Spatio-temporal Change Detection Research Articles

Monitoring yellow rust progression during spring critical wheat growth periods using multi-temporal Sentinel-2 imagery.

Yellow rust (Puccinia striiformis f. sp. tritici) is a devastating hazard to wheat production, which poses a serious threat to yield and food security in the main wheat-producing areas in eastern China. It is necessary to monitor yellow rust progression during spring critical wheat growth periods to support its prediction by providing timely calibrations for disease prediction models and timely green prevention and control. Three Sentinel-2 images for the disease during the three wheat growth periods (jointing, heading, and filling) were acquired. Spectral, texture, and color features were all extracted for each growth period disease. Then three period-specific feature sets were obtained. Given the differences in field disease epidemic status in the three periods, three period-targeted monitoring models were established to map yellow rust damage progression in spring and track its spatiotemporal change. The models' performance was then validated based on the disease field truth data during the three periods (87 for the jointing period, 183 for the heading period, and 155 for the filling period). The validation results revealed that the representation of the wheat yellow rust damage progression based on our monitoring model group was realistic and credible. The overall accuracy of the healthy and diseased pixel classification monitoring model at the jointing period reached 87.4%, and the coefficient of determination (R2) of the disease index regression monitoring models at the heading and filling periods was 0.77 (heading period) and 0.76 (filling period). The model-group-result-based spatiotemporal change detection of the yellow rust progression across the entire study area revealed that the area proportions conforming to the expected disease spatiotemporal development pattern during the jointing-to-heading period and the heading-to-filling period reached 98.2% and 84.4% respectively. Our jointing, heading, and filling period-targeted monitoring model group overcomes the limitations of most existing monitoring models only based on single-phase remote sensing information. It performs well in revealing the wheat yellow rust spatiotemporal epidemic in spring, can timely update disease trends to optimize disease management, and provide a basis for disease prediction to timely correct model. © 2024 Society of Chemical Industry.

The effectiveness of natural reserves from the perspective of habitat quality in the southern section of the Hengduan Mountains, Southwestern China

Natural reserves (NRs) play key roles in habitat integrity conservation and biodiversity loss mitigation, and the assessment of the conservation effectiveness of NRs is needed to better manage them. Habitat quality (HQ) comprehensively reflects habitat integrity and biodiversity, but the conservation effectiveness of NRs from the perspective of HQ has rarely been determined at high spatial resolution. Taking the southern section of the Hengduan Mountains (SSHM) in Southwest China as an example, combining an InVEST-HQ model and spatiotemporal change detection methods, the effectiveness of NRs from the perspective of HQ at 30-m spatial resolution was assessed in this study. The effectiveness disparities of NRs across different properties (i.e., management level, conservation target, size, and establishment age) was analyzed and the human pressures on NRs was investigated. The results showed that the HQ of the NRs is good in the SSHM, with the area ratio of the Higher and Highest HQ ≥ 93%. Most of the NR area (94.11%) was effective at improving or maintaining a good HQ. With regard to NR properties, county NRs, NRs designated to conserving wild animals, middle NRs, and younger NRs were more effective, corresponding to management level, conservation target, size, and establishment age, respectively. The human footprint for an effective area is significantly lower than that for an ineffective area, consistent with higher HQ in the effective area and lower HQ in the ineffective area. These findings support the management and zoning of NRs in the SSHM to ensure their effectiveness.

Deep Siamese Network for annual change detection in Beijing using Landsat satellite data

The detection of spatiotemporal changes in land use/land cover (LULC-SC) plays a paramount role in the analysis of smart cities, it can describe complex urban distribution, functions, and patterns. Chinas capital city has been developing rapidly in the past two decades, however, there are only few long-term studies on an annual scale of LULC-SC. To fill this research gap, we propose a remote sensing parallel framework for the detection of LULC-SC based on the combination of the Deep Siamese Network and long time series, which focuses on the spatial semantic information at the object level. A Landsat time series from 2002 to 2022 serves as input satellite data. First, we focus on building graph constructions at the object level and then use an autonomously constructed deeper-feature graph convolutional network (DF-GCN) to mine deeper features, spatial semantic, and relationships at the object level. Finally, the Siamese Network recognizes the changes in the spatial semantic tensors of long time series of Landsat images and quickly maps LULC-SC. The results prove that the proposed spatiotemporal change detection framework is effective in LULC-SC in Beijing. Compared with other networks, the optimal accuracy of semantic mining based on DF-GCN can reach about 90%. Over the past two decades, the LULC-SC of Beijing has changed in a complex way, with urbanization occurring primarily through the replacement of farmland. Consequently, the proposed framework can generate accurate LULC change maps at high temporal frequencies, which can contribute to a better comprehension of sustainable urban development and planning.

Increasing seasonal variation in the extent of rivers and lakes from 1984 to 2022

Abstract. Knowledge of the spatial and temporal distribution of surface water is important for water resource management, flood risk assessment, monitoring ecosystem health, constraining estimates of biogeochemical cycles and understanding our climate. While global-scale spatiotemporal change detection of surface water has significantly improved in recent years due to planetary-scale remote sensing and computing, it has remained challenging to distinguish the changing characteristics of rivers and lakes. Here we analyze the spatial extent of permanent and seasonal rivers and lakes globally over the past 38 years based on new data of river system extents and surface water trends. Results show that while the total permanent surface area of both rivers and lakes has remained relatively constant, the areas with intermittent seasonal coverage have increased by 12 % and 27 % for rivers and lakes, respectively. The increase is statistically significant in over 84 % of global water catchments based on Spearman's rank correlations (rho) above 0.05 and p values less than 0.05. The seasonal river extent is nearly 32 % larger than the previously observed annual mean river extent, suggesting large seasonal variations that impact not only ecosystem health but also estimations of terrestrial biogeochemical cycles of carbon. The outcomes of our analysis are shared as the Surface Area of Rivers and Lakes (SARL) database, serving as a valuable resource for monitoring and research of hydrological cycles, ecosystem accounting, and water management.

Spatio temporal change detection analysis of land use/land cover (LULC) using geospatial tools in the Kuzhithuraiyar sub-basin, Kodayar River, Kanniyakumari district, Tamilnadu, India


 
 
 Land use land cover (LULC) has altered dramatically because of anthropogenic activities, particularly in places where climate change and population growth are severe. The geographic information system (GIS), Global Positioning System (GPS) and Remote sensing (RS) are widely used techniques for monitoring LULC changes. This study aimed to assess the LULC changes in Kuzhithuraiyar sub basin of Kanyakumari district, Tamil Nadu, India. The satellite images of four decades, that is 1992, 2002, 2012 and 2022, are collected and worked with the help of ArcGIS and ENVI Software. The findings have been validated in the field using GPS. The sub basin is classified based on Maximum Likelihood Classification (MLC) into nine classes. They are (i) Barren land, (ii) Built-up land, (iii) Crop land, (iv) Dense forest, (v) Fallow land, (vi) Grass land, (vii) Open forest, (viii) Shrub land and (ix) Water body and the gain and loss of each classes have been mapped. Land use/land cover change has occurred due to the direct and indirect involvement of population growth. Land use changes beyond the land’s capability played a significant role in triggering land degradation. To minimize these adverse consequences of land use change, environmentally friendly management measures must be implemented. Proper land use planning is essential for a sustainable development of the study area. Hence, this study will help in future sustainable planning and management.
 
 

A novel framework for fine-grained spatio-temporal change detection in satellite images

A novel framework for fine-grained spatio-temporal change detection in satellite images

Spatio-Temporal Change Detection Analysis of Land Use Land Cover of Bathinda District, Punjab, India

ABSTRACT: Due to rapid industrialization and urban sprawl in the last few decades, the land use pattern and its consumption takes place at a large scale that could lead to problems such as over-exploitation of land resources, food insecurity and pollution. It becomes imperative to carry out monitoring and subsequent modelling of land use land cover (LULC) changes. An attempt was made to study the changes in the LULC pattern of district of Bathinda, Punjab, India. Remote sensing (RS) and geographical information system (GIS) were used to perform the analysis of satellite data using image processing and classification procedures. For preparing LULC maps, supervised classification was carried out using maximum likelihood classification (MLC) algorithm, aided with Earth Resources Data Analysis System (ERDAS) Imagine 2014 and ArcGIS 10.3 software. Further, change detection study was done using multi-temporal Linear Imaging Self Scanning Sensor-III (LISS-III) data sets of the year 2006 and 2018 to analyze the temporal changes. It was observed that the region is occupied by various ground features such as water, built-up area, agricultural land, vegetation/trees and fallow land. The results revealed that the area under water bodies have increased by 0.413km2 in 2018. The built-up areas including human settlements, commercial infrastructures, roads and other pavements, have increased from 584.448km2 to 852.140km2 between 2006 and 2018, whereas the agricultural land has reduced from 2686.121km2 to 2398.384km2 during the period. The area under vegetation (trees) indicated that there was an increasing trend from 28.490km2 to 54.678km2 during 12years of time span whereas, the fallow land/barren land showed a decreasing trend from 26.361km2 to 18.367km2. It is suggested that the LULC change detection studies are very significant to conserve the land resources and to avoid further degradation.

Spatiotemporal Change Detection Analysis of Land Cover and Land Use using Multi-Temporal Remote Sensed Imagery and Landscape Metrics: A Sub-Catchment Level Case Study of the Bhagirathi-Hugli River, West Bengal, India

To analyse and characterise land use-land cover (LULC) change, remote sensing (RS) data and geographic information system (GIS) approaches are required. The landscape metrics are also extensively used to explain the structure and pattern of the spatial landscape. This case study has been conducted in the area of the Bhagirathi-Hugli River banks in West Bengal. The study area’s satellite image data from various periods (1989-2017) were used to analyse the spatiotemporal variation of LULC change. The LULC fragmentation was analysed using a total of six landscape measures. Landscape metrics such as Number of Patches (NP), Largest Patch Index (LPI), Landscape Shape Index (LSI), Mean Shape Index (MSI), Contagion (CONTAG), and Shannon’s Diversity Index were used to determine the changes over four different periods. The main features of land use change were the continuous expansion of built-up area (9.30 km2), vegetation (189.14 km2), and the decrease in agricultural fallow land and water bodies from 1989 to 2017. The landscape pattern in the study area has also changed significantly. The landscape metrics indicate that the studied region had a highly fragmented landscape during the study period. The proposed technique is easy to use and can address more than one metric while selecting the appropriate setting can prove problematic.

Spatio-temporal trend and change detection of rainfall for Kosi River basin, Uttarakhand using long-term (115 years) gridded data

Spatio-temporal trend and change detection of rainfall for Kosi River basin, Uttarakhand using long-term (115 years) gridded data

Spatiotemporal change detection of land use land cover (LULC) in Fashiakhali wildlife sanctuary (FKWS) impact area, Bangladesh, employing multispectral images and GIS

Land cover change has posed significant concerns to biodiversity and climate change in Bangladesh and globally. Despite the country’s designation of forest regions as protected areas to conserve their valuable resources, deforestation and forest conversion remained unabated. Fashiakhali Wildlife Sanctuary (FKWS), a protected area in the Chittagong Hill Tracts, and its surrounding forested impact area have experienced considerable changes over the years, yet are deficient in extensive assessment. This study evaluated the land use land cover (LULC) changes in the FKWS impact area over almost 3 decades (1994–2021) using multispectral remotely sensed data. The Landsat images of 1994, 2001, 2010, and 2021 were classified using a maximum likelihood algorithm and analyzed for change detection. The comparative potential of vegetation indices, including Normalized Difference Vegetation Index (NDVI) and Soil Adjusted Vegetation Index (SAVI), in forest cover assessment, and the relationship between Land Surface Temperature (LST) and NDVI was also assessed. A significant forest cover loss of around 1117.17 ha (16%) was recorded in the FKWS impact area between 1994 and 2021, with the hugest proportion of 867.78 ha (12.24%) deforested in the first period (1994–2001). Agricultural land also declined by 593.73 ha (8.37%) within the entire period, despite its initial increase of 392.04 ha (5.53%) between 2001 and 2010, being the primary driver of earlier deforestation. However, in the recent decade (2010–2021), settlement expansion of 963.90 ha (13.59%) due to massive human migration in the area contributed to the most remarkable overall land cover change of 1731.51 ha (24.42%). Furthermore, NDVI provided a better and more accurate forest cover assessment than SAVI and was recommended to aid in the quick evaluation and monitoring of the future impacts of agriculture, settlement, and other sorts of land use on the forest cover. In tandem with the widely acknowledged issue of increased temperature due to climate change, an absolute negative correlation was found between the NDVI and LST, confirming the negative impact of climate change on forest loss in the FKWS impact area.

The current land use dynamics are dependent on the previous land conversion legacies in farming system of west oromia, Ethiopia

The land use analyses most often focused on the spatio-temporal change detection and how the current conversion status is associated with the previous legacies of land use dynamics are little understood. Here, we examined how the forest land use has been converted to other land use types, and if the intermittent and the current extent of the dynamics are associated with the previous extent of land conversions along spatio-temporal scale. For this, we performed the land use change detection with respective to simplified, moderate and complex landscapes forest cover gradients. These analyses were carried out for the periods of forty-five years (1973–2018) by employing supervised classification method. Our results showed that the forestland has been converted to other land use types at the annual rates of -1.23, -1.34 and -0.81 in simplified, moderate and complex landscapes, respectively. Especially, the rate of forest conversion to crop land has been higher in simplified and moderate landscapes when compared with that of the complex landscapes. These results show that the current statuses of land conversions are largely dependent on the landscape complexity gradients occurred as the result of land use legacies (varying intensity of land management regime in the past). Altogether, a landscape complexity per se is not only a guarantee, and the previous land conversion legacies and the current land use dynamics as well need to be considered in developing landscape management strategies.

Assessment of Iran’s Mangrove Forest Dynamics (1990–2020) Using Landsat Time Series

Mangrove forests distributed along the coast of southern Iran are an important resource and a vital habitat for species communities and the local people. In this study, accurate mapping and spatiotemporal change detection were conducted on Iran’s mangroves for three decades, using the Landsat imagery available for the years 1990, 2000, 2010, and 2020. Four general vegetation indices and eight mangrove-specific indices were employed for mangrove mapping in three study sites. Additionally, six important landscape metrics were implemented to quantify the spatiotemporal alteration of the mangrove forests during the study period. Our results showed the robustness of the submerged mangrove recognition index (SMRI), validated as the most effective index (F1-score ≥ 0.89), which was used for mangrove identification within all nine sites. The mangrove area of southern Iran was estimated at approximately 13,000 ha in 2020, with an overall increase of 2313 ha over the whole period. A similar trend could be observed for both the landscape connectivity and complexity. Our results revealed that a stronger connectivity and higher complexity could be detected in most sites, while there was increased fragmentation and a weaker connection in some locations. This study provides an accurate map of Iran’s mangrove forests over time and space.

Spatio-temporal trend and change detection of temperature and precipitation of Kashafroud basin

Spatio-temporal trend and change detection of temperature and precipitation of Kashafroud basin

Spatio-temporal landslide inventory and susceptibility assessment using Sentinel-2 in the Himalayan mountainous region of Pakistan.

The 2005 Kashmir earthquake has triggered widespread landslides in the Himalayan mountains in northern Pakistan and surrounding areas, some of which are active and are still posing a significant risk. Landslides triggered by the 2005 Kashmir earthquake are extensively studied; nevertheless, spatio-temporal landslide susceptibility assessment is lacking. This can be partially attributed to the limited availability of high temporal resolution remote sensing data. We present a semi-automated technique to use the Sentinel-2 MSI data for co-seismic landslide detection, landslide activities monitoring, spatio-temporal change detection, and spatio-temporal susceptibility mapping. Time series Sentinel-2 MSI images for the period of 2016-2021 and ALOS PALSAR DEM are used for semi-automated landslide inventory map development and temporal change analysis. Spectral information combined with topographical, contextual, textural, and morphological characteristics of the landslide in Sentinel-2 images is applied for landslide detection. Subsequently, spatio-temporal landslide susceptibility maps are developed utilizing the weight of evidence statistical modeling with seven causative factors, i.e., elevation, slope, geology, aspect, distance to fault, distance to roads, and distance to streams. The results reveal that landslide occurrence increased from 2016 to 2021 and that the coverage of areas of relatively high susceptibility has increased in the study area.

Spatio-temporal change detection of land use and land cover in Malakand Division Khyber Pakhtunkhwa, Pakistan, using remote sensing and geographic information system.

The land use land cover (LULC) change due to the rapidly growing population is a common feature of the urban area. The rapidly growing population in Malakand Division is a greater threat to the LULC of the area due to its negative impact on environment and ecology. This research aims to detect the variations in LULC from 1991 to 2017 in the Malakand Division, Khyber Pakhtunkhwa (KP) province of Pakistan. The study relies on secondary dataset downloaded from the US Geological Survey (1991, 2001, 2011, and 2017 imageries) and the United Nations Office for the Coordination of Humanitarian Affairs (UN OCHA) website. Maximum likelihood technique under supervised image classification was opted to analyze the LULC changes in between 1991 and 2017. The results were based on six major land use classes including agriculture built-up area, vegetation cover, water bodies, snow cover, and barren land. The results from 1991 to 2017 show a substantial reduction in snow cover and barren land which is consequence of climate change. A known change has been recorded in built-up area which shows an increase from 1.02 to 6.2% with a change of 5.18% of the total land. The vegetation cover water bodies were also showing increase in area. The vegetation cover increased from 28.89 to 44.67% while barren land decreased from 45.68 to 40.29% of the total area. Furthermore, the built-up area increased from 1.02 to 6.2%, whereas water covers increased from 0.63% (1991) to 0.86% (2017) of the total area. The study concludes that there is an immense need for planning to preserve the natural habitat for sustainable development in the area.

Spatio-Temporal Change Detection of Built-up Areas in Ilorin Metropolis and Implications for Green Space Conservation

Urbanization is a leading driver of green space depletion, and accurate land cover mapping is an essential, cost-effective technique for monitoring this phenomenon. In this study, we analyzed the spatio-temporal changes (1986 to 2020) of built-up areas in the city of Ilorin- Nigeria, and discussed the findings’ implications for green spaces. A supervised pixel-based SVM algorithm was adopted for image classification to produce the land cover maps with multi-temporal Landsat images. The analyzed post-classification result yielded an overall accuracy (OA) of 99.87 percent and a Kappa coefficient (KA) of 0.998. The change metrics showed that built-up areas witnessed a decadal increase in spatial extent from 1986 to 2020. Whereas, vegetation land cover witnessed a decadal decline during these periods. The net-growth rate (%) for built-up areas and vegetation land cover between 1986 to 1996, 1996 to 2006 and 2016 to 2020 were +11.67, +5.04, +6.647, +0.06, and −14.1, −36.52, +25.81, −0.10 respectively. Meanwhile, the trend forecast predicted a decline in vegetation of 9510 hectares by the year 2030 if the current land use pattern continues unabated. The land consumption ratio (LCR) and land absorption coefficient (LAC) results were both positive, indicating a rise in the urban population correlating with larger consumption of new land areas over time.

Change detection of Jamuna River and its impact on the local settlements

ABSTRACT Spatio-temporal change detection analysis of Jamuna River using low and medium-resolution satellite data between the years of 1972 and 2013 are dynamic to assess river spatio-temporal changes and their impacts on the local settlements over the specific time. This study employs remote sensing and geographic information system (GIS) proficiencies to identify the river bank shifting, erosion, and depositional features along with the population displacement. The satellite image of LANDSAT multispectral (MSS), thematic mapper (TM), and operational land imager (OLI) and thermal infrared sensor (TIRS) were used in this study. Supervised and unsupervised classification techniques have been used to identify land cover types and detect the changes using ArcGIS Pro 2.3.2. Supervised classifications performed well by validating ≥90% overall accuracy. The result shows total erosion was 3356 ha from 1972 to 2013, while the deposition was 5342 ha. Erosion and deposition cause the bank line shifting, river widening, and settlement displacement. Hundreds of displaced people suffered from loss of land, livelihood, increased poverty, food insecurity, lack of sanitation, and drinking water facilities. Thus, this study will be earnest in assisting the environmental management and associated planning including necessary measures.

ANALYSIS OF SPATIO-TEMPORAL URBAN DYNAMICS IN 11 SMART CITIES OF UTTAR PRADESH, INDIA

Abstract. Urban planning in smart cities needs to be done in a “Smart” way. One way is to analyze the urbanisation pattern by spatio-temporal change detection techniques. Classified data such as, for years 1985, 1995 and 2005 Decadal Land use data for India and for year 2015, Copernicus Global Land service Dynamic Land Cover layers (CGLS-LC100 products) are used to perform multi-temporal analysis of the 11 smart cities of Uttar Pradesh state of India namely "Agra", "Aligarh", "Bareilly", "Jhansi", "Kanpur", "Lucknow", "Moradabad", "Prayagraj", "Rampur", "Saharanpur" and "Varanasi". Dynamics of Urban expansion are studied utilizing concepts of Landscape Metrics calculated by FRAGSTATS and also Shannon’s Entropy Values (Hn) over the 11 smart cities. Largest Patch Index (LPI), Landscape Shape Index (LSI), Aggregation Index (AI) and Mean Euclidean Nearest Neighbor Distance (ENN_MN) are metrics used to characterize urbanisation. Results indicate rise in value of LSI over the years from 1985 and with sudden increase in year 2015 for Built-up patches, corroborating more complexity in shapes of Built-up patches in all 11 cities. Kanpur, showing large values of LPI indicates the sudden increase of Built-up land use class over the years. The decreasing value of ENN_MN over the years indicates less centrality for built-up pixels in urbanisation. AI is unchanged for Built-up patches for 1985–1995 but decrease in year 2015 indicates less compactness which is due to dispersion of built-up pixels. High values of Hn over the years indicating dispersion of urbanisation in all 11 smart cities except Agra, also validates results.

Detecting spatio-temporal urban surface changes using identified temporary coherent scatterers

Synthetic aperture radar (SAR) is able to detect surface changes in urban areas with a short revisit time, showing its capability in disaster assessment and urbanization monitoring. Most presented change detection methods are conducted using couples of SAR amplitude images. However, a prior date of surface change is required to select a feasible image pair. We propose an automatic spatio-temporal change detection method by identifying the temporary coherent scatterers. Based on amplitude time series, x <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> -test and iterative single pixel change detection are proposed to identify all step-times: the moments of the surface change. Then the parameters, e.g., deformation velocity and relative height, are estimated and corresponding coherent periods are identified by using interferometric phase time series. With identified temporary coherent scatterers, different types of temporal surface changes can be classified using the location of the coherent periods and spatial significant changes are identified combining point density and F values. The main advantage of our method is automatically detecting spatio-temporal surface changes without prior information. Experimental results by the proposed method show that both appearing and disappearing buildings with their step-times are successfully identified and results by ascending and descending SAR images show a good agreement.

Spatiotemporal change detection of carbon storage and sequestration in an arid ecosystem by integrating Google Earth Engine and InVEST (the Jiroft plain, Iran)

Our study uses regional-scale maps to quantify carbon storage and sequestration from different land use types to evaluate the effects of future land use scenarios. We developed an integrated modeling approach to assess the spatiotemporal impacts of land use/cover change (LUCC) on the provision and value of the carbon storage and sequestration during the historical period (2000–2019) and predicted scenarios (2019–2046) in the Jiroft plain, Iran. We integrated several analytic tools for our analysis, which was comprised of Google Earth Engine (GEE), Cellular Automata Markov Chain (CA-MC) model, Intensity Analysis (IAA), and the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model. Our results demonstrate that: (1) agriculture and urban expansion led to a considerable decrease in carbon storage, mainly due to rapid deforestation from 2000–2019; (2) if the historical trend continues under the business as usual (BAU) scenario, it will lead to considerable social costs due to the loss of stored carbon in the plain (2,624,113 Mg) with an annual average sequestration loss of −475,547 Mg; (3) the downward carbon sequestration trend could potentially be reversed by employing the environmentally sound planning (ESP) scenario that is estimated to save 3,705,491 Mg in carbon storage, with annual average sequestration gain of + 605,830 Mg. The design scenarios provide a useful guide for policymakers and local governments to help understand the potential outcomes of the various development strategies, which will ultimately lead to more effective ecosystem management.