GIS Fundamentals: A First Text on Geographic Information Systems

History, definitions, components, and applications of GIS

Geographic Information Systems (GIS) represent one of the most transformative technologies of the modern era — a convergence of cartography, database management, spatial statistics, and computer science into a single powerful framework for understanding our world through the lens of location.

The origins of GIS trace back to the 1960s when Roger Tomlinson developed the Canada Geographic Information System (CGIS) for the Canadian government to inventory natural resources. This pioneering work established the fundamental concept that geographic data could be stored, analyzed, and retrieved using computers. Around the same time, Howard Fisher at Harvard Laboratory for Computer Graphics created SYMAP, one of the earliest computer mapping programs.

A GIS consists of five essential components: (1) Hardware — computers, GPS receivers, digitizers, printers; (2) Software — ArcGIS, QGIS, GRASS GIS, PostGIS; (3) Data — spatial and attribute data in vector and raster formats; (4) People — analysts, managers, developers, and end-users; (5) Methods — workflows, analysis procedures, and organizational protocols. Understanding this ecosystem is crucial because GIS is not merely software — it is a complete system for spatial decision-making.

Modern GIS applications span virtually every industry. In urban planning, GIS models traffic flow, zoning compliance, and infrastructure capacity. In public health, disease surveillance systems track outbreaks using spatial clustering algorithms. In agriculture, precision farming uses GPS-guided tractors and satellite-derived vegetation indices to optimize fertilizer application field-by-field. In emergency management, GIS coordinates evacuation routes, resource deployment, and damage assessment. In retail, location analytics determines optimal store placement using demographic and competitor proximity data.

The distinction between GIS as a tool versus GIS as a science (GIScience) is important. GIScience investigates fundamental questions about spatial representation, spatial cognition, uncertainty propagation, and the societal implications of geospatial technology. Michael Goodchild coined the term "GIScience" in 1992, arguing that the technology raises scientific questions that deserve systematic study.

Data quality is a persistent challenge in GIS. Positional accuracy describes how closely coordinates match real-world locations. Attribute accuracy measures whether descriptive data is correct. Completeness assesses whether all features are represented. Logical consistency checks for topological errors like polygon gaps or overlaps. Temporal accuracy considers whether data reflects current conditions. Understanding these dimensions prevents costly errors in spatial analysis — a flood risk map built on inaccurate elevation data could endanger lives.

• GIS is a complete system (hardware + software + data + people + methods), not just software
• The field evolved from 1960s government mapping to today's ubiquitous location intelligence
• Five components of data quality must be evaluated for every dataset
• GIScience asks fundamental scientific questions about spatial representation
• Nearly every industry now depends on geospatial technology

• Install QGIS and load a shapefile of your country's administrative boundaries
• Identify 5 GIS applications relevant to your field of study
• Create a simple point layer of 10 locations you visit regularly and add attribute data
• Compare the same area in OpenStreetMap vs Google Maps — note data differences

• Treating GIS as just a map-making tool rather than an analytical framework
• Ignoring coordinate reference systems when combining datasets
• Assuming digital data is automatically accurate
• Not documenting metadata for spatial datasets

• Tomlinson, R.F. (1968). A Geographic Information System for Regional Planning
• Goodchild, M.F. (1992). Geographical Information Science. International Journal of GIS
• Longley, P.A. et al. (2015). Geographic Information Science and Systems, 4th ed.

Vector and raster data models, topology, TIN

The fundamental challenge of GIS is representing the continuous, infinitely complex real world within a discrete, finite computer system. Two primary data models address this challenge: vector and raster, each with distinct strengths and limitations that make them suitable for different types of geographic phenomena.

The vector data model represents geographic features as discrete geometric objects: points (wells, hospitals, GPS waypoints), lines (roads, rivers, utility networks), and polygons (parcels, lakes, administrative boundaries). Each feature has a precise geometric definition stored as coordinate pairs (x,y or x,y,z) and associated attribute data in a relational table. Vector data excels at representing features with crisp boundaries — a property line either exists at a location or it does not.

Topology is the mathematical framework that defines spatial relationships between vector features. In a topological data model, polygons share boundaries (no gaps or overlaps), lines connect at nodes, and features maintain adjacency, containment, and connectivity relationships. The US Census Bureau's TIGER/Line files are a classic example of topological data — every road segment connects at intersections, and every census block is bounded by road segments, creating a seamless fabric of geographic features.

The raster data model represents space as a regular grid of cells (pixels), each containing a single value. Satellite imagery, digital elevation models (DEMs), temperature surfaces, and land cover maps are all raster datasets. Resolution — the ground area covered by each cell — determines both spatial detail and file size. A 30-meter Landsat pixel covers 900 square meters; a 0.5-meter drone image pixel covers 0.25 square meters but requires 3,600 times more storage per unit area.

Triangulated Irregular Networks (TINs) offer a third approach, particularly useful for terrain modeling. TINs connect irregularly spaced elevation points into a mesh of non-overlapping triangles, concentrating detail in areas of complex terrain while using fewer triangles in flat areas. This makes TINs more efficient than regular grids for elevation representation, though they are computationally more complex to work with.

The choice between vector and raster depends on the phenomenon being modeled, the analysis to be performed, and practical considerations like storage and processing requirements. Continuous phenomena (temperature, elevation, rainfall) map naturally to raster. Discrete phenomena (buildings, roads, boundaries) suit vector representation. Many analyses require converting between models — rasterizing vectors for overlay analysis or vectorizing rasters for boundary extraction.

• Vector = discrete features (points, lines, polygons) with precise geometry
• Raster = continuous surfaces as grids of cells with resolution-dependent detail
• Topology defines spatial relationships and ensures data integrity
• TINs are efficient for terrain modeling with variable detail density
• Model choice depends on phenomenon type, analysis needs, and practical constraints

• Compare vector (shapefile) and raster (GeoTIFF) representations of the same area
• Build a simple TIN from survey points and compare to a DEM
• Check topology on a polygon dataset — identify gaps, overlaps, and slivers
• Calculate storage requirements for raster data at different resolutions

• Using raster data for discrete boundary analysis
• Ignoring topology errors that cause incorrect area calculations
• Confusing resolution with accuracy
• Not considering NoData values in raster analysis

Earth shape, coordinate systems, map projections

Geodesy — the science of measuring Earth's shape, orientation, and gravitational field — underpins every GIS operation. The Earth is not a perfect sphere; it is an oblate ellipsoid, slightly flattened at the poles and bulging at the equator. The equatorial radius is approximately 6,378.137 km while the polar radius is 6,356.752 km — a difference of about 21 km. This seemingly small difference has profound implications for accurate positioning.

A geodetic datum defines a reference surface for measuring positions on Earth. A datum consists of an ellipsoid (mathematical approximation of Earth's shape) and its orientation relative to Earth's center of mass. The World Geodetic System 1984 (WGS 84) is the global standard, used by GPS. Regional datums like NAD83 (North America), ETRS89 (Europe), and Everest 1830 (South Asia) are optimized for specific areas. Mixing datums without transformation can introduce positional errors of hundreds of meters — a critical concern when combining datasets from different sources.

Map projections transform the 3D Earth surface onto a 2D plane. Every projection distorts at least one property: area (equal-area projections preserve area but distort shape), shape (conformal projections preserve local angles but distort area), distance (equidistant projections preserve distances from one or two points), or direction (azimuthal projections preserve directions from a central point). No projection can preserve all properties simultaneously — this is a mathematical impossibility proved by Euler in the 18th century.

Common projections include: Mercator (conformal, used for navigation and web maps — Google Maps uses Web Mercator/EPSG:3857); UTM (Universal Transverse Mercator, divides Earth into 60 zones of 6° longitude, each using a Transverse Mercator projection — ideal for large-scale mapping); Albers Equal Area (preserves area, used for thematic mapping of continents); Lambert Conformal Conic (preserves shape, widely used for aeronautical charts and regional mapping).

The choice of projection depends on the mapping purpose, geographic extent, and the property that must be preserved. For thematic maps showing population density or land cover, equal-area projections are essential — using Mercator would dramatically exaggerate areas near the poles. For navigation, conformal projections preserve angles needed for bearing calculations. For distance measurement, equidistant projections centered on the point of interest provide accurate results.

In Bangladesh, the commonly used coordinate system is Bangladesh Transverse Mercator (BTM) based on the Everest 1830 ellipsoid, or UTM Zone 46N (WGS 84). When working with GPS data (WGS 84) and historical survey maps (Everest datum), transformation parameters must be applied. Failure to do so can introduce errors of 200-500 meters — enough to place a road in the middle of a river.

• Earth is an oblate ellipsoid, not a perfect sphere
• Datums define reference surfaces — mixing datums causes positional errors
• All map projections distort reality — choose based on which property to preserve
• UTM divides Earth into 60 zones for large-scale mapping
• Always verify and document the coordinate system of every dataset

• Reproject a dataset from WGS 84 to your local UTM zone and compare coordinates
• Display the same polygon in 3 different projections and measure the area in each
• Transform points between two datums and measure the positional shift
• Create a world map in Mercator, Albers, and Robinson — compare Greenland's apparent size

• Overlaying layers in different CRS without reprojection
• Using Web Mercator (EPSG:3857) for area measurements
• Confusing geographic coordinates (degrees) with projected coordinates (meters)
• Ignoring datum transformations between old survey data and GPS data

Digitizing, scanning, geocoding, GPS import

Data entry is the process of creating digital spatial data from analog sources or field observations. The quality of GIS analysis is fundamentally constrained by the quality of input data — the "garbage in, garbage out" principle applies with particular force in spatial analysis where errors propagate through every subsequent operation.

Digitizing — tracing features from scanned maps or imagery on screen — remains one of the most common data entry methods. Heads-up digitizing involves tracing features displayed on a computer screen, while tablet digitizing uses a physical digitizing tablet with a puck or stylus. Modern digitizing leverages high-resolution satellite imagery as a base layer, with tools like snapping, streaming, and auto-complete polygons improving both speed and accuracy.

Geocoding transforms text addresses into geographic coordinates, enabling spatial analysis of business data, customer locations, and incident reports. Reverse geocoding does the opposite — converting coordinates to human-readable addresses. The accuracy of geocoding depends on the quality of the reference address database. In developing countries like Bangladesh, where formal addressing systems are incomplete, geocoding accuracy is limited, and alternative approaches like what3words or Plus Codes may be necessary.

GPS data import brings field-collected coordinates directly into GIS. Modern GNSS receivers achieve sub-meter accuracy with differential corrections and centimeter accuracy with RTK methods. The workflow typically involves: mission planning (checking satellite geometry via PDOP), field data collection with attribute recording, post-processing differential corrections, and importing into GIS with quality checks. Common formats include GPX, KML, and CSV with coordinate columns.

• Data quality determines analysis quality — invest in accurate data entry
• Digitizing from imagery requires understanding of scale and resolution limitations
• Geocoding accuracy varies dramatically by region and address quality
• GPS data requires post-processing and quality assessment before GIS import

• Digitize 20 buildings from satellite imagery and assess your accuracy against a reference dataset
• Geocode 50 addresses and map the results — identify failures and their causes
• Collect GPS points in the field and import them into QGIS

Global positioning systems, LiDAR, remote sensing data

The Global Positioning System (GPS) — more broadly, Global Navigation Satellite Systems (GNSS) — has revolutionized spatial data collection. GPS uses a constellation of 31+ satellites orbiting at approximately 20,200 km altitude, transmitting precisely timed signals. A receiver calculates its position by measuring the time delay from at least 4 satellites (3 for position + 1 for clock correction), a process called trilateration.

LiDAR (Light Detection and Ranging) has emerged as a transformative data source for high-resolution 3D mapping. Airborne LiDAR systems emit 100,000+ laser pulses per second, measuring the round-trip time of each pulse to calculate distance with centimeter-level accuracy. The result is a dense 3D point cloud that captures terrain, vegetation, buildings, and infrastructure. In forestry, LiDAR can estimate canopy height, biomass, and forest structure with unprecedented accuracy.

Remote sensing provides synoptic, repeatable coverage of Earth's surface. Key data sources include: Landsat (30m resolution, since 1972 — the longest continuous satellite record), Sentinel-2 (10m resolution, 5-day revisit), MODIS (250m-1km, daily global coverage), and commercial satellites like WorldView (0.3m resolution). Each source serves different analysis scales — MODIS for global monitoring, Landsat for regional change detection, Sentinel for agricultural mapping, and commercial imagery for urban feature extraction.

Open data initiatives have democratized access to geospatial data. OpenStreetMap provides crowd-sourced vector data globally. USGS EarthExplorer offers free Landsat and other satellite imagery. Copernicus Open Access Hub distributes Sentinel data. NASA's Earthdata Search provides access to thousands of Earth science datasets. For Bangladesh, LGED provides local infrastructure data, and SPARRSO offers national remote sensing resources.

• GNSS provides the positional foundation for nearly all modern spatial data
• LiDAR captures high-resolution 3D surfaces for terrain and vegetation analysis
• Remote sensing satellites provide data at scales from global to sub-meter
• Open data initiatives have made vast geospatial datasets freely available

Database operations, joins, queries, SQL basics

Every GIS feature has two inseparable components: geometry (where it is) and attributes (what it is). The attribute table — a structured relational database linked to spatial features — is where most GIS analysis begins. Understanding database operations is as essential to GIS as understanding spatial analysis.

Relational database management systems (RDBMS) store attribute data in tables with rows (records) and columns (fields). Each row corresponds to one geographic feature, and each column stores a specific attribute. Data types include integers, floating-point numbers, text strings, dates, and binary objects. Choosing appropriate data types affects storage efficiency, query performance, and analytical capabilities.

SQL (Structured Query Language) enables powerful data retrieval and manipulation. SELECT queries extract specific records; WHERE clauses filter by conditions; JOIN operations combine data from multiple tables; GROUP BY aggregates data by categories; and ORDER BY sorts results. In GIS, SQL queries often incorporate spatial predicates like ST_Intersects, ST_Within, and ST_Buffer for spatial filtering.

Table joins link attribute data from external sources to spatial features. A common key field (like a district code or census tract ID) connects the two tables. Join types include inner joins (only matching records), left joins (all spatial features, matched attributes where available), and spatial joins (linking records based on geographic proximity or containment). Proper join design prevents data loss and duplication.

• GIS data = geometry + attributes — both are essential
• SQL is the universal language for querying and manipulating attribute data
• Table joins connect external data to spatial features via common key fields
• Data types affect storage, performance, and analytical capabilities

Buffering, overlay, proximity, network analysis

Spatial analysis is the heart of GIS — the ability to pose geographic questions and derive answers that would be impossible with maps or databases alone. Basic spatial analysis operations form the toolkit that every GIS analyst uses daily, from simple proximity queries to complex multi-criteria decision models.

Buffering creates zones of specified distance around features. Point buffers produce circles, line buffers produce corridors, and polygon buffers expand or contract boundaries. Applications include establishing setback zones around wetlands, creating service areas around hospitals, and defining noise impact zones around airports. Variable-width buffers can reflect different protection standards — for example, wider riparian buffers along steeper slopes.

Overlay analysis combines two or more layers to derive new information. The fundamental overlay operations are: Union (combines all features from both layers), Intersect (retains only overlapping areas), Difference (subtracts one layer from another), and Identity (splits features of one layer by boundaries of another while preserving all input features). Overlay is the basis for suitability analysis — combining soil quality, slope, water access, and zoning layers to identify optimal locations for development.

Network analysis models flow through connected linear systems like road networks, utility grids, and stream channels. Common operations include shortest path routing (finding optimal routes considering distance, time, or cost), service area analysis (determining the area reachable within a specified travel time or distance), and location-allocation modeling (optimizing facility placement to serve demand points). Emergency response planning uses network analysis to ensure all areas can be reached within a critical time threshold.

Proximity analysis measures spatial relationships between features. Near analysis calculates the straight-line distance from each feature to the closest feature in another layer. Point-in-polygon analysis determines which polygon each point falls within. Thiessen (Voronoi) polygons partition space so that every location is assigned to its nearest input point — useful for defining service areas when no road network is available.

• Buffering, overlay, and proximity form the core spatial analysis toolkit
• Overlay analysis combines layers to answer multi-criteria spatial questions
• Network analysis models flow and connectivity through linear systems
• Proximity analysis quantifies spatial relationships between features

• Buffer all schools by 500m and identify residential areas within the buffer
• Perform an overlay of land use and soil type layers to identify suitable farmland
• Calculate the shortest route between two points on a road network
• Generate Thiessen polygons for weather stations and interpolate rainfall

Map algebra, surface analysis, cost paths

Raster analysis operates on the grid-cell data model, treating geographic space as a continuous surface of values. The mathematical framework underlying raster analysis is map algebra, introduced by Dana Tomlin in the 1980s. Map algebra defines operations on rasters at four levels: local (cell-by-cell), focal (neighborhood-based), zonal (group-based), and global (entire surface).

Local operations combine corresponding cells from two or more rasters: adding rainfall layers, computing vegetation indices from spectral bands (NDVI = (NIR - Red) / (NIR + Red)), or reclassifying land cover values. Focal operations compute values based on neighboring cells — smoothing filters average surrounding values, slope calculations examine elevation differences between adjacent cells, and edge detection highlights boundaries.

Surface analysis derives new information from elevation data (DEMs). Slope maps show the steepness of terrain in degrees or percent, essential for erosion modeling and construction planning. Aspect maps show the compass direction a slope faces, important for solar energy potential and microclimate analysis. Hillshade creates an illuminated surface for cartographic visualization. Viewshed analysis determines which areas are visible from one or more observation points — critical for telecommunications tower placement, scenic quality assessment, and military planning.

Cost path analysis finds the least-cost route between two points across a surface where different areas have different traversal costs. The cost surface might combine slope (steeper = higher cost), land cover (forest = higher cost than road), and water bodies (impassable). Applications include wildlife corridor design, pipeline routing, and search-and-rescue path planning. The accumulated cost surface can also generate least-cost corridors — zones of near-optimal routes that provide alternatives.

• Map algebra operates at local, focal, zonal, and global levels
• Surface analysis extracts slope, aspect, hillshade, and viewshed from DEMs
• Cost path analysis finds optimal routes across heterogeneous landscapes
• Raster analysis treats geography as continuous surfaces rather than discrete features

DEMs, slope, aspect, watershed delineation

Terrain analysis uses Digital Elevation Models (DEMs) to extract geomorphological information about Earth's surface. DEMs can be derived from multiple sources: contour line interpolation, photogrammetric processing of stereo imagery, LiDAR point clouds, radar interferometry (SRTM, TanDEM-X), and GPS surveys. Resolution ranges from 90m (SRTM) to sub-meter (LiDAR-derived), with resolution determining the scale of features that can be detected.

Hydrological analysis is one of the most powerful applications of terrain analysis. The workflow begins with filling sinks (artificial depressions that would trap water), then computing flow direction (the steepest downhill path from each cell), flow accumulation (counting upstream cells draining through each cell), and finally delineating watersheds (the drainage area above a specified pour point). This automated process replaced weeks of manual delineation from contour maps.

Geomorphometric parameters derived from DEMs include: curvature (profile curvature affects flow acceleration; plan curvature affects flow convergence/divergence), terrain roughness (variability of elevation within a neighborhood), topographic position index (relative elevation compared to surroundings — ridges, valleys, slopes), and topographic wetness index (potential for water accumulation based on upslope area and local slope).

In Bangladesh, terrain analysis has specific applications: flood inundation modeling using SRTM and ALOS PALSAR DEMs, identifying elevated "shelter sites" for flood refugees, mapping drainage networks in the Ganges-Brahmaputra delta, and assessing land suitability for aquaculture based on elevation and water table proximity.

• DEMs are the foundation of terrain analysis — resolution determines analytical capability
• Hydrological analysis automates watershed delineation from elevation data
• Geomorphometric parameters describe terrain shape for hazard and resource assessment
• Terrain analysis is critical for flood management in low-lying countries like Bangladesh

Cartographic principles, symbology, layouts

Cartographic design transforms spatial data into visual communication. A well-designed map tells a story that is immediately understood; a poorly designed map misleads or confuses. The principles of cartographic design — visual hierarchy, figure-ground contrast, balance, and legibility — are as important as the underlying data.

Visual hierarchy ensures readers see the most important information first. Primary features (the map's subject) should be visually dominant through color saturation, line weight, or symbol size. Secondary features (context like roads and boundaries) should be present but subordinate. Background elements (basemaps, labels) should be subtle. The five visual variables — size, shape, color hue, color value (lightness), and color saturation — are the cartographer's tools for creating hierarchy.

Color selection in cartography follows specific principles: sequential color schemes (light-to-dark) represent ordered data (population density); diverging schemes (two hues with neutral midpoint) show deviation from a mean (temperature anomaly); qualitative schemes (distinct hues) distinguish categories (land cover types). ColorBrewer, developed by Cynthia Brewer, provides scientifically validated color schemes that are perceptually uniform and accessible to colorblind viewers.

Map layout composition integrates the map face with essential marginalia: title (concise, descriptive), legend (symbol explanation), scale bar (graphic scale that remains accurate when printed at different sizes), north arrow (orientation reference), data source citation, projection information, and map author/date. Professional layouts often include inset maps (showing location context), charts (summarizing spatial patterns), and photographs (adding human interest).

Digital map output requires understanding of resolution, color modes, and file formats. Print maps typically need 300 DPI resolution in CMYK color mode, exported as PDF or high-resolution TIFF. Web maps may use lower resolution (72-150 DPI) in RGB color mode, optimized as PNG (lossless, best for maps) or JPEG (lossy, smaller files). Vector formats like SVG preserve quality at any zoom level.

• Visual hierarchy guides the reader's eye to the most important information
• Color schemes must match data type: sequential, diverging, or qualitative
• Professional layouts include title, legend, scale bar, north arrow, and source citation
• Output format depends on medium: print (300 DPI, CMYK, PDF) vs web (72 DPI, RGB, PNG)