PEC EPE Geoinformatics common mistakes
PEC EPE Geoinformatics · Understand the traps and how to avoid them.
When computing a geodetic problem, the candidate mixes ellipsoidal quantities (geodetic latitude φ, ellipsoidal height h) with geocentric/spherical ones (geocentric latitude ψ, radius vector r) and plugs them into the same formula, e.g. applying a spherical Vincenty or great-circle distance to coordinates given as φ, λ on a specific ellipsoid without reducing to the auxiliary sphere correctly.
Many textbooks and online tools present spherical formulas, and the candidate memorizes one formula without tracking which latitude definition and which ellipsoid it assumes. The numerical difference (up to ~0.19° between φ and ψ, and tens of km in distance over long lines) is invisible in a quick check, so the error goes unnoticed.
What to do instead
Before any computation, write down the reference surface (sphere vs. GRS80/WGS84 ellipsoid) and the latitude type explicitly. Convert geodetic to geocentric latitude with tan ψ = (1−e²) tan φ when a spherical formula is used, and use ellipsoidal formulas (Vincenty/Karney) for ellipsoidal inputs. Sanity-check a long line against a known baseline (e.g. equator degree ≈ 110.6 km vs. polar degree ≈ 111.7 km) to catch surface mismatches.
The candidate ignores the epoch and reference frame of coordinates, treating a coordinate triple as timeless - e.g. comparing or transforming coordinates given in ITRF2014 epoch 2010.0 with others in ETRS89 or in ITRF2000 epoch 2000.0 as if they were directly comparable, or omitting the plate-motion/velocity term in a datum transformation.
Coordinates are usually presented as bare numbers, and the candidate focuses on the projection or the transformation parameters while overlooking the frame realization and epoch. Time-dependent frames are a small part of most courses, so the habit of checking epoch is never built.
What to do instead
Adopt a fixed checklist for every coordinate set: frame name (ITRF/ETRS89/national), realization/year, epoch, and whether velocities must be applied. Practice one full ITRF-to-ETRS89 transformation including the epoch propagation (v·Δt) and state the residual. In exam answers, always write the frame and epoch next to the coordinates.
In engineering surveying setting-out, the candidate computes coordinates and then sets out directly from them without closing the traverse or checking the control, and reports the computed setting-out elements as final without an independent check (e.g. no redundant measurement, no check to a second control point).
Coursework rewards arriving at the numerically correct coordinate, so the candidate treats computation as the end of the task. The professional habit of redundancy and independent verification is procedural and rarely tested by calculation drills, so it is not internalized.
What to do instead
For every setting-out scenario, structure the answer as: (1) control check (known baseline or resection with redundancy), (2) computation of setting-out elements, (3) an independent check measurement (second known point, check distance, or repeated observation), (4) stated tolerances. Memorize typical engineering tolerances and quote them; graders award marks for the verification step, not only the number.
The candidate treats a GNSS position as a single point with one accuracy figure, ignoring the distinction between code (pseudorange) and carrier-phase solutions, between absolute and relative (differential) positioning, and between precision, accuracy and integrity - e.g. quoting a 2 cm RTK figure for a standalone code solution or claiming RTK works without a data link or fixed ambiguity.
Manufacturer specs and marketing language blur these categories, and the candidate memorizes headline numbers ("RTK = cm") without the conditions that produce them (fixed integers, short baseline, good geometry, correction link). Exam questions that vary one condition then defeat the memorized number.
What to do instead
Build a table of GNSS modes (SPP, SBAS, DGNSS, RTK float/fixed, PPP, static relative) with inputs, typical accuracy, latency, and failure conditions. For each answer, name the mode and the conditions. Check geometry via DOP and state that RTK requires a fixed ambiguity and a correction link; state that PPP needs long convergence and precise orbits/clocks.
When interpreting photogrammetric or remotely sensed products, the candidate reports a map/point accuracy equal to the nominal GSD or pixel size, and treats classification accuracy as if it were positional accuracy - e.g. stating "10 cm GSD therefore 10 cm accuracy" or quoting overall accuracy from a confusion matrix as the map's positional quality.
GSD is the most visible number in a product description and is easily conflated with accuracy. The candidate does not separate radiometric/thematic accuracy (confusion matrix, kappa) from geometric accuracy (RMSE against check points), and does not account for factors like image matching, terrain, and control that dominate real accuracy.
What to do instead
For any product, state separately: GSD (sampling), geometric accuracy (RMSE_x, RMSE_y, RMSE_z against independent check points), and thematic accuracy (overall accuracy, user's/producer's accuracy, kappa) with the validation sample design. Learn that planimetric accuracy is typically 1–3× GSD depending on matching and control, and never derive one metric from another.
The candidate applies a map projection formula without checking whether the point lies inside the projection's valid domain or zone, and without distinguishing grid convergence, magnetic declination and true bearing - e.g. using a UTM scale factor of 0.9996 everywhere, ignoring the zone width limit, or treating grid north as true north in an azimuth computation.
Projection formulas are learned as algebra, and the validity conditions (zone limits, hemisphere, scale factor variation with easting, convergence γ) are learned separately or not at all. In a time-pressured exam the candidate substitutes numbers into the formula and never asks whether the result is meaningful.
What to do instead
For each projection, memorize its domain and distortion pattern: UTM zones 6° wide, k₀ = 0.9996 on the central meridian, scale error growing with easting (≈0.9996 + x²/2R²), convergence γ ≈ Δλ·sin φ. Always compute and state grid convergence and, where relevant, magnetic declination with its epoch and annual change, and show the bearing relationship (true = grid + γ; magnetic = true − declination).
In GIS analysis, the candidate runs distance, area, buffer or slope operations on data in a geographic (angular) CRS, or reprojects raster data with nearest-neighbour when a continuous variable requires resampling, and reports the resulting numbers as valid.
GIS software accepts any layer in any CRS and silently produces numbers, so nothing flags the error. The candidate has learned the operations and the CRS concept as separate topics and does not connect them at the moment of analysis.
What to do instead
Adopt a rule: any metric operation requires a projected CRS appropriate to the extent (state why, e.g. UTM zone, equal-area projection for area statistics). For rasters, match resampling to data type (nearest for categorical, bilinear/cubic for continuous). In exam answers, state the CRS choice and its justification, and note the distortion if an equal-area or conformal projection is used for the wrong purpose.
The candidate designs or critiques a spatial database without addressing spatial indexing, topology, and coordinate storage explicitly - e.g. storing geometries as text without SRID, omitting a spatial index, or ignoring that a query with mixed SRIDs returns wrong or no results. Also common: confusing the roles of WKT/WKB, GeoJSON and GML, or assuming a spatial database enforces topology automatically.
Database questions are often answered from general SQL knowledge, and spatial extensions (PostGIS, Oracle Spatial, OGC Simple Features) are treated as an add-on rather than a design constraint. The candidate has read about spatial indexes but never had to justify one.
What to do instead
For any schema question, explicitly state: geometry column type with SRID, spatial index (GiST/R-tree) and why, coordinate precision, and whether topology is enforced at the database level or by application rules. Learn the OGC Simple Features model and the format roles (WKT/WKB for storage and interchange, GeoJSON for web, GML for OGC services, GeoPackage for portable containers). Include a query plan reasoning: bounding-box filter first, then exact predicate.
In land administration and cadastral questions, the candidate answers with technical geometry only and omits the legal/administrative layer - e.g. describing boundary re-establishment purely by coordinates without addressing the legal boundary, monumentation hierarchy, adverse possession, or the difference between fixed and general boundaries, and without naming the applicable statutory framework.
The candidate comes from a technical surveying background and treats cadastral questions as coordinate problems. Legal concepts (title vs. boundary, registration principles, Torrens vs. deeds registration) are studied as isolated definitions rather than as the context in which survey measurements have legal effect.
What to do instead
Structure cadastral answers in two layers: (1) legal - what the boundary is in law, what the register records, which statute and authority govern, what evidence re-establishes an obliterated boundary; (2) technical - measurement method, accuracy, and how it supports the legal boundary. Practice distinguishing fixed vs. general boundary systems and the hierarchy of evidence (original monuments, then measurements, then occupation).
In professional-practice, ethics and project-management items, the candidate answers with generic good intentions ("act honestly", "communicate with the client", "follow the contract") instead of identifying the specific obligation, the conflicting duties, the correct escalation path, and the documented action - e.g. failing to state that a discovered error must be reported to the client and the regulating body, or that a scope change requires written variation before work continues.
Ethics and management are perceived as common sense and therefore under-prepared. The candidate has not internalized codes of conduct, procurement types, or the contractual mechanics (variation, delay, limitation of liability), so answers stay at the level of values rather than professional procedure.
What to do instead
Prepare a decision framework for ethics scenarios: identify the facts, the duties owed (public, client, profession, employer), the conflict, the disclosure and escalation route, and the record to be kept. For project management, memorize the contract types (lump sum, schedule of rates, cost-plus, design-build), the variation and delay-notification mechanics, and the components of a QA/QC plan. Answer in numbered actions, not adjectives.