GISP Remote Sensing and Image Analysis 2 — Questions and Answers
Question 1: Which classification approach assigns each pixel to a land cover class based purely on spectral statistics without using any training samples?
- Supervised classification
- Unsupervised classification (Correct answer)
- Object-based image analysis
- Decision tree classification
Correct answer: Unsupervised classification
Unsupervised classification algorithms like ISODATA and K-means group pixels into spectral clusters automatically without analyst-provided training data.
Question 2: What does the term 'geometric correction' in image processing address?
- Removing atmospheric haze from imagery
- Correcting distortions caused by sensor tilt, Earth curvature, and terrain relief (Correct answer)
- Adjusting brightness and contrast
- Filtering speckle noise from radar images
Correct answer: Correcting distortions caused by sensor tilt, Earth curvature, and terrain relief
Geometric correction involves transforming an image to match a map projection by correcting for sensor geometry, platform motion, Earth rotation, and topographic displacement.
Question 3: LiDAR point clouds classified as 'ground returns' are primarily used to generate which type of surface model?
- Digital Surface Model (DSM)
- Digital Terrain Model (DTM) (Correct answer)
- Canopy Height Model (CHM)
- Normalized Digital Surface Model (nDSM)
Correct answer: Digital Terrain Model (DTM)
Ground-classified LiDAR returns represent bare-earth elevation and are interpolated to create a Digital Terrain Model that excludes vegetation and structures.
Question 4: In image classification accuracy assessment, what does the producer's accuracy measure?
- The probability that a classified pixel actually represents that class
- The probability that a reference site is correctly classified on the map (Correct answer)
- The overall percentage of correctly classified pixels
- The disagreement due to quantity differences between maps
Correct answer: The probability that a reference site is correctly classified on the map
Producer's accuracy (recall) measures how well reference sites of a given class are identified correctly on the classified map, indicating errors of omission.
Question 5: What is the purpose of pan-sharpening in remote sensing image processing?
- Removing noise from panchromatic bands
- Merging high-resolution panchromatic data with lower-resolution multispectral data (Correct answer)
- Converting thermal bands to visible wavelengths
- Correcting for bidirectional reflectance effects
Correct answer: Merging high-resolution panchromatic data with lower-resolution multispectral data
Pan-sharpening fuses a high spatial resolution panchromatic band with coarser multispectral bands to produce a color image at the higher spatial resolution.
Question 6: Which radar backscatter characteristic makes SAR (Synthetic Aperture Radar) particularly useful for flood mapping?
- High backscatter from open water surfaces
- Very low or specular backscatter from smooth water surfaces (Correct answer)
- Thermal emission from inundated vegetation
- Increased coherence over flooded areas
Correct answer: Very low or specular backscatter from smooth water surfaces
Open water acts as a specular reflector in SAR imagery, directing microwave energy away from the sensor and resulting in very dark (low backscatter) pixels that are easily distinguished from land.
GISP Remote Sensing and Image Analysis 2
Which classification approach assigns each pixel to a land cover class based purely on spectral statistics without using any training samples?