Lecture4_Class(1)
Lecture Overview
Course: Digital Terrain Modelling (LSGI3242A)
Institution: Department of Land Surveying and Geo-Informatics, The Hong Kong Polytechnic University
Instructor: Tiangang Yin (tiangang.yin@polyu.edu.hk)
Lab Sessions: Every Monday, 3:30 – 4:20 pm
Lab Sessions Schedule
Lab 1: Weeks 1-2 (5%) - Search of DTM from the Internet | Tutors: Qi & Weiwei
Lab 2: Weeks 3-5 (5%) - Terrain Analysis | Tutors: Cailin & Weiwei
Lab 3: Week 6 (10%) - Comparison of Interpolation Methods | Tutors: Qi & Weiwei
Lab 4: Weeks 7-8 (10%) - Comparison between TIN and constrained TIN / TIN with different data sets | Tutors: Cailin & Weiwei
Lab 5: Weeks 9-10 (10%) - Contouring from DTM and Flow Analysis | Tutors: Qi & Cailin
Lab 6: Weeks 11-13 (10%) - Effect of DTM accuracy and random noise | Tutors: Cailin & Qi
Mid-Term Quiz
Date: 28 October 2024
Time: 10:00 am to 10:45 am
Location: PQ303
Format: Closed book; 20-30 True/False or Multiple-Choice questions
Weight: 10% of final marks
Content: Covers Lectures 1 to 5
Introduction
DTM Objectives:
Identification of objectives, resources, and budget
Understanding terrain through historic records and site reconnaissance
Data acquisition including field surveys, aerial photos, remote sensing, and digitization of historic maps
Terrain sampling and surface modelling
Digital Terrain Surface Modelling
Definition: A mathematical representation of the terrain surface after data collection
Process: Surface reconstruction or modelling is achieved through interpolating known data points to estimate unknown values
Concept of Interpolation:
Essential for estimating elevation values based on discontinuous measured data
Continuous surface reconstruction demands interpolation methods to estimate unknown height values
Mathematical Functions Used in Modelling
Spatial Domain Representation:
Basic Form: Z = f(x, y) (elevation based on coordinates)
General Polynomial Functions include terms based on the terrain complexity
Characteristic Plots for Polynomial:
1st plot: Horizontal plane
2nd & 3rd plots: Inclined planes
Higher terms result in curves or complex surfaces
Surface Modelling Cases
Case 1: Point-Based Surface Modelling
Single sampled point; discontinuous surface, works for any data pattern
Case 2: Triangle-Based Surface Modelling
Utilizes three points; creates a continuous but not necessarily smooth surface, applicable in various sampling contexts
Case 3: Grid-Based Bilinear Surface Modelling
Four unknowns based on grid; smooth surface but may not be continuous depending on truncation
Case 4: Advanced Grid-Based Surface
Utilizes higher-order terms to achieve smooth global and regional surfaces
Surface Classification
Based on size:
Local: Small, complex areas
Global: Simplified surfaces covering large areas
Regional: Between local and global
Based on smoothness:
Discontinuous: Step-like surfaces
Continuous: Linked local surfaces
Smooth: Utilizes polynomials ensuring continuous first and higher-order derivatives
Interpolation Properties
Why Interpolation?
Measurements can't cover every terrain location; must estimate heights from known adjacent points using techniques:
Exact Fitting: Passes through all points
Best Fitting: Approximates data using least squares, allows redundancy
Classification of Interpolation Methods
Area vs Point-based Interpolation:
Area-based: Based on grids
Point-based: Utilizes nearby data points to estimate unknowns
Exact vs Best Fitting:
Overfitting can cause instability; lower-order fitting is often preferred for a more stable interpolation
Summary of Key Concepts
Higher-order polynomial fitting can lead to overfitting, causing issues such as oscillation due to noise in data. Best fitting prefers a lower order with least squared errors.
Various interpolation techniques (nearest neighbor, bilinear, bicubic) yield different surface characteristics depending on the requirement of smoothness or continuity.
The ongoing evolution of digital terrain modelling continues to influence practices in land surveying and geo-informatics.