Accurate and reliable informal settlement maps are fundamental decision-making tools for planning, and for expediting informed management of cities. However, extraction of spatial information for informal settlements has remained a mammoth task due to the spatial heterogeneity of urban landscape components, requiring complex analytical processes. To date, the use of Google Earth Engine platform (GEE), with cloud computing prowess, provides unique opportunities to map informal settlements with precision and enhanced accuracy. This paper leverages cloud-based computing techniques within GEE to integrate spectral and textural features for accurate extraction of the location and spatial extent of informal settlements in Durban, South Africa. The paper aims to investigate the potential and advantages of GEE’s innovative image processing techniques to precisely depict morphologically varied informal settlements. Seven data input models derived from Sentinel 2A bands, band-derived texture metrics, and spectral indices were investigated through a random forest supervised protocol. The main objective was to explore the value of different data input combinations in accurately mapping informal settlements. The results revealed that the classification based on spectral bands + textural information yielded the highest informal settlement identification accuracy (94% F-score). The addition of spectral indices decreased mapping accuracy. Our results confirm that the highest spatial accuracy is achieved with the ‘textural features’ model, which yielded the lowest root-mean-square log error (0.51) and mean absolute percent error (0.36). Our approach highlights the capability of GEE’s complex integrative data processing capabilities in extracting morphological variations of informal settlements in rugged and heterogeneous urban landscapes, with reliable accuracy.
6 April 2021 in South African Geographical Journal
This research investigates the distribution and magnitude of vegetation loss in the Save catchment. The main objective is to map and quantify human-induced land degradation in the Save catchment, using Residual trend analysis (RESTREND) method. The investigation was done using Moderate Resolution Imaging Spectroradiometer (MODIS) NDVI time series data, and gridded precipitation datasets from Climate Research Unit, recorded between 2000 and 2015. NDVI and rainfall time series, as well as ordinary least squares regression models used in the analysis, were computed in R statistical program. Geographic Information System (GIS) techniques were used in the quantification of degradation trends. The study revealed that approximately 18% of Save catchment experienced declining residual trends whilst increasing residual trends covered 34% of the area. These trends covered 1.7 Mha and 3.1 Mha, respectively. The study calculated the statistical significance of the declining trends. The results demonstrate that about 3.6 Mha experienced significant human-induced land degradation during the study period. This area represents 39% of the Save Catchment, 4%, 13%, and 22% of which were classified as severely, moderately, and lightly degraded, respectively. The results indicate the vulnerability of Save catchment to human-induced degradation. Severe degradation was noted in the central districts of the Save Catchment, notably Bikita, Chiredzi, and most parts of Chipinge. These findings demonstrate the effectiveness of RESTREND in removing influence of precipitation changes from land degradation. Based on these results, we recommend for the use of RESTREND method in detecting land degradation that is triggered by human actions.
Texture-based informal settlement (IS) mapping has in recent years gained attention due to the introduction of very high-resolution (VHR) imagery and advances in spatial analytics. Numerous studies conducted on the use of texture analysis for IS mapping have investigated wide-ranging sensors, algorithms, scale, classifiers, feature selection and other factors of interest. No study has systematically investigated all those relevant factors affecting texture-based classification processes. This work thus presents a detailed synthesis of the scientific progress in texture analysis in IS mapping. Results of the analysis revealed that grey level co-occurrence matrix (GLCM) was the most popularly used approach, with 73% of published articles exploiting various GLCM texture features in different geographical settings. Quickbird imagery, which was used in 31.7% of the studies, was the widely adopted sensor. For classification, the use of machine-learning classifiers, particularly, support vector machine (SVM) and Random Forest (RF) yielded high accuracy levels (>80%). Also, deep learning was identified as having the potential to advance informal settlement identification. Results of multi-city comparisons demonstrated the need for texture features to be locally specific in order to enhance the transferability of tested algorithms. Considering the relevance of morphology in informal settlement analysis, integration of remote sensing data and field survey statistics would enhance understanding of region-specific morphological variations that would enhance informal settlement mapping accuracy.
2 September 2022 in Transactions of the Royal Society of South Africa
The diversity of informal settlement morphologies across locales makes their mapping inherently challenging in heterogeneous urban landscapes. The aim of this study was to evaluate the potential of pansharpening techniques on Sentinel 2A data, and textural features, in enhancing informal settlement identification accuracy in a fragmented urban environment. Brovey transform, intensity, hue and saturation transform, Environmental Systems Research Institute (ESRI), simple mean, and Gram–Schmidt techniques were employed to pansharpen multispectral bands of Sentinel 2A, bands 5, 6, and 7 in the first group, and bands 8A, 11 and 12 in another, using an average of bands 4 and 8 as the panchromatic band. The main objective was to investigate the efficacy of pansharpening Sentinel 2A imagery and texture analysis in automated mapping of morphologically varied informal settlements. An evaluation of the quality of fused images was undertaken through computation of the correlation between the spectral values of the original multispectral and pansharpened image. Grey-level-co-occurrence matrix texture features were extracted from the pansharpened images, and subsequently incorporated in the classification process, using a support vector machine classifier. Our results confirm that the Gram–Schmidt fusion technique yielded the highest informal settlement identification accuracy (F-score 95.2%; overall accuracy 91.8%). The experimental results demonstrated the potential of pansharpening Sentinel 2A, and the added value of image texture for a more nuanced characterisation of informal settlements.