The production of thematic maps using an image classification is one of the most common applications of remote sensing; such as the production of Land Use and Land cover maps which derived from photo-interpretation of satellite images and Aerial photos. Considerable research has been directed at the various components of the mapping process, including the assessment of accuracy, which is the focused subject of the Thesis. As the result of the “Global Land Cover Mapping at Finer Resolution” project led by National Geomatics Center of China, the first 30-meters resolution Open Global Land Cover Dataset (named GlobeLand30) have been produced for the years 2000 and 2010. This important dataset is free, open and downloaded through Website: http://www.globallandcover.com:6677/InfoServer/GLC/DownLoad.aspx?No=0. The objective of the current study is the assessment of the thematic accuracy of these data on the Italian area by means of a benchmarking with the more detailed land cover open datasets available for many Italian regions. These datasets are independent each other and provided with different thematic classification and resolutions. The accuracy assessment is based on the cell-by-cell comparison between Italian regional maps and the GlobeLand30 in order to obtain the confusion matrix and all its derived agreement statistics (overall accuracy, producer’s and user’s accuracy, allocation and quantity disagreement, etc.), which help to understand the Classification Quality. The Thesis presents the Methodology and Procedures for assessing open GlobeLand30. The analysis has been performed in 8 regions across Italy by taking advantage of GRASS, an open-source GIS which offers advanced features for geospatial data processing and analysis. The results of the assessment was very good. This has proven that the combination of open data and open source software provides us a new way for spatial analyses and geographical applications.
La produzione di carte tematiche mediante tecniche di classificazione d’immagine rappresenta una delle applicazioni più comuni del telerilevamento; un tipico esempio in questo senso è la realizzazione di carte di copertura e uso del suolo a partire da fotointerpretazione di immagini satellitari o aeree. Tra i numerosi studi di ricerca condotti in tale ambito notevole importanza riveste la valutazione dell’accuratezza del processo di classificazione, tema sul quale è incentrato il presente lavoro. Nell’ambito del progetto "Global Land Cover Mapping at Finer Resolution" guidato dal National Geomatics Center of China, è stato prodotto, per gli anni 2000 e 2010, il primo dataset globale di copertura del suolo con risoluzione 30 metri (chiamato GlobeLand30); questo importante dataset è gratuito, aperto e scaricabile tramite il sito web: http://www.globallandcover.com:6677/InfoServer/GLC/DownLoad.aspx?No=0. L'obiettivo del presente studio è la valutazione dell’accuratezza tematica dei dati GlobeLand30 sul territorio italiano attraverso un’analisi comparativa con le più dettagliate carte di copertura del suolo rese disponibili come “open data” da varie regioni italiane; tali dataset sono indipendenti tra loro e caratterizzati da classificazione tematica e risoluzioni diverse. La stima dell’accuratezza è stata condotta tramite una procedura che, attraverso un confronto cella per cella tra le carte regionali e la GlobeLand30, ha permesso di calcolare la matrice di confusione e derivare da essa una serie di statistiche (accuratezza globale, accuratezza del produttore, accuratezza dell’utente, grado di discordanza tematica e spaziale etc.) in grado di descrivere la qualità della classificazione. La tesi presenta la metodologia e le procedure adottate per la valutazione di GlobeLand30. Le analisi, condotte in 8 regioni italiane con risultati molto soddisfacenti, sono state eseguite mediante GRASS, GIS open-source che offre funzionalità avanzate per l'elaborazione e l’analisi di dati geospaziali. Lo studio dimostra come la combinazione di open data e software open source possa fornire una nuova possibilità nello svolgimento di analisi spaziali e nello sviluppo di applicazioni geografiche.
A first complete benchmarking of the new Chinese 30 m. resolution global land cover dataset (GLC30) and regional land coverage datasets in Italy
HUSSEIN, EMAN HUSSEIN MOHAMED
2013/2014
Abstract
The production of thematic maps using an image classification is one of the most common applications of remote sensing; such as the production of Land Use and Land cover maps which derived from photo-interpretation of satellite images and Aerial photos. Considerable research has been directed at the various components of the mapping process, including the assessment of accuracy, which is the focused subject of the Thesis. As the result of the “Global Land Cover Mapping at Finer Resolution” project led by National Geomatics Center of China, the first 30-meters resolution Open Global Land Cover Dataset (named GlobeLand30) have been produced for the years 2000 and 2010. This important dataset is free, open and downloaded through Website: http://www.globallandcover.com:6677/InfoServer/GLC/DownLoad.aspx?No=0. The objective of the current study is the assessment of the thematic accuracy of these data on the Italian area by means of a benchmarking with the more detailed land cover open datasets available for many Italian regions. These datasets are independent each other and provided with different thematic classification and resolutions. The accuracy assessment is based on the cell-by-cell comparison between Italian regional maps and the GlobeLand30 in order to obtain the confusion matrix and all its derived agreement statistics (overall accuracy, producer’s and user’s accuracy, allocation and quantity disagreement, etc.), which help to understand the Classification Quality. The Thesis presents the Methodology and Procedures for assessing open GlobeLand30. The analysis has been performed in 8 regions across Italy by taking advantage of GRASS, an open-source GIS which offers advanced features for geospatial data processing and analysis. The results of the assessment was very good. This has proven that the combination of open data and open source software provides us a new way for spatial analyses and geographical applications.| File | Dimensione | Formato | |
|---|---|---|---|
|
09.2014.Final Thesis, Eman Hussein Mohamed Hussein.pdf
accessibile in internet per tutti
Descrizione: Eman Hussein Mohamed Hussein-Final Thesis
Dimensione
8.04 MB
Formato
Adobe PDF
|
8.04 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/95085