closest class
nearest neighbour classification
use actual training data BV in each band and nearest neighbour distance measures aka kmeans
max likelihood classification algorithm
assumes norm distance based on probabilities
determines probabilities for each class then assigns them
what is the isolate classification method
self organised
select c max clusters numbers t max, M number of times it classifies pixel
what is fuzzy classification
based on imprecise geographic info. hard classifier (yes=1, no=0)
what is GEOBIA
2 procedures
first is object oriented image segmentation
second is traditional classification to the objects
