Automating land cover classification using time series NDVI : a case study in the Berg River Catchment Area
| dc.contributor.advisor | Munch, Zahn | en_ZA |
| dc.contributor.author | Adesuyi, Ayodeji Steve | en_ZA |
| dc.contributor.other | Stellenbosch University. Faculty of Arts and Social Sciences. Dept of Geography and Environmental Studies. | en_ZA |
| dc.date.accessioned | 2016-12-22T13:05:40Z | |
| dc.date.available | 2016-12-22T13:05:40Z | |
| dc.date.issued | 2016-12 | |
| dc.description | Thesis (MA)--Stellenbosch University, 2016. | en_ZA |
| dc.description.abstract | ENGLISH ABSTRACT: The processing of large volumes of geographic information system (GIS) and remote sensing (RS) data necessitates the development of automated techniques which are cost-effective, faster and user-friendly in order to aid spatial decision making. In this study, an automated technique for identifying agricultural land cover was developed using a custom tool. Multiple ensemble classifiers in ArcGIS workflow automation tool (MEAWAT) was tested on time-series MODIS normalised difference vegetation (NDVI) data using the Berg River catchment area of Western Cape, South Africa as a case study. Although the tool was developed to perform agricultural land cover classification using MODIS input data, the tool was subsequently applied to Landsat NDVI data of the same study extent. A few modifications to the tool were implemented to accommodate the different satellite imagery. The tool was built on an ArcGIS/Python platform, and various GIS & RS functions usually performed in a variety of different software packages were integrated, including study area selection, reprojection, classification and accuracy assessment. The NDVI phenology curve was used to create training data for the classification. Different parameters were tested which allow users to engage with different rules and derive a suitable land cover map for their purpose. MEAWAT uses decision tree and ensemble classifiers such as random forest and extra-tree as well as boosting using a meta-estimator (AdaBoost). Classification accuracies of 70.5%, 75.5%, 76.3% and 78.7% were achieved respectively with MODIS data, while an accuracy of 89% was achieved using the boosted random forest classifier on the Landsat data. It was observed that a better classification output can be derived using MEAWAT on higher resolution satellite imagery provided good training data are available. These findings highlight the potential of MEAWAT for large dataset land cover classification using different satellite imagery. In addition, it exposed limitations of the tool, indicating that various adjustments will be needed on the tool when working with other satellite imagery different from MODIS and Landsat. | en_ZA |
| dc.description.abstract | AFRIKAANS OPSOMMING: Die verwerking van groot volumes geografiese inligtingstelsel- (GIS) en afstandswaarnemingsdata noodsaak die ontwikkeling van outomatiese tegnieke wat koste-doeltreffend, vinnig en gebruikersvriendelik is ten einde ruimtelike besluitneming te ondersteun. In hierdie studie is ʼn geoutomatiseerde tegniek vir die identifisering van landbou-verwante landbedekking met behulp van ʼn pasgemaakte instrument ontwikkel. Veelvuldige geheelklassifiseerder in ArcGIS outomatiese instrument (MEAWAT) is op die MODIS genormaliseerde verskil plantegroei-indeks (GVPI) tydreeksgegewens van die Bergrivieropvangsarea in die Wes-Kaap, Suid-Afrika, getoets. Alhoewel die instrument ontwikkel is om landbou-verwante landbedekking met behulp van MODIS-data te klassifiseer, is die instrument ook op Landsat GVPI-data vir dieselfde studiegebied toegepas. Die instrument is effens aangepas sodat verskillende satellietbeeldtipes geakkommodeer kon word. Die instrument is op die ArcGIS/Python-platform gebou en die GIS- en afstandswaarnemingfunksies wat gewoonlik deur ʼn verskeidenheid sagtewarepakkette vervul word, is geïntegreer, insluitende die seleksie van die studie-area, herskatting, klassifikasie en assessering van akkuraatheid. Die GVPI-fenologiekurwe is gebruik om opleidingsdata vir die klassifikasie te skep. Verskillende parameters, watgebruikers in staat stel om verskeie reëls te gebruik om ʼn geskikte grondbedekkingkaart vir hulle doeleindes te ontwikkel, is getoets. Die MEAWAT-instrument gebruik beslissingsbome en geheelklassifiseerders soos ewekansige-woud en ekstra boom, asook versterking deur middel van ʼn meta-beramer (AdaBoost). Klassifikasie-akkuraatheid van onderskeidelik 70.5%, 75.5%, 76.3% en 78.7% is met die MODIS-data verkry, terwyl 89% akkuraatheid van die Landsat-data met behulp van die versterkte ewekansige-woudklassifiseerder verkry is. Dit is waargeneem dat ʼn beter klassifikasie afgelei kan word deur MEAWAT op hoër resolusie satellietbeelde toe te pas, maar slegs indien goeie opleidingsdata beskikbaar is. Hierdie bevindinge beklemtoon die potensiaal van MEAWAT vir die klassifikasie van groot landbedekkingdatastelle deur van verskillende satellietbeelde gebruik te maak. Dit het ook beperkings van die instrument aan die lig gebring, wat aandui dat verskeie aanpassings nodig sal wees wanneer satellietbeelde wat van MODIS en Landsat verskil gebruik word. | af_ZA |
| dc.description.version | Masters | en-ZA |
| dc.format.extent | xii, 127 pages | en_ZA |
| dc.identifier.uri | http://hdl.handle.net/10019.1/100017 | |
| dc.language.iso | en_ZA | en_ZA |
| dc.publisher | Stellenbosch : Stellenbosch University | en_ZA |
| dc.rights.holder | Stellenbosch University | en_ZA |
| dc.subject | MODIS (Spectroradiometer) | en_ZA |
| dc.subject | Landsat satellites | en_ZA |
| dc.subject | Normalized Difference Vegetation Index | en_ZA |
| dc.subject | NDVI | en_ZA |
| dc.subject | Land cover | en_ZA |
| dc.subject | MEAWAT | en_ZA |
| dc.subject | Multiple Ensemble Classified in ARcGIS Workflow Automation Tool | en_ZA |
| dc.subject | Remote sensing | en_ZA |
| dc.subject | UCTD | |
| dc.title | Automating land cover classification using time series NDVI : a case study in the Berg River Catchment Area | en_ZA |
| dc.type | Thesis | en_ZA |