dc.contributor.author	Bashir, Rokni Deilami
dc.date.accessioned	2023-04-19T01:08:32Z
dc.date.available	2023-04-19T01:08:32Z
dc.date.issued	2018-04
dc.description	Thesis (Ph.D (Remote Sensing))
dc.description.abstract	Land use and land cover change maps are usually utilized in planning and decision making processes for environmental monitoring and management. Therefore, identification of the certainty and reliability of these maps are very important. Unfortunately, in many studies only the pixel-based spectral and probabilistic measures as obtained from the classification approaches are used for uncertainty estimation. In this study, a new approach has been introduced which is based on the pixel-based as well as the spatial parameters to achieve the objective of this study which is to determine uncertainties and their propagation in dynamic change detection based on classified remotely-sensed images. Landsat TM and ETM+ images of Iskandar Malaysia, acquired in 2006 and 2016 have been coregistered using the first order polynomial. This study uses Maximum Likelihood classification procedure. A post classification comparison change detection technique was used to estimate major change between different land classes. This study revealed that built-up area increases and agriculture decreases by 8,736.48 hectare and 9,531.54 hectare respectively for the whole area. For calculation of uncertainty, a smaller area was selected. Using different spatial analysis functions for modeling of uncertainty of change from agriculture to urban areas, the relevant spectral parameters that include probability for each class (probabilistic) and uncertainty of classification for each year (classification uncertainty measures), differences of Normalized Differencing Vegetation Index (NDVI) and Normalized Differencing Build-up Index (NDBI) and spatial parameters that include distance from road and urban area were extracted. The spectral and spatial parameters have been integrated through the logistic regression modelling approach to model uncertainty of change from agriculture to urban areas. According to regression coefficient results, the probabilistic measures indicated by the probability of urban area in 2006 and NDVI and NDBI differencing displays the highest contribution for modelling of uncertainty in change detection. The spatial parameters do not indicate main contribution in the model; their signs of coefficients are in good agreements with those of the theoretical expectations. The resulting data of uncertainty for change from the agriculture to urban areas indicates that uncertainty is low in the change area. The Relative Operating Characteristics (ROC) index has been used for validation of the model and it has been estimated to be 0.9979, which is an indicator of a good model fitting. Also, the relationship between intensity of change and uncertainty map shows that in high intensity change area, uncertainty is low and accuracy is high
dc.description.sponsorship	Faculty of Geoinformation and Real Estate
dc.identifier.uri	http://openscience.utm.my/handle/123456789/242
dc.language.iso	en
dc.publisher	Universiti Teknologi Malaysia
dc.subject	Geoinformation and real estate
dc.subject	Land use
dc.subject	Agriculture
dc.title	Uncertainty determination in land use and land cover change detection based on remotely sensed images
dc.type	Thesis
dc.type	Dataset

Uncertainty determination in land use and land cover change detection based on remotely sensed images

Files

Original bundle

License bundle

Collections