Development of a method for mapping the highest coastline in Sweden using breaklines extracted from high resolution digital elevation models

Abstract (Swedish)

Utveckling av en metod för att kartera högsta kustlinjen i Sverige med hjälp av brytlinjer extraherade från högupplösta höjdmodeller

Högsta kustlinjens (HK) geografiska läge i landskapet markerar gränsen mellan subakvatiskt och supraakvatiskt avsatta sediment. Idag varierar HKs höjd över havet från några få m.ö.h. i Skåne till omkring 289 m.ö.h. i Ångermanland på grund av den glacialisostatiska landhöjningen. Högkvalitativ HK-data är av intresse bland annat vid markanvändnings- och samhällsplanering och vid rekonstruktion av historiska havsnivåer och händelser i Östersjöns havsbassäng.

Den här studien har undersökt möjligheten att använda brytlinjer i terrängen, extraherade från så kallade ”land surface parameters” (LSPs) som genererats ur högupplöst höjddata, för att automatisera karteringen av HK i svallade områden i Sverige. För att uppskatta lämpliga skaldimensioner som framhäver terrängbrytlinjer av intresse för studien testades ett intervall av fönsterstorlekar för terrängparameterberäkningarna. Fyra stycken semi-automatiserade metoder för att extrahera brytlinjer baserade på markytans krökning utvecklades i ESRIs ArcGIS 10.2.2 for Desktop och tillämpades på två pilotområden i Sverige. Metoderna är uppbyggda av ett gemensamt steg där brytlinjerna extraheras och ett individuellt steg där brytlinjerna klassificeras med olika grad av automation. För att jämföra HK-kartorna genererade från de utvecklade metoderna med HK-kartor som manuellt karterats från högupplösta terrängskuggningskartor och med HK-data som idag finns tillgänglig hos Sveriges geologiska undersökning (SGU) beräknades klassificeringsnoggrannhet och höjdfel för samtliga kartor i förhållande till referensdata.

Studien fann att en 22x22m fönsterstorlek var framgångsrik för att extrahera brytlinjer av markytans krökning relaterad till strandvallar och glaciala lineationer som används för att kartera HK i pilotområdena. Noggrannhetsutvärderingen indikerar att tre av de utvecklade metoderna genererar HK-kartor med högre noggrannhet än SGUs HK-data, och liknande till högre noggrannhet än HK-kartor baserade på HK-punkter manuellt karterade från högupplösta terrängskuggningskartor. Metoderna som använder en manuell klassificering av brytlinjerna visade en högre noggrannhet än metoderna som använder en automatiserad klassificering av brytlinjerna.

Denna studie visar att brytlinjer av markytans krökning genererade från högupplösta höjdmodeller kan användas för att göra kartering av HK i svallade områden mer automatiserad, strukturerad och reproducerbar samtidigt som man uppnår liknande noggrannhet som vid manuell kartering baserad på terrängskuggningskartor.

Abstract

Development of a method for mapping the highest coastline in Sweden using breaklines extracted from high resolution digital elevation models

The geospatial position of the highest coastline (HCL) defines the boundary between subaquatic and supra-aquatic deposited sediments. Today the HCL is located at different elevations throughout Sweden, a few m.a.s.l. in southern Scania to around 289 m.a.s.l. at the coast of Ångermanland, due to the glacio-isostatic rebound. High quality data of the HCL is of interest e.g. in land-use and spatial planning and when reconstructing historical sea levels and events within the Baltic Sea Basin.

In this study the use of land surface parameter (LSP) breakline extraction methods applied on high resolution DEMs for automating mapping of the HCL in wave washed areas in Sweden was investigated. Appropriate scale dimensions for enhancing breaklines of interest was estimated by testing a range of moving window sizes for the LSP computations. Four semi-automated mapping methods based on curvature breakline extraction was developed in ESRI’s ArcGIS 10.2.2 for Desktop and applied on two pilot areas in Sweden. The methods consist of a common breakline extraction step and individual breakline classification steps with differing grade of automation. To compare the HCL maps generated by the developed methods with HCL maps manually mapped from high resolution hillshade maps and with the current HCL data supplied by the Geological Survey of Sweden (SGU) classification accuracies and elevation errors were computed using a reference data set.

A 22x22m moving window size was found successful for the extraction of curvature breaklines related to wave washed features and glacial flow lineation features used to map the HCL within the pilot areas under investigation. The accuracy assessment indicates that three of the developed methods generate HCL maps with accuracies above the current HCL data provided by SGU and accuracies similar to or above HCL maps based on manually mapped HCL data points. Higher accuracies were found for the methods using a manual classification of the extracted breaklines than for the methods using an automated classification of the extracted breaklines.

This study found that, by applying curvature breakline extraction methods on high resolution DEMs, HCL mapping in wave washed areas can be made more automated, structured and reproducible while still reaching similar accuracies as manual hillshade mapping methods.

Popular Abstract

Mapping the highest coastline using high resolution elevation data

The highest position the coastline had after the last ice age is called the highest coastline (HCL). Today the HCL is located at different elevations throughout Sweden, a few m.a.s.l. in southern Scania to around 289 m.a.s.l. at the coast of Ångermanland, due to the glacio-isostatic rebound. The glacio-isostatic rebound occurs at different rates in Sweden, depending on to what extent the ground at a location was depressed by the weight of the ice-sheet. Today the HCL can be mapped by determining the elevation of relict landforms connected to the HCL e.g. the highest beach ridge formed by wave washing action of a historical sea. High quality data of the HCL is of interest e.g. in land-use and spatial planning and when reconstructing historical sea levels and events within the Baltic Sea Basin.

Using a high resolution digital elevation model (DEM), landforms connected to the HCL can be identified with high precision. Manual digital mapping using a detailed DEM can be time-demanding and therefore partly automated mapping methods using Geographical Information Systems (GIS) are of interest when working with large areas. This study investigates the possibilities to automate the mapping of the HCL by extracting information of landforms connected to the HCL from Sweden’s national high resolution DEM. The main aim of this study was to develop a structured and reproducible method for mapping the HCL while reaching similar accuracies as manual digital mapping methods and higher accuracies than the current HCL data available at the Geological Survey of Sweden (SGU).

Four mapping methods with differing automation were developed using GIS. The methods consists of two main parts: part 1 where maximum and minimum curvature information of the topography is extracted from the DEM, generating curvature breaklines which describe concave and convex surfaces of landforms, and part 2 where the breaklines are classified into landforms connected to the HCL. HCL maps produced by the developed methods were compared with manually mapped HCL maps and the HCL data supplied by the SGU. Three of the developed methods were found to generate HCL maps with the desired accuracies aimed for. Higher accuracies were found for the developed methods using a manual classification of the extracted breaklines in part 2 than for the methods using an automated classification of the breaklines.

This study shows that HCL mapping, in areas where wave washed landforms are dominant, can be made more automated while still reaching similar accuracies as manual digital mapping methods.

• author: Lundgren, Anna
• supervisor: Harry Lankreijer
• organization: Dept of Physical Geography and Ecosystem Science
• course: GISM01 20161
• year: 2016
• type: H2 - Master's Degree (Two Years)
• subject: Earth and Environmental Sciences
• keywords: Physical Geography and Ecosystem analysis, Geographical Information Systems, GIS, the highest coastline, Geological Survey of Sweden, SGU, high resolution digital elevation model, LiDAR, land surface parameter, curvature, breakline extraction, geomorphological mapping
• publication/series: Master Thesis in Geographical Information Science
• report number: 47
• language: English
• additional info: External supervisor: Tore Påsse, Geological Survey of Sweden; Download the thesis in a low resolution version: "PDF_Anna_Lundgren_GISM01_MasterThesis_low_reso.pdf", or in a high resolution version: "PDF_Anna_Lundgren_GISM01_MasterThesis_higher_r.pdf".
• id: 8871937
• date added to LUP: 2016-05-03 17:13:45
• date last changed: 2016-05-03 17:13:45