We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Data Integration for Lithological Mapping Using Machine Learning Algorithms.
- Authors
Manap, Hatice Seval; San, Bekir Taner
- Abstract
The aim of this study is to compare and evaluate the performances of different classification algorithms (Maximum Likelihood Classification [MLC], Random Forest [RF], Support Vector Machine [SVM], and Neural Network [NN]) with data integration for lithological classification in partly vegetated areas. The optical (ASTER), SAR (Sentinel-1) images and Digital Elevation Model (DEM) data were used as classification inputs. The study area used in this work is located in the western part of Antalya Province of Turkey and is characterized by relatively high relief and good exposure of numerous lithological classes. Five data sets were used for input to the classifiers: (1) Sentinel-1 image with its polarimetric bands (i.e. VV, HH, VH, HV), (2) ASTER image, (3) ASTER image with DEM, (4) the integration of ASTER and Sentinel-1 images, and (5) the integration of ASTER, DEM and Sentinel-1 data. Classification of the five input data sets by each of the four classifiers generated a total of 20 output images. The SVM algorithm produced the best overall classification accuracies (averaging 94.18% of three models for all data integrated) and the MLC algorithm produced the worst classification accuracy (averaging 35.75% of three models when using the Sentinel-1 polarimetric images as input). Regardless of the classifier algorithms, the accuracy values obtained in the classification using only ASTER data were found to be between 81.34% and 88.09% for averaging all models. On the other hand, when Sentinel-1 and DEM data were added to the optical data sets, the classification accuracies for the average of all models increased to values between 90.18% and 94.18%.
- Subjects
SENTINEL-1 (Artificial satellite); DATA integration; SUPPORT vector machines; CLASSIFICATION algorithms; DIGITAL elevation models; RANDOM forest algorithms; MACHINE learning
- Publication
Earth Science Informatics, 2022, Vol 15, Issue 3, p1841
- ISSN
1865-0473
- Publication type
Article
- DOI
10.1007/s12145-022-00826-3