Variant A

Variant A — mandatory Section II: Hyperspectral + Landsat; choose 2 more from Max Likelihood / IKONOS / GOES. Mandatory Section III: Landsat applications; choose 1 from Drone / Moderate resolution / Supervised.

Total: 140 points. Star (⭐) any question you want to carry into the final cram.

Section I — Multiple choice / matching

edit

2 pts In the EVI formula, the empirically-determined values are…

○ a. C1 = 6.0, C2 = 7.5, L = 1.0
○ b. C1 = 6.0, C2 = 1.5, L = 0
○ c. C1 = 1.0, C2 = 1.0, L = 1.0
○ d. C1 = 6.0, C2 = 7.5, L = 0

Reveal answer

Correct: C1 = 6.0, C2 = 7.5, L = 1.0

Model answer

MODIS-standard EVI coefficients.

💡

6 - 7.5 - 1 (six, seven-point-five, one). L ≠ 0 (that would kill the (1+L) multiplier).

edit

2 pts Stefan–Boltzmann: total emitted radiation from a blackbody scales with…

○ a. T⁴
○ b. T
○ c. T²
○ d. T³

Reveal answer

Correct: T⁴

Model answer

M = σT⁴. Small temperature changes = big radiance changes.

💡

Stefan has 4 letters. T⁴. Or: FOURTH power. Double temp → 16× radiance.

edit

2 pts On Landsat TM, Band 4 corresponds to…

○ a. NIR (0.76–0.90 µm)
○ b. Red (0.63–0.69 µm)
○ c. SWIR (1.55–1.75 µm)
○ d. Thermal (10.4–12.5 µm)

Reveal answer

Correct: NIR (0.76–0.90 µm)

Model answer

TM band order: 1 Blue, 2 Green, 3 Red, 4 NIR, 5 SWIR-1, 7 SWIR-2, 6 Thermal.

💡

TM Band 4 = NIR (for NDVI: (B4−B3)/(B4+B3)). Don't confuse with OLI Band 4 = Red (OLI shifted numbering by 1).

edit

2 pts Rayleigh scattering primarily affects…

○ a. Shorter wavelengths (why the sky is blue)
○ b. Longer wavelengths (why the sky is red)
○ c. All wavelengths equally
○ d. Only thermal IR

Reveal answer

Correct: Shorter wavelengths (why the sky is blue)

Model answer

Particles ≪ λ (air molecules); 1/λ⁴ dependence.

edit

2 pts Class C airspace typically extends to…

○ a. 4 000 ft AGL
○ b. 10 000 ft MSL
○ c. 2 500 ft AGL
○ d. 1 200 ft AGL

Reveal answer

Correct: 4 000 ft AGL

Model answer

Inner 5 NM ring SFC–4000 AGL; outer 10 NM shelf 1200–4000 AGL.

edit

2 pts Supervised classification REQUIRES…

○ a. A priori knowledge of land-use types and sample locations.
○ b. No training data — the computer infers classes.
○ c. Only an image and a decision rule.
○ d. A DEM and atmospheric profile.

Reveal answer

Correct: A priori knowledge of land-use types and sample locations.

Model answer

Unsupervised (ISODATA) is the one that needs no a priori knowledge.

💡

Supervised = teacher needed (training samples). If the question says 'needs NO training data' that's UNSUPERVISED (ISODATA).

edit

2 pts Part 107 minimum flight visibility is…

○ a. 3 statute miles
○ b. 1 statute mile
○ c. 5 statute miles
○ d. 10 statute miles

Reveal answer

Correct: 3 statute miles

Model answer

Measured at a slant from the control station.

💡

3 SM visibility. Pair with 500 ft below + 2000 ft horizontal cloud clearance — the '3-5-2' rule.

edit

2 pts Landsat 7 / 8 / 9 orbit at…

○ a. 705 km altitude, sun-synchronous, 16-day repeat
○ b. 832 km altitude, 26-day repeat
○ c. 35 786 km altitude, geostationary
○ d. 919 km altitude, 18-day repeat

Reveal answer

Correct: 705 km altitude, sun-synchronous, 16-day repeat

Model answer

919 km was Landsat 1–3; dropped to 705 km from L4 onward.

edit

2 pts In ISODATA, the **T** parameter is…

○ a. A convergence threshold — % of pixels unchanged between iterations.
○ b. Maximum number of iterations.
○ c. Number of clusters.
○ d. Spectral distance metric.

Reveal answer

Correct: A convergence threshold — % of pixels unchanged between iterations.

Model answer

N = max clusters, M = max iterations, T = convergence threshold.

💡

N-T-M: Number, Threshold, Max-iterations. T is a PERCENT (e.g., 95% pixels stable). M is an iteration COUNT. Don't swap T and M.

edit

2 pts The Sun's peak emission wavelength (Wien's Law, T ≈ 6000 K) is approximately…

○ a. 0.48 µm
○ b. 9.66 µm
○ c. 1.0 µm
○ d. 0.10 µm

Reveal answer

Correct: 0.48 µm

Model answer

λ_max = 2897.8 / 6000 ≈ 0.483 µm (visible green-blue).

💡

Sun peak = 0.48 µm (visible). Earth peak = 9.66 µm (thermal). Don't swap them. Big 9.66 goes with COOL Earth.

edit

2 pts TM's Band 6 (thermal) has a spatial resolution of…

○ a. 120 m
○ b. 30 m
○ c. 60 m
○ d. 100 m

Reveal answer

Correct: 120 m

Model answer

TM thermal was 120 m; ETM+ improved to 60 m; TIRS is 100 m.

💡

Thermal got BETTER over time: TM 120 → ETM+ 60 → TIRS 100 (resampled to 30). Don't confuse 30 m (VSWIR bands) with the much coarser thermal.

edit

2 pts What is the formula for NDVI?

○ a. (NIR − Red) / (NIR + Red)
○ b. NIR / Red
○ c. (Red − NIR) / (Red + NIR)
○ d. (NIR + Red) / (NIR − Red)

Reveal answer

Correct: (NIR − Red) / (NIR + Red)

Model answer

NDVI = (NIR − Red) / (NIR + Red). Range −1 to +1.

💡

NIR comes FIRST (it's bigger for vegetation). Difference on top, sum on bottom. NIR/Red is the Simple Ratio — a different index.

edit

2 pts Landsat MSS, TM, and ETM+ use which scanning geometry?

○ a. Across-track (whiskbroom, discrete detectors + mirror)
○ b. Along-track (pushbroom, linear array)
○ c. Staring array
○ d. Conical scan

Reveal answer

Correct: Across-track (whiskbroom, discrete detectors + mirror)

Model answer

Landsat 8/9 OLI is the first Landsat pushbroom.

edit

2 pts Who is credited with the first true vegetation index (Simple Ratio)?

○ a. Cohen (1991)
○ b. Rouse et al. (1974)
○ c. Anderson (1976)
○ d. Hord (1982)

Reveal answer

Correct: Cohen (1991)

Model answer

SR = NIR / Red — Cohen (1991).

💡

Cohen = SR (first VI). Rouse = NDVI (1974). Anderson = LULC classes (1976). Hord = Bayesian decision rule (1982). Don't mix these four.

edit

2 pts Clouds appear white because…

○ a. Non-selective scattering — water droplets ≫ wavelength, all wavelengths scattered equally
○ b. Rayleigh scattering favors blue
○ c. Mie scattering affects only IR
○ d. Clouds absorb all visible light

Reveal answer

Correct: Non-selective scattering — water droplets ≫ wavelength, all wavelengths scattered equally

Model answer

B, G, R scattered in equal proportion → white.

💡

'Non-selective' literally means ALL colors treated equally → white. Big particles (droplets) = non-selective. Tiny particles (molecules) = Rayleigh = blue sky.

Section II — Short answer

edit

10 pts mandatory Hyperspectral remote sensing — definition + example sensor + orbit altitude + band count.

Reveal answer

Model answer

Imagine taking a photo with hundreds of really narrow color filters all at once. Each pixel becomes a full rainbow signature of whatever’s at that spot.

That signature is unique enough to identify specific minerals, specific plant species, or pollutants — things that look identical with a normal camera but have distinct fingerprints when you slice the spectrum thinly enough.

The classic example is Hyperion on the EO-1 satellite — used in mining, agriculture, and pollution monitoring.

🔬 Show the science / technical version

Collects data in hundreds of narrow (5–10 nm) contiguous bands, producing a full reflectance spectrum per pixel.
Used to identify specific minerals, vegetation species, and pollutants via spectral matching.
Example: Hyperion on EO-1 — 220 bands, 0.4–2.5 µm, 30 m, 705 km orbit.
Trade-off: massive data volumes, low SNR per band → redundancy reduced via PCA / MNF before classification.

edit

10 pts mandatory Landsat — altitude, repeat, sensors, band cheat sheet.

Reveal answer

Model answer

Landsat is the granddaddy of Earth observation — running continuously since 1972, longer than any other civilian satellite series.

It captures the whole Earth every 16 days (or every 8 days now, with Landsat 8 + 9 working together) at moderate detail (30 m). The data is free, which is why it’s the backbone of every long-term land-change study you’ve heard of: Amazon deforestation, glacier retreat, urban sprawl, agriculture monitoring.

Modern sensors are OLI-2 (visible/NIR/SWIR) and TIRS-2 (thermal infrared).

🔬 Show the science / technical version

Longest continuous EO record (1972–present, USGS / NASA).
Orbit 705 km (L4 onward), sun-synchronous, 16-day repeat (L8 + L9 combined ≈ 8-day).
Modern sensors: OLI-2 (9 bands @ 30 m, pan @ 15 m) + TIRS-2 (2 thermal @ 100 m).
Heritage TM/ETM+ numbering: 1 Blue, 2 Green, 3 Red, 4 NIR, 5 SWIR-1, 7 SWIR-2, 6 Thermal.
NDVI: TM B4/B3 → OLI B5/B4 (numbering shifted by 1).

edit

10 pts choice Maximum Likelihood classifier — how it works and its main assumption.

Reveal answer

Model answer

Picture each class as a fuzzy cloud floating in color-space. Max Likelihood asks: “Which cloud is this pixel most likely from?”

Crucially, it accounts for each cloud’s shape — not just its center. A tight class is treated differently from a sprawling one.

The catch: it assumes the clouds are roughly Gaussian blobs. If your classes are weird-shaped (banana-shaped, donut-shaped), ML’s accuracy drops.

It’s the most popular supervised classifier — but only when the data fits the assumption.

🔬 Show the science / technical version

Assumes each class’s pixels follow a multivariate Gaussian distribution.
For each pixel, compute probability of membership in every class (from class mean + covariance).
Assign pixel to the class with the highest probability.
Strength: statistically principled, handles correlated bands.
Weakness: needs enough training samples to estimate covariance; struggles if classes aren’t Gaussian.

edit

10 pts choice IKONOS — operator, altitude, bands, radiometric resolution.

Reveal answer

Model answer

IKONOS was the first U.S. commercial sub-meter satellite — launched 1999, retired 2015. Sub-meter means you can pick out individual cars and rooftops from space.

Big deal because before IKONOS, that resolution was military-only.

It carries a high-detail black-and-white panchromatic band plus four lower-detail color bands, which you fuse together (“pan-sharpening”) to get color at the pan resolution.

It paved the way for today’s commercial high-res constellation — WorldView, GeoEye, Pléiades, SkySat.

🔬 Show the science / technical version

First commercial sub-meter satellite (Space Imaging, Sept 1999 – Mar 2015).
Altitude ~681 km, sun-synchronous.
Bands: Pan 0.45–0.90 µm at 1 m; 4 MS bands (B/G/R/NIR) at 4 m.
Radiometric 11-bit → 2 048 levels.
Off-nadir up to ±45°, swath ~11–13 km, revisit 1–3 days.

edit

10 pts choice GOES satellites — orbit, instrument, example use.

Reveal answer

Model answer

GOES = Geostationary Operational Environmental Satellite. NOAA’s weather satellites.

They sit way up high, parked over the equator, orbiting at exactly Earth’s rotation rate — so they appear stationary over one hemisphere all day.

That continuous stare is what makes them perfect for weather: hurricane tracking, severe storms, wildfires, fog, lightning. The newest generation (GOES-R series) takes a full-disk image every 10 minutes.

Spatial resolution is coarse (~½ km to 2 km), but you get time in exchange.

🔬 Show the science / technical version

NOAA weather satellite series.
Geostationary orbit ~36 000 km above the equator — continuously images the same hemisphere.
Current instrument: ABI (Advanced Baseline Imager) — 16 bands, full-disk every 10 min, CONUS every 5 min.
Used for hurricane tracking, wildfires, fog, storm prediction.

Section III — Explanation / essay

edit

15 pts mandatory Applications of the Landsat satellite series, with examples.

Reveal answer

Model answer

What Landsat is good for — basically watching the Earth’s surface change, decade after decade.

Forests — tracking deforestation in the Amazon, Borneo, Congo. Map every year, see what’s gone.
Cities — urban growth, sprawl, paving over farmland.
Agriculture — crop type maps, yield forecasts (track how green fields get during the season).
Water — reservoir levels, glacier retreat, algal blooms in the Great Lakes.
Disasters — wildfire burn scars, flood extent, volcanic activity.
Climate — long-term surface-temperature trends, snow-cover decline.

Why it works: Landsat has been running since 1972, the imagery has been FREE since 2008, and a single scene covers ~185 km × 170 km. Hard to find a serious land-change study that doesn’t lean on Landsat somewhere.

🔬 Show the science / technical version

Continuity since 1972 — decadal change detection unmatched by any other system.
Land-cover / land-use change — Amazon deforestation, urban growth, wetland loss.
Agriculture — crop-type mapping, yield forecasting via NDVI, irrigation monitoring.
Water — reservoir extent, glacier retreat, algal blooms (thermal + VSWIR).
Disaster response — wildfire burn scars, flood extent, volcanic activity.
Climate — long-term surface-temperature records, snow-cover trends.
Free & open data policy (since 2008) — the single biggest multiplier of Landsat’s scientific impact.

edit

15 pts choice Advantages and disadvantages of drone (UAS) technology, with examples.

Reveal answer

Model answer

Drones (UAS) — strengths and weaknesses for remote sensing.

Strengths

Centimeter-level detail — way beyond any satellite.
Fly on demand — your schedule, under cloud cover, on any day.
Cheap per flight — orders of magnitude less than tasking a satellite or chartering a plane.
Swap sensors freely — RGB, thermal, multispectral, LiDAR, hyperspectral.
Perfect for small AOIs — a farm field, a construction site, a wildlife habitat.

Weaknesses

Tiny footprint — covers only a few acres at a time, so a regional project means many flights.
Heavy regulation — FAA Part 107: stay under 400 ft, in line-of-sight, away from controlled airspace, daylight only.
Weather-dependent — wind grounds you, rain damages the gear, cold kills the battery.
Short flights — typically 20–40 minutes per battery.
Big files — gigapixel orthomosaics fill drives fast.
Privacy / safety — public reactions are mixed.

🔬 Show the science / technical version

Advantages

Cm-level spatial resolution — far beyond any satellite.
On-demand scheduling, under clouds, on your schedule.
Low cost per flight vs. aircraft or satellite tasking.
Flexible payloads — RGB, multispectral, thermal, LiDAR, hyperspectral.
Ideal for small AOIs: field-scale ag, construction, infrastructure, wildlife.

Disadvantages

Regulatory burden — FAA Part 107, altitude / line-of-sight / no-fly zones.
Weather-limited — wind, rain, cold batteries.
Small footprint → many flights to cover a region.
Battery endurance ~20–40 min per flight.
Data management — gigapixel orthomosaics strain storage & processing.
Privacy and public-safety concerns.

edit

15 pts choice Advantages and disadvantages of moderate-resolution sensors (10 – 250 m), with examples.

Reveal answer

Model answer

Moderate-resolution sensors — the workhorses of regional and global Earth observation. Roughly 10 m to 250 m per pixel.

Examples: Landsat (30 m), Sentinel-2 (10 m), MODIS (250 m+).

Advantages

Wide swaths — one pass covers an entire state.
Frequent revisit when you combine satellites — Landsat + Sentinel-2 ≈ every 2–3 days globally.
Long free archives — Landsat back to 1972, Sentinel-2 since 2015.
Solid band selection for vegetation indices, water indices, fire mapping.

Disadvantages

Can’t see individual objects — buildings, trees, single fields blur out.
Mixed pixels at edges — a 30 m pixel might be half-forest, half-grassland.
Clouds wreck many scenes — even with 16-day repeat, a lot of acquisitions are unusable.
Thermal is coarser still — Landsat thermal is 100 m, resampled to 30 m for display.

The trade-off: you sacrifice fine detail for wide coverage and frequent revisit. For continental-scale science, that’s the right trade.

🔬 Show the science / technical version

Examples: Landsat OLI (30 m), Sentinel-2 MSI (10–60 m), MODIS (250–1000 m).

Advantages

Large swath — Landsat 185 km, Sentinel-2 290 km — regional/global coverage.
Frequent revisit when combined (L8+L9+S2 ≈ every 2–3 days).
Long free archives (Landsat → 1972).
Multispectral breadth — NDVI, EVI, NBR, water indices.

Disadvantages

Can’t resolve individual objects (buildings, small fields, trees).
Mixed pixels at class boundaries hurt classification accuracy.
Cloud contamination — 16-day repeat means many scenes unusable.
Thermal resolution is coarser still (TIRS 100 m resampled to 30 m).

edit

15 pts choice Supervised classification — details, advantages, disadvantages.

Reveal answer

Model answer

Supervised classification — you teach the computer what each class looks like, then it sorts every pixel.

How it works

Find places you know — forest patches, water bodies, urban blocks.
Sample those — give the computer color signatures of each class.
Pick a decision rule — Max Likelihood is the most common.
Run it — every pixel in the scene gets a class label.
Check accuracy — compare to ground-truth points you didn’t use for training (confusion matrix, kappa).

Strengths

Higher accuracy than unsupervised when your training data is good.
You control the class scheme — useful when you have a specific question.
Works well when classes are spectrally distinct.

Weaknesses

Training-sample collection is slow — fieldwork, photo interpretation, ground truth.
Bias risk — if you forget a class, the computer can’t predict it.
Max Likelihood assumes Gaussian classes — fails on weird-shaped distributions.

🔬 Show the science / technical version

Procedure

Define training sites (AOIs) for each known class.
Compute class signatures (mean + covariance per band).
Apply a decision rule — Min Distance, Mahalanobis, Max Likelihood, Parallelepiped — to every pixel.
Accuracy assessment (confusion matrix, kappa).

Advantages

Analyst controls the class scheme — matches the question.
Higher accuracy than unsupervised when training data is good.
Works well when classes are spectrally distinct + well-sampled.

Disadvantages

Labor-intensive — good ground truth takes time.
Bias risk — unrepresented classes won’t be predicted.
Max Likelihood assumes multivariate-normal class distributions.

Back to practice tests · Final cram