Comparative Analysis of Machine Learning Algorithms in Automatic Identification and Extraction of Water Boundaries

Abstract

Monitoring open water bodies accurately is important for assessing the role of ecosystem services in the context of human survival and climate change. There are many methods available for water body extraction based on remote sensing images, such as the normalized difference water index (NDWI), modified NDWI (MNDWI), and machine learning algorithms. Based on Landsat-8 remote sensing images, this study focuses on the effects of six machine learning algorithms and three threshold methods used to extract water bodies, evaluates the transfer performance of models applied to remote sensing images in different periods, and compares the differences among these models. The results are as follows. (1) Various algorithms require different numbers of samples to reach their optimal consequence. The logistic regression algorithm requires a minimum of 110 samples. As the number of samples increases, the order of the optimal model is support vector machine, neural network, random forest, decision tree, and XGBoost. (2) The accuracy evaluation performance of each machine learning on the test set cannot represent the local area performance. (3) When these models are directly applied to remote sensing images in different periods, the AUC indicators of each machine learning algorithm for three regions all show a significant decline, with a decrease range of 0.33–66.52%, and the differences among the different algorithm performances in the three areas are obvious. Generally, the decision tree algorithm has good transfer performance among the machine learning algorithms with area under curve (AUC) indexes of 0.790, 0.518, and 0.697 in the three areas, respectively, and the average value is 0.668. The Otsu threshold algorithm is the optimal among threshold methods, with AUC indexes of 0.970, 0.617, and 0.908 in the three regions respectively and an average AUC of 0.832.

Assessment of Riverbank Erosion Hotspots along the Mekong River in Cambodia Using Remote Sensing and Hazard Exposure Mapping

“…Large-scale riverbank monitoring requires intensive resources and continuity, which is vital for the Mekong River Basin, where several developments on the main river and its tributaries significantly affect sedimentation. In recent years, machine learning algorithms have helped to improve the accuracy of water body detection; however, they are still in the developmental stage [50]. Hence, the water index method is more suitable for detecting large-scale riverbanks along the Mekong River and assessing changes in the river body over time.…”

“…Machine learning methods were found to outperform the water index-based method in specific areas and terrains [42][43][44][45][46]. However, when water detection was performed in the same location at different image acquisition times using the same machine learning algorithm, accuracy was significantly reduced [50], requiring the model to be retrained, which was resource intensive. Machine learning algorithms help to improve accuracy; however, they are still in the developmental stage [50].…”

“…However, when water detection was performed in the same location at different image acquisition times using the same machine learning algorithm, accuracy was significantly reduced [50], requiring the model to be retrained, which was resource intensive. Machine learning algorithms help to improve accuracy; however, they are still in the developmental stage [50]. Based on the above literature, large scale riverbank detection along the Mekong River and the assessment of river body changes over time, the water index method was selected for application in this study.…”

Development of sensor system and data analytic framework for non-invasive blood glucose prediction

“…Predictive models are selected by comparing and analyzing various regression algorithms from the literature. From the profound literature that discussed on the rationale behind the selection of regression algorithms, it was reported that the performance of predictive models depends on the methodologies implemented, the dataset created 50 , 64 – 69 , the size and heterogeneity of the dataset 70 . Regression algorithms differ in their principles of operation, advantages, and limitations.…”

DUPnet: Water Body Segmentation with Dense Block and Multi-Scale Spatial Pyramid Pooling for Remote Sensing Images

“…Machine learning-based water body segmentation algorithms build a relationship between water body samples and masks, which reduces the reliance on segmentation thresholds. Many popular algorithms such as Support Vector Machine (SVM) [9], Random Forests (RF) [10], Decision Tree (DT), and Deep Learning (DL) have been developed in remote sensing image segmentation [11,12]. DL has attracted more attention in image segmentation mainly due to its strong ability to extract variables to express feature information [13], which boosts the intelligent and automatic interpretation of remote sensing images.…”