UAV Topographic Map With 0.25 m Contours
You are commissioned to produce a 1:1,000 topographic map with 0.25 m contours for a 60 ha housing site in Trelawny. A UAV photogrammetric survey was flown at 120 m above ground level with 12 ground control points measured by RTK, but the GCPs are clustered along the access road. The orthomosaic and dense cloud were processed in ITRF2014 geographic coordinates and then transformed to the Jamaica National Grid. Independent total-station checkpoints at 15 well-distributed locations show horizontal residuals of 0.10 to 0.35 m and vertical residuals of -0.55 to +0.20 m; the largest vertical errors occur under secondary vegetation. The road centreline in the orthophoto plots 1.8 m east of the design centreline. The client wants the map for earthworks quantities and drainage design and asks you to 'smooth the contours' so they match the road design.
1. Evaluate whether the current deliverable meets the stated accuracy, and explain the likely causes of the residuals and the road-centreline shift.
Read the worked answer
The deliverable does not meet a 0.25 m contour specification on the evidence given. Independent checkpoints show horizontal residuals up to 0.35 m and vertical residuals from -0.55 to +0.20 m; the vertical spread is more than twice the contour interval, so the contours would be misleading for drainage and earthworks quantities. The 1.8 m road-centreline shift is a separate red flag. A systematic horizontal shift usually points to georeferencing or transformation problems, GCP coordinate error, or a datum and projection issue rather than random photogrammetric noise. The GCPs are clustered along the access road, so the model has poor control across the 60 ha site and can rotate or bow away from the road. Vegetation reduces ground visibility and biases the dense cloud upward, which fits the largest vertical errors under secondary vegetation. The ITRF2014-to-Jamaica-Grid transformation, including epoch and scale, must also be checked.
2. What processing and field checks would you require before accepting a revised surface?
Read the worked answer
Reject the current surface for design until it is re-processed. Add well-distributed GCPs and independent checkpoints, especially at the far corners and in vegetated areas. Use RTK or total station for ground checkpoints and cross-sections, verify the coordinate transformation and projection, check camera calibration and flight overlap, classify ground points, and remove vegetation returns. Compare the DEM with total-station cross-sections along proposed drains and roads. If vegetation prevents reliable ground modelling, supplement the UAV data with RTK or total-station spot levels or with LiDAR. Report both absolute and relative accuracy, because the design team needs to know whether the surface is fit for earthworks quantities and hydraulic design. A check that only compares the model with its own GCPs is not independent and should not be accepted.
3. How should you respond to the client's request to smooth the contours so they match the road design?
Read the worked answer
I should not smooth the contours to match the road design. That would alter the survey evidence and misrepresent the terrain, especially where the errors affect drainage and quantities. It is a professional integrity issue, not a cosmetic one. I can produce a clearly labelled design or visualisation surface, or a generalised map for presentation, but the topographic deliverable must show the measured surface with its accuracy statement and limitations. I would explain to the client that smoothing hides the very discrepancies that matter, and offer a re-survey or targeted ground check instead. If the client insists on presenting the smoothed surface as the survey, I would decline to certify it.