Sürdürülebilir Kalkınma Amaçları
Yazarlar
Yayınlar
- MakaleAçık Erişim2021Satellite-derived shallow wetland bathymetry using different classification algorithms and datasets
Dervisoglu, Adalet; Yagmur, Nur; Bilgilioglu, Burhan Baha
Remote sensing (RS) effectively identifies, analyzes, and monitors wetlands. In addition to these two-dimensional studies, RS is used with several techniques in determining shallow water depths. The primary purpose of this study is to obtain shallow wetland bathymetry utilizing spectral reflections obtained at different water depths by field study and satellite images. Machine learning (ML) algorithms, which are widely used in remote sensing, are used in this study. Four algorithms were selected as random forest (RF), support vector machine (SVM), Neural Networks (NN), and Maximum Likelihood Classification (MLC). Since machine learning algorithms use training samples/datasets, the classification accuracy is directly related to selecting these data. The effect of pixel counts on classification was investigated by using two different training data set also. Duden (Kulu) Lake, which is a shallow wetland, was chosen as the study area. The Iterative Self-Organizing Data Analysis Technique (ISODATA) classification algorithm divided into as many clusters as possible was applied on Sentinel-2 multispectral images. All classes were rede-fined using measured spectral signatures and were created a bathymetric map. This map was used as reference data in creating training sets and the accuracy assessment of ML algorithms. When the water surface areas obtained from algorithms were compared with the bathymetric map and Normalized Difference Water Index, the best result was obtained with RF. According to the accuracy assessment results, it was seen that the number of training data affects the accuracy, and the best results were obtained with SVM and RF algorithms with training data containing more pixels (overall accuracy 93.87% and 92.64, kappa 0.89 and 0.87, respectively).
- MakaleAçık Erişim2021Investigation of the Effect of the Dataset Size and Type in the Earthquake-Triggered Landslides Mapping: A Case Study for the 2018 Hokkaido Iburu Landslides
Comert, Resul
Rapid mapping of landslides that occur after an earthquake is important for rapid crisis management. In this study, experimental research was conducted on the size of the model area and the data types used in developing classifiers for the supervised classification approaches used in rapid landslide mapping. The Hokkaido Iburu earthquake zone that occurred on September 6, 2018, was selected as the study area. PlanetScope pre-event and post-event images and ALOS-PALSAR Digital Elevation Model (DEM) were used in the analysis processes. In this context, five model areas with different sizes and one test area were determined. Object-based image analysis (OBIA) was used as a landslide mapping approach. Random Forest classifier, which is a supervised classification algorithm, was performed in the mapping of image objects produced by the segmentation stage of OBIA. Two different data sets were created for landslide mapping: change-based dataset and post-event dataset. The change-based dataset is generated from change data such as the difference of normalized difference vegetation index (δNDVI), change detection Image (CDI), princiable component analysis (PCA), and Independent component analysis (ICA) which are used in change detection applications. The post-event dataset was created from data generated from post-event image bands. When the obtained results were examined, higher accuracy results were obtained with the post-event dataset. Increasing the size of the model area, in other words, increasing the training data slightly increases the accuracy of landslide mapping. However, a model area that represents the region to be mapped in small sizes to make rapid decisions provides a 94% F-measure accuracy for earthquake-triggered landslide detection.
- MakaleAçık Erişim2019OBJECT BASED BURNED AREA MAPPING WITH RANDOM FOREST ALGORITHM
Comert, Resul; Matcı, Dilek Kucuk; Avdan, Ugur
It is very important to map the burned forest areas economically, quickly and with the high accuracy of issues such as damage assessment studies, fire risk analysis, and management of forest regeneration processes. Remote sensing methods give advantages such as fast, easy-to-use and high accuracy for burned area mapping. Recent years machine learning algorithms have become more popular in satellite image classification, due to the effective solutions for the analysis of complex datasets which have a large number of variables. In this study, the success of object based random forest algorithm was investigated for burned forest area mapping. For this purpose, Object based image analysis (OBIA) was performed using Landsat 8 image of the Adrasan and Kumluca fires which occurred in 24 – 27 June 2016. The study consisted of five steps. In the first step, the multi-resolution image segmentation was performed for obtaining image objects from Landsat 8 spectral bands. In the second step, the image object metrics such as spectral index and layer values were calculated for all image objects. In the third step, a random forest classifier model was developed. Then, the developed model applied to the test site for classification of the burned area. Finally, the obtained results evaluated with confusion matrix based on the randomly sampled points. According to the results, we obtained 0.089 commission error (CE) with 0.014 omission error (OE). An overall accuracy was obtained as 0.99. The results show that this approach is very useful to be used to determine burned forest areas.
- BildiriAçık Erişim2017THE EFFICIENCY of RANDOM FOREST METHOD for SHORELINE EXTRACTION from LANDSAT-8 and GOKTURK-2 IMAGERIES
Bayram, B.; Erdem, F.; Akpinar, B.; Ince, A.K.; Bozkurt, S.; Catal Reis, H.; Seker, D.Z.; Karas, I.R.; Rahman, A.A.; Isikdag, U.
Coastal monitoring plays a vital role in environmental planning and hazard management related issues. Since shorelines are fundamental data for environment management, disaster management, coastal erosion studies, modelling of sediment transport and coastal morphodynamics, various techniques have been developed to extract shorelines. Random Forest is one of these techniques which is used in this study for shoreline extraction.. This algorithm is a machine learning method based on decision trees. Decision trees analyse classes of training data creates rules for classification. In this study, Terkos region has been chosen for the proposed method within the scope of "TUBITAK Project (Project No: 115Y718) titled "Integration of Unmanned Aerial Vehicles for Sustainable Coastal Zone Monitoring Model – Three-Dimensional Automatic Coastline Extraction and Analysis: Istanbul-Terkos Example". Random Forest algorithm has been implemented to extract the shoreline of the Black Sea where near the lake from LANDSAT-8 and GOKTURK-2 satellite imageries taken in 2015. The MATLAB environment was used for classification. To obtain land and water-body classes, the Random Forest method has been applied to NIR bands of LANDSAT-8 (5th band) and GOKTURK-2 (4th band) imageries. Each image has been digitized manually and shorelines obtained for accuracy assessment. According to accuracy assessment results, Random Forest method is efficient for both medium and high resolution images for shoreline extraction studies.