Remote sensing Study Guide

📖 Core Concepts Remote sensing – gathering information about an object or phenomenon without physical contact, using detected electromagnetic (EM) energy. Active sensor – carries its own energy source (e.g., radar, LIDAR) and measures the returned signal. Passive sensor – records naturally occurring EM radiation (e.g., reflected sunlight, emitted thermal energy). Resolution types – spatial (smallest distinguishable object), spectral (number/width of wavelength bands), radiometric (intensity gradations), temporal (revisit frequency). Orbit families – Low‑Earth Orbit (LEO, polar), Sun‑synchronous (constant solar angle), Geostationary (fixed over one point, ideal for weather). Key products – NDVI, digital elevation models (DEMs), analysis‑ready data (ARD). 📌 Must Remember Active vs. Passive – Active: emits → measures return (SAR, LIDAR). Passive: detects ambient EM (optical, thermal). Frequency bands – Visible, infrared, microwave, gamma‑ray (UV rarely used). Spatial resolution – Determines minimum object size detectable (e.g., Landsat 30 m, Sentinel‑2 10 m). Spectral resolution – More bands → better material discrimination (hyperspectral > multispectral). Temporal resolution – LEO polar: revisit ≈ 16 days (single satellite), Sun‑synchronous constellations reduce to 2–3 days. Confusion matrix metrics – Overall accuracy, Producer’s accuracy, User’s accuracy; derived from error matrix. NDVI formula – $NDVI = \frac{NIR - Red}{NIR + Red}$ (range ‑1 to +1). 🔄 Key Processes Data acquisition Choose platform (satellite, aircraft, drone) → sensor type (active/passive) → schedule based on temporal needs. Pre‑processing Radiometric correction → remove sensor noise. Atmospheric correction → compensate scattering/absorption. Georeferencing → align image to ground coordinates using control points. Image analysis Visual interpretation or automated classification (pixel‑based vs. object‑based). Compute indices (e.g., NDVI) → derive vegetation health, crop stress. Accuracy assessment Collect reference data → build confusion matrix → compute accuracy metrics. Product generation Generate DEMs (e.g., InSAR), land‑cover maps, time‑series cubes (ARD). 🔍 Key Comparisons SAR vs. LIDAR – SAR: microwave, penetrates clouds, works day/night, provides surface roughness & interferometric DEMs. LIDAR: laser (near‑infrared), requires clear sky, yields high‑precision 3‑D point clouds. Pixel‑based vs. Object‑based classification – Pixel: uses spectral values per pixel; prone to speckle in high‑resolution data. Object‑based: groups pixels into meaningful objects, incorporates shape, texture, context → higher accuracy for heterogeneous scenes. LEO polar vs. Sun‑synchronous – Polar: any local time, broader revisit window. Sun‑synchronous: same solar illumination each pass → ideal for change detection. Geostationary vs. LEO – GEO: constant view of large area, coarse spatial resolution, excellent for real‑time weather. LEO: finer spatial detail, global coverage, intermittent view. ⚠️ Common Misunderstandings “Higher spatial resolution always means better data.” – Fine resolution may increase noise; for some analyses (e.g., climate trends) coarse, frequent data are more valuable. “Passive sensors can’t work at night.” – Thermal infrared is passive but emits from the Earth itself, enabling night‑time observations. “All satellites have the same revisit time.” – Revisit depends on orbit, swath width, and constellation size. “More spectral bands = higher classification accuracy automatically.” – Redundant bands can add noise; proper band selection matters. 🧠 Mental Models / Intuition “Sensor as a camera + lens” – The sensor records EM energy (camera) and the platform’s altitude & viewing angle act as the lens determining resolution and coverage. “Resolution trade‑off triangle” – You can improve two of three: spatial, spectral, temporal – the third will suffer (e.g., high spatial + high spectral → lower temporal). “Active = “ping”; Passive = “listen.” – Active sensors send a signal and listen for echo (like sonar); passive sensors only listen to what the target already emits/reflected. 🚩 Exceptions & Edge Cases Ultraviolet sensors – Rarely used for Earth observation due to atmospheric absorption; only useful for specific atmospheric studies. Microwave penetration – SAR can see through vegetation canopy and, under dry conditions, shallow soil; not true for all frequencies. Sun‑synchronous drift – Over years, the local solar time drifts slightly; corrections may be needed for long‑term change detection. 📍 When to Use Which Mapping flood extent – Use passive optical (high spatial) when cloud‑free; switch to SAR (active) if clouds present. Estimating canopy height – Prefer LIDAR point clouds for precise 3‑D structure; use InSAR for large‑area elevation change when LIDAR unavailable. Crop health monitoring – NDVI from multispectral (Sentinel‑2) for regular phenology; hyperspectral if specific pigment discrimination is required. Urban change detection – High‑resolution optical (WorldView) for detail; SAR for structural change regardless of weather. 👀 Patterns to Recognize Spectral signature “V‑shape” in NDVI → healthy vegetation (high NIR, low Red). Speckle noise pattern in SAR images → indicates need for multilooking or filtering. Linear phase ramps in interferograms → signal of topographic change (use for DEM generation). Consistent temporal spikes in thermal bands → possible fire or industrial heat source. 🗂️ Exam Traps Confusing active vs. passive – A question may describe a sensor that measures returned laser pulses (LIDAR) but label it “passive”; remember any emitted energy makes it active. Mixing up resolution types – “Higher temporal resolution means finer spatial detail” is false; they trade off. NDVI sign errors – NDVI cannot be > 1; a value > 1 indicates a calculation mistake (wrong band order). Orbit misconceptions – Assuming a GEO satellite can provide 10 m resolution; GEO platforms typically have coarse resolution (> 1 km). Accuracy metrics – Selecting “overall accuracy” as the sole metric for imbalanced classes; you should also cite producer’s and user’s accuracies. --- Study tip: Turn each bullet into a flashcard. Review the mental models and patterns right before the exam to let the concepts click quickly. Good luck!