LESSON 12.2 — Remote Sensing and GIS

A. Standard Map

B. Mechanism in Words

1. Define the spatial question — identify whether the planning task requires (a) synoptic area coverage of a continuously varying phenomenon (RS is primary), (b) spatially precise discrete feature mapping (vector GIS with field survey), or (c) 3D urban form data (LiDAR); most plans require all three in combination.
2. Select sensor and platform — match spatial resolution to minimum mapping unit (building = <1m; ward land use = 5–10m; regional = 30m+); match temporal resolution to rate of change; verify orbit type provides required revisit and illumination consistency.
3. Acquire and preprocess imagery — apply radiometric correction (convert digital numbers to reflectance values), geometric correction (orthorectification using DEM), and atmospheric correction (remove scattering/absorption effects) before analysis.
4. Classify and extract information — apply supervised or unsupervised classification to assign every pixel to a land cover category; validate accuracy against field samples (minimum 80% overall accuracy for planning-grade LULC maps); compute spectral indices (NDVI for vegetation, NDBI for built-up).
5. Build the GIS database — import classified RS outputs as raster layers; digitise or import vector features (roads, boundaries, parcels); georeference all layers to a common coordinate system; link attribute tables.
6. Run spatial analysis — execute task-specific GIS operations: buffer zones for proximity analysis, overlay for constraint mapping, network analysis for accessibility, weighted overlay for suitability scoring.
7. Validate and integrate into plan — cross-check GIS outputs against field surveys and secondary data; present results as thematic maps aligned with URDPFI 2015 planning phases; document metadata for reproducibility.

C. Core Concept Explanations

C1. Remote Sensing Basics — Passive vs Active; EM Spectrum; Resolution Types

Active vs Passive Sensing

Electromagnetic Spectrum — Key Bands for Planning

Four Resolution Parameters

C2. Satellite Platforms — ISRO, Landsat, SPOT, IKONOS, WorldView

Indian Platforms (ISRO)

International Platforms

C3. GIS Data Models — Raster vs Vector

A GIS represents geographic space using one of two data models. The choice is not aesthetic — it determines what analysis is possible and what accuracy is achievable.

C4. GIS Operations — Buffer, Overlay, Network Analysis, Suitability Analysis

C5. LiDAR — Point Cloud, DEM/DSM, Urban Feature Extraction

LiDAR (Light Detection and Ranging) is an active remote sensing technology that emits rapid laser pulses (typically NIR, 1064nm) and measures the time for each pulse to return from the target. Distance = (speed of light × time) / 2. Multiple return pulses from a single emitted beam allow simultaneous capture of the top of a tree canopy (first return) and the ground beneath (last return).

LiDAR Workflow in Urban Planning:

Aerial LiDAR flight (airborne platform or UAV)
        ↓
Raw point cloud (x, y, z + intensity + return number)
        ↓
Point cloud classification
    - Ground points → DEM (bare earth)
    - Vegetation points → canopy height model
    - Building roof points → DSM
        ↓
nDSM = DSM − DEM  (normalised Digital Surface Model)
    = height of objects above ground
        ↓
Building height extraction: nDSM values over building footprints
        ↓
Urban planning outputs:
    - Building height map (FAR verification)
    - Tree canopy cover (green space assessment)
    - Flood depth modelling (DEM + water level)
    - Solar potential mapping (rooftop slope/orientation from DSM)

C6. Planning Applications — Land-Use Change, Sprawl Mapping, Suitability

Land-Use Change Detection
Multi-temporal satellite imagery (same sensor, same season, different years) is classified independently and then compared pixel-by-pixel. Change matrix shows which land use categories converted to which others.

Urban Sprawl Mapping
Sprawl is identified by comparing urban built-up extent across census years using classified satellite imagery. The Shannon Entropy index (derived from land use distribution across zones) quantifies sprawl: higher entropy = more dispersed, fragmented growth.

Key satellite data for India: Landsat archive (1972–present, free) provides longest available time series. Sentinel-2 (2015–present) provides higher temporal frequency (5-day revisit) at moderate resolution (10m).

Suitability Analysis Workflow
A weighted overlay suitability analysis for a new settlement zone typically follows:

1. Identify criteria (slope, flood risk, distance from existing infrastructure, land cost, agricultural quality)
2. Convert each criterion to a raster layer with a standardised score (e.g., 1–5, where 5 = most suitable)
3. Assign weights to each criterion based on planning priorities (AHP or expert judgement)
4. Compute weighted sum: Suitability = Σ(weight_i × score_i)
5. Classify the output into suitability zones (High, Medium, Low, Excluded)
6. Validate against field reconnaissance and stakeholder input

GIS Applications Aligned to URDPFI 2015 Phases

D. Worked Numericals and Parameter Tables

No NAT questions are prescribed for Lesson 12.2. Two reference tables fulfil Section D.

Table D1 — Raster vs Vector Decision Table

Table D2 — Satellite Resolution Comparison by Planning Task

E. Common Confusions

• LiDAR is always active — not passive, not satellite-only. LiDAR emits laser pulses; it does not depend on sunlight. Airborne and UAV-mounted LiDAR are more common for urban mapping than satellite LiDAR (ICESat-2 is a profiling instrument, not an area-mapping tool).
• Network analysis cannot be performed on a raster. Road routing, service area calculation, and O-D matrix analysis require topologically connected vector line networks with traversal costs. Raster cost-distance is an approximation for regional analysis only — it does not respect turn restrictions, one-way streets, or signal delays.
• DSM ≠ DEM. DSM includes building roofs and tree canopy. DEM is bare ground only. The difference (nDSM = DSM − DEM) gives object height above ground. Using DSM for drainage modelling produces incorrect flow paths because buildings appear as terrain obstacles.
• NDVI is not a land use map. NDVI measures relative vegetation vigour (NIR − Red) / (NIR + Red). A high NDVI pixel is vegetated — but it could be a park, farm, forest, or vegetated road median. Converting NDVI directly to a land use map without further classification is a methodological error.
• Spatial resolution ≠ accuracy. A 0.25m image is not automatically more accurate than a 10m image — accuracy depends on geometric correction, atmospheric correction, and classification method. A well-processed 10m Sentinel-2 scene may have higher classification accuracy than a poorly processed 0.5m image.
• SSO ≠ GEO. Sun-Synchronous orbit (700–800 km) provides consistent illumination and is used for change detection. Geostationary orbit (~36,000 km) provides continuous monitoring of a fixed region and is used for weather and disaster warning. They serve different planning functions — never substitute one for the other.

F. Exam Traps

G. Answer-Writing Cues

MCQ / MSQ — Active vs passive sensor identification:

MCQ — GIS operation selection:

MSQ — Satellite selection for a given task:

MCQ / MSQ — LiDAR product identification:

H. PYQ Linkage Note

I. Mini-Check — Lesson 12.2

Q1. (MSQ — select ALL correct) Which of the following statements about remote sensing sensors are correct?

(A) SAR (Synthetic Aperture Radar) can image through cloud cover during monsoon season
(B) LiDAR depends on reflected sunlight and cannot be used at night
(C) Landsat 8 OLI is a passive multispectral sensor operating in Sun-Synchronous orbit
(D) INSAT-3D in Geostationary orbit is suitable for multi-temporal urban land use change detection
(E) Thermal infrared sensors are classified as passive sensors

Answer: A, C, E

Explanation: (A) SAR is active, microwave — cloud-penetrating. Correct. (B) LiDAR is active — emits its own laser pulses; operates day and night. Incorrect. (C) Landsat 8 OLI is passive multispectral in SSO. Correct. (D) INSAT-3D in GEO provides continuous coverage of the same region at ~4km spatial resolution — not suitable for city-level land use change detection which requires 5–30m resolution and temporal archive. Incorrect. (E) Thermal IR sensors record surface-emitted radiation without their own energy source — passive. Correct.

Q2. (MSQ — select ALL correct) A planner needs to perform a suitability analysis to identify land parcels for a new affordable housing project. The criteria are: slope gradient, distance from proposed metro station, flood risk zone, and existing land use. Which of the following statements about the GIS methodology are correct?

(A) All criterion layers must be in raster format with identical cell size before weighted overlay can be applied
(B) Network analysis is the correct GIS operation to compute distance from the metro station to each parcel
(C) The flood risk zone layer (a vector polygon) must be rasterised before inclusion in the weighted overlay
(D) A buffer operation around the metro station can generate a proximity score layer for the weighted overlay
(E) Weighted overlay requires all input layers to be in vector format

Answer: A, C, D

Explanation: (A) Weighted overlay (raster map algebra) requires all layers in raster format with matching extent and cell size. Correct. (B) Distance from metro station to parcels is a Euclidean proximity (buffer/distance raster) task, not a network analysis task. Network analysis computes travel time/distance along roads — the criterion here is proximity, not routed distance. Incorrect. (C) Vector flood polygon must be rasterised to participate in raster map algebra. Correct. (D) A buffer generates a zone of distance; converting it to a scored proximity raster (e.g., 5 = within 300m, 1 = beyond 1,500m) is a valid approach to generate the criterion layer. Correct. (E) Weighted overlay is a raster operation — not vector. Incorrect.

Q3. (MCQ) A planner wishes to extract building heights for an entire city to verify FAR compliance. After obtaining airborne LiDAR data, the first step is to classify the point cloud into ground and non-ground points, producing a DSM and a DEM. What derived product directly represents building heights above ground?

(A) DTM
(B) TIN
(C) nDSM (DSM − DEM)
(D) NDVI

Answer: (C) nDSM (DSM − DEM)

Explanation: nDSM (normalised Digital Surface Model) = DSM − DEM. DSM records the top of all surface objects (building roofs, tree canopy). DEM records bare ground elevation. Their difference isolates the height of above-ground objects. Over building footprints, nDSM values equal building height above terrain — directly applicable for FAR and height-limit compliance checking. DTM is another name for bare-ground terrain (similar to DEM). TIN is a terrain representation format. NDVI is a vegetation spectral index with no relationship to building heights.

Q4. (MCQ) Which satellite orbit type ensures that a sensor passes over the same latitude at the same local solar time on every revisit, enabling consistent illumination conditions for multi-temporal land use change detection?

(A) Geostationary (GEO)
(B) Polar orbit
(C) Sun-Synchronous orbit (SSO)
(D) Low Earth Orbit (LEO) with random inclination

Answer: (C) Sun-Synchronous orbit (SSO)

Explanation: In SSO (700–800 km altitude), the orbital plane precesses at the same rate as Earth orbits the Sun, ensuring the satellite passes over a given latitude at the same local solar time on every pass. This guarantees consistent illumination angle and shadow geometry across all images in a time series — essential for valid multi-temporal comparison. GEO provides continuous coverage of one region but at ~4km spatial resolution — unsuitable for urban LULC mapping. Polar orbit achieves global coverage but does not maintain consistent solar time. (GATE 2017)

Q5. (MCQ) A GIS analyst attempts to compute the shortest travel route between two hospitals using a raster DEM layer as the input network. This approach is technically flawed because:

(A) DEMs store elevation data in floating point format which network solvers cannot read
(B) Raster data has no topological connectivity — it cannot represent road junctions, one-way constraints, or turn costs required for network analysis
(C) Network analysis requires data in GEO coordinate system, not projected coordinates
(D) The DEM resolution of 30m is too coarse for routing calculations

Answer: (B)

Explanation: Network analysis (shortest path, service area) requires a topologically connected vector line network where edges carry impedance values (travel time, distance) and nodes represent junctions with turn restrictions. A raster DEM has no topology — it stores elevation values in a grid of cells with no connectivity rules, no direction attributes, and no junction logic. The conceptual mismatch is the data model, not the resolution, coordinate system, or data type. Option (D) is a secondary practical concern but is not the fundamental flaw — even a 1m DEM would fail for the same structural reason.