Mapping CORINE Land Cover from Sentinel-1A SAR and SRTM Digital Elevation Model Data using Random Forests

Balzter, Heiko; Cole, Beth; Thiel, Christian; Schmullius, Christiane

doi:10.3390/rs71114876

Cited by 160 publications

(139 citation statements)

References 39 publications

Supporting

Mentioning

119

Contrasting

Unclassified

Order By: Relevance

“…RF was selected for this study because it generally outperforms conventional classifiers such as the Gaussian maximum likelihood classifier [61,62], while performing favorably, or equally well, to other non-parametric approaches; e.g., CART [63,64], Support Vector Machines [32,65,66], Artificial Neural Networks [67], and K-Nearest Neighbor [68]. It is a powerful non-linear and non-parametric classifier that allows for fusion and aggregation of high-dimensional data from various sources (e.g., optical, SAR, and topography [30,69,70]; SAR and topography [21,58,71]; and optical and topography [72][73][74]). RF produces independently constructed classification trees, similar to the Classification and Regression (CART) method, using bootstrapped samples of the original data [75,76].…”

Section: Image Classificationmentioning

confidence: 99%

Mapping the Dabus Wetlands, Ethiopia, Using Random Forest Classification of Landsat, PALSAR and Topographic Data

et al. 2017

View full text Add to dashboard Cite

Abstract:The Dabus Wetland complex in the highlands of Ethiopia is within the headwaters of the Nile Basin and is home to significant ecological communities and rare or endangered species. Its many interrelated wetland types undergo seasonal and longer-term changes due to weather and climate variations as well as anthropogenic land use such as grazing and burning. Mapping and monitoring of these wetlands has not been previously undertaken due primarily to their relative isolation and lack of resources. This study investigated the potential of remote sensing based classification for mapping the primary vegetation groups in the Dabus Wetlands using a combination of dry and wet season data, including optical (Landsat spectral bands and derived vegetation and wetness indices), radar (ALOS PALSAR L-band backscatter), and elevation (SRTM derived DEM and other terrain metrics) as inputs to the non-parametric Random Forest (RF) classifier. Eight wetland types and three terrestrial/upland classes were mapped using field samples of observed plant community composition and structure groupings as reference information. Various tests to compare results using different RF input parameters and data types were conducted. A combination of multispectral optical, radar and topographic variables provided the best overall classification accuracy, 94.4% and 92.9% for the dry and wet season, respectively. Spectral and topographic data (radar data excluded) performed nearly as well, while accuracies using only radar and topographic data were 82-89%. Relatively homogeneous classes such as Papyrus Swamps, Forested Wetland, and Wet Meadow yielded the highest accuracies while spatially complex classes such as Emergent Marsh were more difficult to accurately classify. The methods and results presented in this paper can serve as a basis for development of long-term mapping and monitoring of these and other non-forested wetlands in Ethiopia and other similar environmental settings.

show abstract

Section: Image Classificationmentioning

confidence: 99%

Mapping the Dabus Wetlands, Ethiopia, Using Random Forest Classification of Landsat, PALSAR and Topographic Data

et al. 2017

View full text Add to dashboard Cite

show abstract

“…As the volume scattering at C-Band in forests (representing permanently vegetated areas) and dense agricultural crops (representing seasonally vegetated areas) may cause similar backscatter values over these two classes in summer time acquisitions and thus make the two classes less separable (Balzter et al, 2015), only the winter period (1 st December 2014 to 31 st M arch 2015) data were selected. All acquisitions from the specified time frame were used, regardless of the environmental conditions (snow, precipitation, temperature) or differences in orbit and thus in local incidence angles.…”

Section: Datamentioning

confidence: 99%

“…The data from Sentinel-1 satellite were already demonstrated to be useful for land cover classification as a possibility to complement the cloud-covered areas (Balzter et al, 2015) or to be used for a forest change detection to identify clear-cuts (Olesk et al, 2015). Similarly, the dual-polarization data from Radarsat C-Band sensor were already used to complement forest cover maps from ALOS-PALSAR L-Band SAR systems (Anthea et al, 2014).…”

Section: Introductionmentioning

confidence: 99%

Forest Area Derivation From Sentinel-1 Data

Dostálová¹,

Hollaus²,

Milenković³

et al. 2016

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABS TRACT:The recently launched Sentinel-1A provides the high resolution Synthetic Aperture Radar (SAR) data with very high temporal coverage over large parts of European continent. Short revisit time and dual polarization availability supports its usability for forestry applications. The following study presents an analysis of the potential of the multi-temporal dual-polarization Sentinel-1A data for the forest area derivation using the standard methods based on Otsu thresholding and K -means clustering. Sentinel-1 data collected in winter season 2014-2015 over a test area in eastern Austria were used to derive forest area mask with spatial resolution of 10m and minimum mapping unit of 500m2. The validation with reference forest mask derived from airborne full-waveform laser scanning data revealed overall accuracy of 92% and kappa statistics of 0.81. Even better results can be achieved when using external mask for urban areas, which might be misclassified as forests when using the introduced approach based on SAR data only. The Sentinel-1 data and the described methods are well suited for forest change detection between consecutive years.

show abstract

“…The RF algorithm [33] is an ensemble classifier that consists of many decision trees that does not overfit. RF can be used for land cover mapping [34] and it often gives better land cover classification accuracies [35,36]. Moreover, RF with its multiple advantages such as handling large numbers of input variables and calculating an error matrix, giving estimates of the importance of the features in the classification fast and robust [32].…”

Section: Introductionmentioning

confidence: 99%

Post-War Land Cover Changes and Fragmentation in Halgurd Sakran National Park (HSNP), Kurdistan Region of Iraq

Hamad

Kolo

Balzter

2018

Land

Self Cite

View full text Add to dashboard Cite

Context:The fundamental driving force of land use and land cover (LULC) change is related to spatial and temporal processes caused by human activities such as agricultural expansion and demographic change. Landscape metrics were used to analyze post-war changes in a rural mountain landscape, the protected area of Halgurd-Sakran National Park (HSNP) in north-east Iraq. Therefore, the present work attempts to identify the temporal trends of the most fragmented land cover types between two parts of the national park. Objectives: The objectives of this study are to compare two land cover classification algorithms, maximum likelihood classification (MLC) and random forest (RF) in the upper and lower parts of HSCZ, and to examine whether landscape configuration in the park has changed over time by comparing the fragmentation, connectivity and diversity of LULC classes. Methods: Two Landsat images were used to analyze LULC fragmentation and loss of habitat connectivity (before and after the Fall of Baghdad in 2003). Seven landscape pattern metrics, percentage of land (PLAND), number of patch (NP), largest patch index (LPI), mean patch size (MPS), euclidian nearest neighborhood distance (ENN_AM), interspersion and juxtaposition (IJI) and cohesion at class level were selected to assess landscape composition and configuration. Results: A significant change in LULC classes was noticed in the lower part of the park, especially for pasture, cultivated and forest-lands. The fragmentation trends and their changes were observed in both parts of the park, however, more were observed in the lower part. The inherent causes of these changes are the socio-economic factors created by the 1991-2003 UN post-war economic sanctions. The changes increased during sanctions and decreased afterwards. The fall of Baghdad in 2003, followed by rapid economic boom, marked the greatest cause in land use change, especially in changes-susceptible cultivated areas. Conclusions: Shrinkage of forest patches in the lower part of the park increases the distance between them, which contributes to a decline in biological diversity from decreasing habitat area. Lastly, the results confirm the applicability of the combined method of remote sensing and landscape metrics.

show abstract

Mapping CORINE Land Cover from Sentinel-1A SAR and SRTM Digital Elevation Model Data using Random Forests

Cited by 160 publications

References 39 publications

Mapping the Dabus Wetlands, Ethiopia, Using Random Forest Classification of Landsat, PALSAR and Topographic Data

Mapping the Dabus Wetlands, Ethiopia, Using Random Forest Classification of Landsat, PALSAR and Topographic Data

Forest Area Derivation From Sentinel-1 Data

Post-War Land Cover Changes and Fragmentation in Halgurd Sakran National Park (HSNP), Kurdistan Region of Iraq

Contact Info

Product

Resources

About