Patch Clustering Research Articles

Surface coal mining in drylands: A multiscale comparison of spatiotemporal patterns and environmental impacts between Inner Mongolia and Mongolia

Surface coal mining (SCM) poses a great threat to the environment. Previous studies have explored the speed and scale of SCM in the Mongolian Plateau, but landscape-based analysis is needed for creating actionable knowledge required for environmental policy-making. Thus, taking a landscape ecological approach, here we compared the spatiotemporal patterns and major environmental impacts of SCM between Inner Mongolia of China and Mongolia during 1975–2015 at multiple administrative levels. We found that the SCM area increased by nearly 40 times in Inner Mongolia and 11 times in Mongolia during the 40 years. The annual increase rate in terms of both area and number was greater in Inner Mongolia than in Mongolia during 1975–2010, but the order was reversed during 2010–2015. At the prefectural or aimag level, the SCM distribution exhibited considerable variations. In 2015, 44 % of the total SCM area was located in Baotou, Wuhai, and Ordos of Inner Mongolia and Ömnögovi of Mongolia in 2015. The spatiotemporal patterns of SCM were characterized by increases in patch size, shape complexity, clustering, and landscape fragmentation. We estimated that the surrounding ecosystems disturbed by mining were 14.72 times larger than the SCM sites themselves in Inner Mongolia and 21.10 times in Mongolia. More threatened species were potentially affected by SCM in Inner Mongolia than in Mongolia. The variations in the scope and speed of SCM between Inner Mongolia and Mongolia may be attributable to multiple factors, including the distribution of coal mines themselves, economic investments, and national and local policies. Our study provides scientific support for promoting China-Mongolia bilateral collaboration for curbing SCM expansion and mitigating its environmental impacts on the plateau.

Test-based patch clustering for automatically-generated patches assessment

Previous studies have shown that Automated Program Repair (apr) techniques suffer from the overfitting problem. Overfitting happens when a patch is run and the test suite does not reveal any error, but the patch actually does not fix the underlying bug or it introduces a new defect that is not covered by the test suite. Therefore, the patches generated by apr tools need to be validated by human programmers, which can be very costly, and prevents apr tool adoption in practice. Our work aims to minimize the number of plausible patches that programmers have to review, thereby reducing the time required to find a correct patch. We introduce a novel light-weight test-based patch clustering approach called xTestCluster, which clusters patches based on their dynamic behavior. xTestCluster is applied after the patch generation phase in order to analyze the generated patches from one or more repair tools and to provide more information about those patches for facilitating patch assessment. The novelty of xTestCluster lies in using information from execution of newly generated test cases to cluster patches generated by multiple APR approaches. A cluster is formed of patches that fail on the same generated test cases. The output from xTestCluster gives developers a) a way of reducing the number of patches to analyze, as they can focus on analyzing a sample of patches from each cluster, b) additional information (new test cases and their results) attached to each patch. After analyzing 902 plausible patches from 21 Java apr tools, our results show that xTestCluster is able to reduce the number of patches to review and analyze with a median of 50%. xTestCluster can save a significant amount of time for developers that have to review the multitude of patches generated by apr tools, and provides them with new test cases that expose the differences in behavior between generated patches. Moreover, xTestCluster can complement other patch assessment techniques that help detect patch misclassifications.

Effect of fluid patch clustering on the P-wave velocity-saturation relation: a critical saturation model

Quantitative analysis of the relationship between seismic wave velocities and fluid saturation in porous media is of great significance for any fluid injection and extraction operation in subsurface rock formations. However, seismic velocities are not only dependent on the amount of saturation, but also on the distribution of fluid patches and their size. The patch size variation during changes in saturation is oftentimes ignored in modeling studies, even though it is natural to assume that with increasing saturation, fluid patches will form larger and, at some critical saturation, percolating clusters. To capture the evolution of patch size with saturation implied in the velocity-saturation relations, we are inspired by percolation theory. By incorporating the connectivity of water-filled patches in the continuous random medium model, we develop a critical saturation model. We apply this critical saturation model to examine recently reported experimental measurements, specifically analyzing the patch size changes. For measurements of drainage or imbibition processes in four sandstone samples, we indeed find a clear indication of growing patch size with water saturation. The predictions of the critical saturation model are in reasonable agreement with observations. Our approach improves the accuracy of the interpretation of the velocity-saturation relations in partially saturated rocks and forms a basis for exploring its underlying mechanisms.

Landscape configuration and storm characteristics drive spatial patterns of wind disturbance in boreal forest landscapes

ContextWind is an important disturbance in circumboreal forests, and its frequency and severity may change with climate change, highlighting the need to understand the drivers of wind disturbance. Currently, how landscape configuration drives wind disturbance is poorly understood.ObjectivesWe investigated whether and how landscape configuration is related to the extent and spatial pattern of wind disturbance, and how these relationships vary between windstorms and thunderstorms.MethodsWe used salvage logging data after 16 storms that occurred in Finland between 2011 and 2021. We placed a total of 301 landscapes, each encompassing an area of 8024 ha, within the storm tracks and used regression models to test how wind disturbance extent, disturbance patch size, number of disturbance patches, and disturbance patch clustering were related to landscape configuration and storm characteristics.ResultsIncreasing mean gap size and edge density, including permanent openings (e.g., lakes) and recent harvest gaps, increased disturbance extent, disturbance patch size, and number of disturbance patches. Conversely, increasing mean harvest gap size decreased disturbance patch clustering. Increasing wind speed had the largest contribution to increasing disturbance extent and number of disturbance patches, and decreasing disturbance patch clustering, with the magnitude of the effect varying between windstorms and thunderstorms.ConclusionsThe extent and spatial pattern of wind disturbances varied with landscape configuration and storm characteristics. Disturbance patches were larger in landscapes with large canopy gaps, resulting in a greater disturbance extent, exacerbated by increasing wind speed and thunderstorm development.

RefQSR: Reference-Based Quantization for Image Super-Resolution Networks.

Single image super-resolution (SISR) aims to reconstruct a high-resolution image from its low-resolution observation. Recent deep learning-based SISR models show high performance at the expense of increased computational costs, limiting their use in resource-constrained environments. As a promising solution for computationally efficient network design, network quantization has been extensively studied. However, existing quantization methods developed for SISR have yet to effectively exploit image self-similarity, which is a new direction for exploration in this study. We introduce a novel method called reference-based quantization for image super-resolution (RefQSR) that applies high-bit quantization to several representative patches and uses them as references for low-bit quantization of the rest of the patches in an image. To this end, we design dedicated patch clustering and reference-based quantization modules and integrate them into existing SISR network quantization methods. The experimental results demonstrate the effectiveness of RefQSR on various SISR networks and quantization methods.

Harvest block aggregation as a driver of intensive moose browsing pressure on hardwood regeneration in a temperate forest

In many regions of the world, browsing pressure by ungulates has increased over the last century, often reaching unsustainable levels. Overbrowsing by ungulates is threatening biodiversity and causing delays in the establishment of forest regeneration. Moose (Alces alces) is considered as an ecological engineer, as it can cause severe impacts on stand structure, composition, quality, and succession through overbrowsing. In conifer-dominated stands, moose browsing pressure has been shown to vary with the distribution of forage and cover patches across the landscape, but such spatial relationships have not been explored in hardwood-dominated landscapes, where intensive forest management can create clustered patches of good-quality forage and cover, ideal conditions for wintering moose. We investigated the influence of local and landscape variables on browsing by North American moose (Alces alces americana) on commercial and non-commercial woody regeneration in an intensively managed forest landscape of northwestern New Brunswick, Canada. We predicted that browsing pressure at the local scale (vegetation sampling plot) would increase with moose winter counts in the surrounding landscape (1 to 3 km radius). We modeled moose counts as a function of environmental features and road density, using a quasi-Poisson regression. According to an aerial survey (25 km2 sampling plots), moose counts increased with the proportion of forage patches (disturbed stands < 20 years old), whereas they decreased with the proportion of mixed mature stands > 80 years old, which offer both forage and cover. We extrapolated this model to a 2770 km2 study area and used predicted counts as an independent variable to assess its influence on browsing pressure. In 720 plots of 4.15 m2, browsing pressure was quantified using occurrence (i.e. proportion of seedlings/saplings with at least one twig browsed during the previous winter) and intensity (i.e. mean proportion of twigs browsed per stem) on commercial and non-commercial tree seedlings and saplings. We explained variation in occurrence and intensity using zero-inflated models with snow precipitation and local vegetation variables as covariates. Browsing pressure increased with predicted moose counts within a 3 km radius, with snow precipitation and with the proportion of coniferous trees in the canopy, whereas it decreased with sapling density. These results highlight the importance of seedlings and saplings in moose winter diet and confirm the attraction of moose toward recent (<20-year-old) harvest blocks during winter. Therefore, we should avoid aggregating harvest blocks to decrease their attractiveness to moose and browsing pressure on natural hardwood regeneration.

PCB defect detection with self-supervised learning of local image patches

This paper presents a defect detection technique in printed circuit boards (PCBs) using self-supervised learning of local image patches (SLLIP). Defect detection in PCBs is often hindered by the problems like a lack of defect data, the existence of tiny components, and the cluttered background. From the observation that some local image patches of a PCB are similar in texture and brightness distribution but are different in semantic features, the proposed self-supervised learning method utilizes the relative position estimation, spatially adjacent similarity, and k-means clustering of patches to learn finely classified semantic features. Then, the feature consistency between the local image patches and the background is learned by a local image patch completion network. The feature differences between the estimated and the original image patches are used to detect anomaly areas in PCBs. Experiment results on the PCB defect dataset demonstrate that the proposed SLLIP outperforms the state-of-the-art methods.

Histopathological bladder cancer gene mutation prediction with hierarchical deep multiple-instance learning.

Gene mutation detection is usually carried out by molecular biological methods, which is expensive and has a long-time cycle. In contrast, pathological images are ubiquitous. If clinically significant gene mutations can be predicted only through pathological images, it will greatly promote the widespread use of gene mutation detection in clinical practice. However, current gene mutation prediction methods based on pathological images are ineffective because of the inability to identify mutated regions in gigapixel Whole Slide Image (WSI). To address this challenge, hereby we propose a carefully designed framework for WSI-based gene mutation prediction, which consists of three parts. (i) The first part of cancerous area segmentation, based on supervised learning, quickly filters out a large number of non-mutated regions; (ii) the second part of cancerous patch clustering, based on the representations derived from contrastive learning, ensures the comprehensiveness of patch selection; and (iii) the third part of mutation classification, based on the proposed hierarchical deep multi-instance learning method (HDMIL), ensures that sufficient patches are considered and inaccurate selections are ignored. In addition, benefiting from a two-stage attention mechanism in HDMIL, the patches that are highly correlated with gene mutations can be identified. This interpretability can help a pathologist to analyze the correlation between gene mutation and histopathological morphology. Experimental results demonstrate that the proposed gene mutation prediction framework significantly outperforms the state-of-the-art methods. In the TCGA bladder cancer dataset, five clinically relevant gene mutations are well predicted.

Part-to-Surface Mesh Segmentation for Mechanical Models Based on Multi-Stage Clustering

In this paper, we propose a part-to-surface mesh segmentation method for engineering models characterized by a multi-stage clustering of surface patches generated by the Variational Shape Approximation (VSA) method. Initially, border types between topologically adjacent surface patches are estimated and surface patches bonded by certain border types are grouped into clusters. Secondly, clusters of surface patches are iteratively aggregated into volumetric part-level sub-meshes according to the convexity variations of intermediate sub-meshes. At last, surface patches within their respective part-level sub-meshes are merged into surface-level sub-meshes that are well-fitted by basic primitive shapes based on an improved surface-aware similarity metric. Experimental results demonstrate that the proposed method can achieve better results on mechanical models both quantitatively and perceptually, especially for objects with non-zero genus topology and rich in tiny features.

Biologically Important Areas for bowhead whales (Balaena mysticetus): Optimal site selection with integer programming

Place-based approaches to marine conservation identify areas that are crucial to the success of populations, species, communities, or ecosystems, and that may be candidates for special management actions. In the United States, the National Oceanic and Atmospheric Administration defined Biologically Important Areas (BIAs) for cetaceans (whales, dolphins, and porpoises) as areas and periods that individual populations or species are known to preferentially use for certain activities or where small resident populations occur. The activities considered to be biologically important are feeding, migrating, and activities associated with reproduction. We present an approach using spatial optimization to refine the BIA delineation process to be more objective and reproducible for conservation planners and decision makers who wish to use various spatial criteria to address conservation or management objectives. We present a case study concerning feeding bowhead whales (Balaena mysticetus) and bowhead whale calves in the western Beaufort Sea to illustrate the mechanics and benefits of our optimization model. In the case study, we incorporate spatial information about whales’ relative density and optimally delineate BIAs under different thresholds for minimum patch (cluster) size and total area encompassed within the BIA network. Results from our case study showed three consistent patterns related to minimum cluster size (contiguity) and maximum area threshold for both BIA types and all months: (1) cells with the highest whale density were selected when contiguity or maximum area thresholds were small; (2) for a given area threshold, the number of whales inside BIAs was inversely proportional to cluster size; and (3) the number of whales inside BIAs initially increased rapidly as the area threshold increased, but eventually approached an asymptote. Additionally, information on temporal variability in a BIA may influence the development of conservation, management, monitoring, or mitigation methods. To provide additional insight into the ecological characteristics of the BIAs selected during the optimization step, we quantified inter-annual variability in whale occurrence and density within individual BIAs using statistical techniques. The bowhead whale BIAs and associated information that we present can be incorporated with other relevant information (e.g., objectives, stressors, costs, acceptable risk, legal constraints) into conservation and management decision-making processes.

Using Marchenko–Pastur SVD and Linear MMSE Estimation for Reducing Image Noise

The degradation in visual quality of images is often seen due to a variety of noise added inevitably at the time of image acquisition. Its restoration has thus become a fundamental and significant problem in image processing. Many attempts are made in recent past to efficiently denoise images. But, the best possible solution to this problem is still an open research problem. This paper validates the effectiveness of one such popular image denoising approach, where an adaptive image patch clustering is followed by the two step denoising algorithm in Principal Component Analysis (PCA) domain. First step uses Marchenko–Pastur law based hard thresholding of singular values in the singular value decomposition (SVD) domain and the second step removes remaining noise in PCA domain using Linear Minimum Mean-Squared-Error (LMMSE), a soft thresholding. The experimentation is conducted on gray-scale images corrupted by four different noise types namely speckle, salt &amp; pepper, Gaussian, and Poisson. The efficiency of image denoising is quantified in terms of popular image quality metrics peak signal-to-noise ratio (PSNR), structural similarity (SSIM), feature similarity (FSIM), and the denoising time. The comprehensive performance analysis of the denoising approach against the four noise models underlies its suitability to various applications. This certainly gives the new researchers a direction for selection of image denoising method.

Research on weed identification method in rice fields based on UAV remote sensing.

Rice is the world's most important food crop and is of great importance to ensure world food security. In the rice cultivation process, weeds are a key factor that affects rice production. Weeds in the field compete with rice for sunlight, water, nutrients, and other resources, thus affecting the quality and yield of rice. The chemical treatment of weeds in rice fields using herbicides suffers from the problem of sloppy herbicide application methods. In most cases, farmers do not consider the distribution of weeds in paddy fields, but use uniform doses for uniform spraying of the whole field. Excessive use of herbicides not only pollutes the environment and causes soil and water pollution, but also leaves residues of herbicides on the crop, affecting the quality of rice. In this study, we created a weed identification index based on UAV multispectral images and constructed the WDVI NIR vegetation index from the reflectance of three bands, RE, G, and NIR. WDVI NIR was compared with five traditional vegetation indices, NDVI, LCI, NDRE, and OSAVI, and the results showed that WDVI NIR was the most effective for weed identification and could clearly distinguish weeds from rice, water cotton, and soil. The weed identification method based on WDVI NIR was constructed, and the weed index identification results were subjected to small patch removal and clustering processing operations to produce weed identification vector results. The results of the weed identification vector were verified using the confusion matrix accuracy verification method and the results showed that the weed identification accuracy could reach 93.47%, and the Kappa coefficient was 0.859. This study provides a new method for weed identification in rice fields.

Mapping, quantifying and comparing seascape heterogeneity of Southwest Indian Ridge seamounts

ContextSeamounts are abundant geomorphological features creating seabed spatial heterogeneity, a main driver of deep-sea biodiversity. Despite its ecological importance, substantial knowledge gaps exist on the character of seamount spatial heterogeneity.ObjectivesThis study aimed to map, quantify and compare seamount seascapes to test whether individual habitats and seamounts differ in geomorphological structuring, and to identify spatial pattern metrics useful to discriminate between habitats and seamounts.MethodsWe mapped and classified geomorphological habitat using bathymetric data collected at five Southwest Indian Ridge seamounts. Spatial pattern metrics from landscape ecology are applied to quantify and compare seascape heterogeneity in composition and configuration represented in resulting habitat maps.ResultsWhilst part of the same regional geological feature, seamounts differed in seascape composition and configuration. Five geomorphological habitat types occurred across sites, which within seamounts differed in patch area, shape and clustering, with ridge habitat most dissimilar. Across seamounts, the spatial distribution of patches differed in number, shape, habitat aggregation and intermixing, and outcomes were used to score seamounts on a gradient from low to high spatial heterogeneity.ConclusionsAlthough seamounts have been conceptualised as similar habitats, this study revealed quantitative differences in seascape spatial heterogeneity. As variations in relative proportion and spatial relationships of habitats within seamounts may influence ecological functioning, the proposed quantitative approach can generate insights into within-seamount characteristics and seamount types relevant for habitat mappers and marine managers focusing on representational ecosystem-based management of seamounts. Further research into associations of sessile and mobile seamount biodiversity with seascape composition and configuration at relevant spatial scales will help improve ecological interpretation of metrics, as will incorporating oceanographic parameters.

Linking Earthquake Magnitude‐Frequency Statistics and Stress in Visco‐Frictional Fault Zone Models

AbstractThe ability to estimate the likelihood of given earthquake magnitudes is critical for seismic hazard assessment. Earthquake magnitude‐recurrence statistics are empirically linked to stress, yet which fault‐zone processes explain this link remains debated. We use numerical models to reproduce the interplay between viscous creep and frictional sliding of a fault‐zone, for which inter‐seismic locking becomes linked to stress. The models reproduce the empirical stress‐dependent earthquake magnitude distribution observed in nature. Stress is related to the likelihood a fault section is near frictional failure, influencing likely rupture lengths. An analytical model is derived of a fault consisting of identical patches, each with a probability of inter‐seismic locking. It reproduces a similar magnitude‐recurrence relationship, which may therefore be caused by probabilistic clustering of locked fault patches. Contrasts in earthquake statistics between regions could therefore be explained by stress variation, which has future potential to further constrain statistical models of regional seismicity.

Deep learning applied to analyze patterns from evaporated droplets of Viscum album extracts

This paper introduces a deep learning based methodology for analyzing the self-assembled, fractal-like structures formed in evaporated droplets. To this end, an extensive image database of such structures of the plant extract Viscum album Quercus10^{-3} was used, prepared by three different mixing procedures (turbulent, laminar, and diffusion based). The proposed pattern analysis approach is based on two stages: (1) automatic selection of patches that exhibit rich texture along the database; and (2) clustering of patches in accordance with prevalent texture by means of a Dense Convolutional Neural Network. The fractality of the patterns in each cluster is verified through Local Connected Fractal Dimension histograms. Experiments with Gray-Level Co-Occurrence matrices are performed to determine the benefit of the proposed approach in comparison with well established image analysis techniques. For the investigated plant extract, significant differences were found between the production modalities; whereas the patterns obtained by laminar flow showed the highest fractal structure, the patterns obtained by the application of turbulent mixture exhibited the lowest fractality. Our approach is the first to analyze, at the pure image level, the clustering properties of regions of interest within a database of evaporated droplets. This allows a greater description and differentiation of the patterns formed through different mixing procedures.

Automated grading of diabetic retinopathy using CNN with hierarchical clustering of image patches by siamese network.

Diabetic retinopathy (DR) is a progressive vascular complication that affects people who have diabetes. This retinal abnormality can cause irreversible vision loss or permanent blindness; therefore, it is crucial to undergo frequent eye screening for early recognition and treatment. This paper proposes a feature extraction algorithm using discriminative multi-sized patches, based on deep learning convolutional neural network (CNN) for DR grading. This comprehensive algorithm extracts local and global features for efficient decision-making. Each input image is divided into small-sized patches to extract local-level features and then split into clusters or subsets. Hierarchical clustering by Siamese network with pre-trained CNN is proposed in this paper to select clusters with more discriminative patches. The fine-tuned Xception model of CNN is used to extract the global-level features of larger image patches. Local and global features are combined to improve the overall image-wise classification accuracy. The final support vector machine classifier exhibits 96% of classification accuracy with tenfold cross-validation in classifying DR images.

Concurrent optimization of topologies and fiber orientations for laminated composite structures

This paper formulates the concurrent optimization of structural topology and anisotropic materials for laminated composite structures by simultaneously optimizing laminate topology and fiber orientations. In this formulation and model, both constant-stiffness and variable-stiffness designs can be obtained. For constant-stiffness designs, ply orientation optimization is studied, and for variable-stiffness designs, element or patch (cluster) orientation optimization is considered, in which both geometry-based and fiber angle-based methods are developed for patch clustering. Analytical solutions to determine optimal material orientations are derived for patch or ply orientation in addition to element orientation. Moreover, out-of-plane loadings are involved, and hence shear deformations and transverse shear effects are taken into consideration. An extended moving iso-surface threshold (MIST) algorithm is proposed to implement the formulation and optimization model, with which the optimization of topology and element/patch/ply orientation is in a fully-coupled concurrent manner.

Quick recovery of a threatened butterfly in well‐connected patches following an extreme drought

AbstractExtreme weather events are becoming more frequent due to climate change. We therefore need to understand how species respond to these events. In 2018, the worst drought ever recorded hit the island of Gotland in the Baltic Sea, which led to a major decline of the threatened marsh fritillary (Euphydryas aurinia) the succeeding summer.We compared yearly occupancy among 256 habitat patches between 2018 and 2020 and analysed the colonisation–extinction dynamics between 2019 and 2020 in relation to patch area, connectivity, and habitat quality. Moreover, using capture‐mark‐release data in two patch clusters differing in connectivity, we also compared population sizes before and after the drought and analysed ‘population growth rates’. We also compared yearly abundance of host plants (Succisa pratensis) over time.Results show a remarkable recovery of the marsh fritillary. Both patch occupancy in 2020 and the number of individuals in the well‐connected patch cluster were higher than before the drought. In contrast, host plants were fewer and smaller, which taken together suggest that the amount of food resources was roughly half in 2020 compared to the pre‐drought conditions. Moreover, the butterfly population in the less connected patch cluster was eight times smaller compared to the population size before the drought. Local colonisations, extinctions, and population growth rates were explained by connectivity.The ability to quickly recover after extreme droughts is promising in times of climate change. The significance of connectivity for the population dynamics during recovery highlights the importance of maintaining well‐connected patch networks.

Assessing spatial–temporal heterogeneity of China’s landscape fragmentation in 1980–2020

Exploring the spatial–temporal heterogeneity of landscape fragmentation contributes to realizing the dynamic evolution of the landscape pattern and understanding the impacts of physical-geographical environment and human activities on landscape systems. Differing from previous studies that focused on using single landscape pattern metrics to characterize the landscape fragmentation, this study proposed a Landscape Fragmentation Index (LFI) to assess and analyze the spatial–temporal heterogeneity of China’s landscape fragmentation in 1980–2020. The result showed that the landscape pattern metrics characterizing patch size, patch shape, patch density, patch type, patch cluster, and patch distance were equipped with obvious phased change characteristics, and all of them have successively undergone the stable phase, fluctuation phase, and adjustment phase. In 1980–2020, China’s LFI was gradually increased to 0.393 from 0.383, annually rising by 0.07 percent points on average. Also, the annual average growth of urbanization rate and GDP respectively reached 1 percent point and 9.26 percent points. Besides, regarding spatial pattern, the landscape fragmentation showed relatively obvious fluctuation features in Southwest-Northeast (Hu Line) and Northwest-Southeast (Bole-Taipei Line) directions and Hu Line was manifested as the dividing line of landscape fragmentation while Bole-Taipei Line was shown as the equilibrium line. Spatial-temporal heterogeneity of China’s landscape fragmentation indicated that the fragmented landscape pattern was the combined effect of the physical-geographical environment and human activity interference.

Dichotomous classification and implications in spatial planning: A case of the Rural-Urban Continuum settlements of Kerala, India

Kerala State in India has a unique Rural-Urban Continuum (RUC) settlement pattern where it is difficult to distinguish between urban from rural. However, like all the Indian States, the RUC settlements of Kerala are also divided into rural and urban, and this dichotomous classification forms the basis of spatial planning, governance, and management. The current situation has resulted in the spread of urban characterized settlements towards the environmentally fragile areas of the state. Despite several discussions regarding the RUC nature of settlements, details about the spatial characteristics of Kerala are missing in the literature. Accordingly, the paper explores the RUC settlement pattern of Kerala in two parts. The first part assesses the RUC pattern based on the existing Indian census definition. The result reveals that the urban and rural definitions do not hold validity in Kerala. The second part explores the settlements based on the topographic distribution, followed by a detailed analysis of the spatiotemporal dynamics of the built areas in three levels of detailing. The study reveals a spread of built-up areas across diverse topography and variation among the built-up areas of different urban areas. While the lowland regions indicated a dominance and clustering of built-up patches, in the midlands and towards the highland study areas, the built-up areas are smaller and more fragmented with an affinity towards the transportation corridors. Therefore the study helped characterize the spread of reclassified settlements and the changes in built-up areas across diverse topography and emphasized the requirement to move away from dichotomous classification as followed in some developed countries. The study recommends an RUC code for Kerala and an Eco-sensitive Regional Planning approach for a better spatial planning process. A modified and refined planning framework is also proposed as a final output from the research.