GISCI 241 Principles of Remote Sensing

How GISCI 241 is assessed

The five numeric components sum to 100%. The published course outline states the exam weight but leaves its timing and operating details to later Canvas confirmation.

What GISCI 241 covers

Follow the course from sensing physics and spatial representation through aerial imagery, resolution, radar and LiDAR, then apply those foundations to classification, change detection, time series and drone surveys.

It then asks how raster and vector representations, scale, projection, image geometry and four different meanings of resolution affect what can be measured. Later modules apply those foundations to supervised and unsupervised classification, neighbourhood analysis, change detection, dense time series and structure from motion.

Practical work uses Google Earth Engine and related spatial workflows to produce images, indices, classifications, elevation products and maps.

The assessment design makes communication as important as computation: five assessed labs require visible outputs and short explanations; three quizzes test conceptual and technical understanding; a group video links remote sensing to a Sustainable Development Goal; and an individual poster project combines rationale, method, results and cartographic presentation.

The final examination is worth 30%, but the published course outline leaves its timing and operating format to later Canvas confirmation. This guide therefore focuses on the content and reasoning that are actually published. Each chapter separates observation from interpretation, works an original numerical or method example, and ends with practice that states what a result can and cannot support.

Worked example · free

Compare pixel support and radiometric levels

Key terms

Remote sensing

Obtaining information about an object or place without direct contact by measuring reflected, emitted or returned energy.

Spectral signature

The pattern of a surface's response across wavelength bands, used as evidence for discrimination under stated conditions.

Spatial resolution

The ground detail represented by an image, commonly expressed through pixel size or resolvable feature size.

Spectral resolution

The ability of a sensor to distinguish narrow wavelength intervals through the number and width of its bands.

Radiometric resolution

The sensitivity of a sensor to intensity differences, often represented by the number of digital bits.

Temporal resolution

The frequency with which a sensor can observe the same location, distinct from the frequency of usable cloud-free data.

Backscatter

The portion of transmitted radar energy returned toward the sensor after interacting with the surface.

Point cloud

A collection of three-dimensional sample points produced by ranging or photogrammetric reconstruction.

Ground control point

A stable visible location with known coordinates used to place imagery or a model into geographic space.

Training sample

A labelled set of representative pixels used to characterise classes in supervised classification.

Change vector

The signed multiband difference between observations at two dates, whose length summarises change magnitude.

Structure from motion

A photogrammetric method that reconstructs camera geometry and three-dimensional structure from overlapping images.

GISCI 241 FAQ

Questions students ask in each GISCI 241 chapter

Is a remote sensor just a camera?

No. Some sensors record reflected visible light, but others measure infrared emission, microwave backscatter or laser travel time. Even optical instruments sample defined wavelength bands and convert radiance into digital values. The displayed picture is a product of that measurement and rendering chain, not a direct copy of human vision. Chapter 01 · Remote Sensing Foundations and Platforms

When should I use raster instead of vector?

Use raster when the phenomenon varies continuously across space or when cell-by-cell arithmetic is central, such as elevation, temperature or multispectral imagery. Use vector when discrete objects, exact geometry, attributes or network relationships dominate. Many workflows combine both, but every conversion changes support, boundaries or interpolation. Chapter 02 · Spatial Data, Scale and Projections

What makes a good control point?

A good point is stable through time, sharply recognisable in both image and reference, accurately located and distributed with other points around the area of interest. Avoid shadows, moving shorelines, vegetation edges and ambiguous corners. Reserve independent features or surveyed points to test the fitted transformation. Chapter 03 · Aerial Photos and Image Interpretation

Why are wavelength and frequency inverse?

Electromagnetic energy travels at a fixed speed in the stated relationship, so more wave cycles must fit into each second when each cycle is shorter. The course expresses this as c equals wavelength times frequency. Keep units consistent and perform a direction check after every calculation. Chapter 04 · Electromagnetic Radiation and Multispectral Images

What does high resolution mean?

The phrase is incomplete unless it names a dimension. Small pixels mean finer spatial sampling, narrow bands mean finer spectral sampling, more digital levels mean finer radiometric encoding, and short revisit means finer temporal sampling. A dataset can be strong in one dimension and weak in another. Chapter 05 · The Four Dimensions of Resolution

Why can radar observe through cloud?

Many radar systems use microwave wavelengths that interact less with ordinary cloud droplets than visible light does, allowing observations during cloudy conditions and at night. Penetration depends on wavelength and target conditions, so radar is not unaffected by atmosphere or precipitation and does not see through every material. Chapter 06 · Active Sensing and Radar

Can LiDAR see through a forest canopy?

Some pulses pass through canopy gaps and produce later returns from lower vegetation or ground. Dense canopy can still block ground observations, so terrain quality varies with canopy, footprint, scan angle and pulse density. A ground model under forest may depend heavily on sparse returns and interpolation. Chapter 07 · LiDAR Returns and Elevation Models

When is supervised classification preferable?

Use it when meaningful target classes are known and representative training areas can be collected across their variability. It gives control over the legend but can fail when training omits locations, seasons or mixed conditions. Keep independent reference data out of training so transfer can be tested honestly. Chapter 08 · Supervised and Unsupervised Classification

What is the difference between local and focal analysis?

A local operation uses the target cell and aligned values at that location, such as band arithmetic or image subtraction. A focal operation uses surrounding cells in a window, such as a mean, median or edge kernel. Window size sets the spatial scale and must be reported. Chapter 09 · Raster Analysis and Change Detection

Why is a time series better than two dates?

Additional observations reveal baseline variation, event timing, seasonality, trend and recovery. Two endpoints can miss a sharp event followed by partial recovery or mistake a seasonal difference for long-term change. The time series adds context, provided acquisition and preprocessing remain comparable. Chapter 10 · Time Series and Disturbance Signals

Why does structure from motion need overlap?

The same feature must appear in several photographs from different positions so parallax can constrain camera geometry and three-dimensional location. Coverage with only one or two weak views leaves holes or unstable depth. Front, side and cross-flight overlap provide redundancy for matching and reconstruction. Chapter 11 · Drone Surveys and Structure from Motion

How to study for the exam

Study in linked passes. First draw the measurement chain for optical, radar, LiDAR and drone photogrammetry, naming what energy or geometry creates the data. Second build a comparison table for raster and vector, vertical and oblique views, the four resolutions, supervised and unsupervised classification, local and focal analysis, and two-date versus time-series change.

Third work the small calculations by hand: wavelength and frequency, pixel area, bit-depth levels, surface-minus-terrain height, focal means and multiband change magnitude. After every number, write the physical direction and one limitation.

Use the lab question types as practice without copying their stems: band combinations, indices, collection filtering, mean versus median composites, classification comparison and elevation differencing. For the poster, audit rationale, method, results and presentation separately.

Before the final examination, check Canvas and the official timetable because the published outline gives the 30% weight but not the operating format. Then revise the full module chain and practise closing each answer with validation and uncertainty.