Research to improve the accuracy of very high-resolution satellite image classification algorithms is still one of the hot topics in the field of remote sensing. Successful results of deep learning methods in areas such as image classification and object detection have led to the application of these methods to remote sensing problems. Recently, Convolutional Neural Networks (CNNs) are among the most common deep learning methods used in image classification, however, the use of CNN’s in satellite image classification is relatively new. Due to the high computational complexity of 3D CNNs, which aim to extract both spatial and spectral information, 2D CNNs focussing on the extraction of spatial information are often preferred. High-resolution satellite images, however, contain crucial spectral information as well as spatial information. In this study, a 3D-2D CNN model using both spectral and spatial information was applied to extract more accurate land cover information from very high-resolution satellite images. The model was applied on a Worldview-2 satellite image including agricultural product areas such as tea, hazelnut groves and land use classes such as buildings and roads. The results of the CNN based model were also compared against those of the Support Vector Machine (SVM) and Random Forest (RF) algorithms. The post-classification accuracies were obtained using 800 control points generated by a web interface created for crowdsourcing purposes. The classification accuracy was 95.6% for the 3D-2D CNN model, 89.2% for the RF and 86.4% for the SVM.

In accordance with current demands, the natural gas transmission pipeline (NGTP) is one of the most appropriate methods used in the distribution of existing reserves. Construction of long-distance pipelines requires large expenditures. Decreasing the time and cost of such construction and minimising environmental damage depend upon identifying the optimum routes from the onset of the project. Route determination is one of the most important steps in NGTP projects. The route determination process requires obtaining the existing graphic and non-graphic data from different institutions and organisations, as well as gathering, storing, querying and analysing non-existing data in an appropriate and efficient manner. Accessing the correct results rapidly by analysing such large data-sets can be achieved with spatial multicriteria decision-making technologies based on the geographic information system as an effective decision support tool. In this study, three methods were implemented for two NGTP projects of 103.60 and 60.89 km in length. At the end of this study, it was concluded that Spatial technique for order preference by similarity to ideal solution was the most effective of the three pipeline routing methods and that it could reduce project costs by approximately 21%.

Hyperspectral images can identify minerals better than Multispectral images because of their high spectral resolutions. However, a pixel might include more than one mineral, as hyperspectral images have low spatial resolution. In these situations, the number of minerals can be estimated in mixed pixels but their spatial positions cannot be known. This is one of the biggest obstacles that prevents effective use of hyperspectral images in mineral exploration. In order to overcome that obstacle, it is known that, fusing hyperspectral images with a better spatial resolution images generates a better content in the fused images. In this study, well known and publicly available Cuprite AVIRIS image is fused with Landsat 8- Panchromatic band. Then, the minerals in the Cuprite site are classified both on the fused image and the original AVIRIS image. When we compared the classification results, it is found that, minerals are classified with higher accuracy on the fused image compared to the AVIRIS image.

This study aims to generate optimum forest route. Contrary to the traditional manual methods, a dynamic route model has been developed. Along with a dynamic route model, an algorithm has been developed. In this study, a landslide susceptibility map that was generated by considering eight factors defined for the study area was used to investigate the risk status of the current forest roads. Then, the optimum forest routes are improved by considering potential landslide areas and several restrictive factors. Finally, the optimum forest road routes were compared to the current forest road routes. The results showed that the developed optimal path algorithm enabled the generation of the optimum forest road route. According to the risk results obtained for current forest roads; 559 km of all roads are risky (%37) and 33 km are very risky (%2). In addition, 39% of the existing roads were found to be under landslide risk.