Redefining Homogeneous Climate Regions in Bangladesh Using Multivariate Clustering Approaches

The knowledge of the climate pattern for a particular region is important to alleviate the impact of climate change and protect the environment by taking appropriate actions based on geographical knowledge. It is also equally important for water resources planning and management. In this study, the regional disparities and similarities have been revealed among different climate stations or regions in Bangladesh based on different climatological factors such as rainfall, temperatures, relative humidity, sea level pressure, cloud cover, wind speed, the sunshine hour. We have selected one of the best-fitted algorithms for particular climate data from three multivariate clustering approaches named hierarchical clustering, partitioning around medoids (PAM), and K-means clustering by using different validation tests. Four homogeneity tests (Mann-Kendall Test, Pettitt's test, Buishand Range Test, Standard Normal Homogeneity Test) also have been performed for each of the clusters created based on several factors. The results suggest that the climate regions or meteorological stations of Bangladesh can be clustered into two groups based on a combination of climatological variables. According to the findings, there is a huge variation between the two groups in terms of climatological factors. The first group (cluster 1) is located in the northern part of the country that includes drought-prone and vulnerable regions, whereas, the second group (cluster 2) contains rain-prone and hilly regions, which are mostly situated in the southern part. All newly defined clusters show homogeneous behavior with few exceptions such as clusters based on sea level pressure are not homogeneous.

Spatial Segmentation of Mass Spectrometry Imaging Data by Combining Multivariate Clustering and Univariate Thresholding

<p>Spatial segmentation partitions mass spectrometry imaging (MSI) data into distinct regions providing a concise visualization of the vast amount of data and identifying regions of interest (ROIs) for downstream statistical analysis. Unsupervised approaches are particularly attractive as they may be used to discover the underlying subpopulations present in the high-dimensional MSI data without prior knowledge of the properties of the sample. Herein, we introduce an unsupervised spatial segmentation approach, which combines multivariate clustering and univariate thresholding to generate comprehensive spatial segmentation maps of the MSI data. This approach combines matrix factorization and manifold learning to enable high-quality image segmentation without an extensive hyperparameter search. In parallel, some ion images inadequately represented in the multivariate analysis are treated using univariate thresholding to generate complementary spatial segments. The final spatial segmentation map is assembled from segment candidates generated using both techniques. We demonstrate the performance and robustness of this approach for two MSI data sets of mouse uterine and kidney tissue sections acquired with different spatial resolutions. The resulting segmentation maps are easy to interpret and project onto the known anatomical regions of the tissue.</p>

Spatial Segmentation of Mass Spectrometry Imaging Data by Combining Multivariate Clustering and Univariate Thresholding

<p>Spatial segmentation partitions mass spectrometry imaging (MSI) data into distinct regions providing a concise visualization of the vast amount of data and identifying regions of interest (ROIs) for downstream statistical analysis. Unsupervised approaches are particularly attractive as they may be used to discover the underlying subpopulations present in the high-dimensional MSI data without prior knowledge of the properties of the sample. Herein, we introduce an unsupervised spatial segmentation approach, which combines multivariate clustering and univariate thresholding to generate comprehensive spatial segmentation maps of the MSI data. This approach combines matrix factorization and manifold learning to enable high-quality image segmentation without an extensive hyperparameter search. In parallel, some ion images inadequately represented in the multivariate analysis are treated using univariate thresholding to generate complementary spatial segments. The final spatial segmentation map is assembled from segment candidates generated using both techniques. We demonstrate the performance and robustness of this approach for two MSI data sets of mouse uterine and kidney tissue sections acquired with different spatial resolutions. The resulting segmentation maps are easy to interpret and project onto the known anatomical regions of the tissue.</p>