Thermal footprint of the urbanization process: analyzing the heat effects of the Urbanization Index (UI) on the Local Climate Zone (LCZ) and Land Surface Temperature (LST) over two decades in Seville
Abstract
Urbanization is a multifaceted process characterized by changes in urban areas through various means, such as sprawl, ribbon development, or infill and compact growth. This phenomenon changes the pattern of the local climate zone (LCZ) and significantly affects the climate, vegetation dynamics, energy consumption, water resources, and public health. This study aims to discern the impacts of changes in urban growth on the LCZ and land surface temperature (LST) over a two-decade period. A comprehensive methodology that integrates statistical analysis, data visualization, machine learning, and advanced techniques, such as remote sensing technology and geospatial analysis systems, is employed. ENVI, GEE, and GIS tools are utilized to collect, process, and monitor satellite data and imagery of temporal and spatial variations in intensive or diffuse urbanization processes from 2003 to 2023 to analyze and simulate land use and land cover (LULC) changes, urbanization index (UI), LCZ patterns, and LST changes over the years and to make overlapping maps of changes to recognize the relation between LULC, LCZ, and LST. This study focuses on Seville’s urban area, which has experienced rapid urbanization and a significant increase in average temperature during the last few decades. The findings of this study will provide actionable recommendations into the interplay between urban growth and climate and highlight the pivotal role of urban growth in shaping resilience and vulnerable areas based on microclimate changes. Urban planners can leverage these insights to predict alternatives for the future development of urban areas and define practical climate mitigation strategies.
Full text
Citation: Falah, N.; Solis-Guzman, J.; Falah, N. Thermal Footprint of the Urbanization Process: Analyzing the Heat Effects of the Urbanization Index (UI) on the Local Climate Zone (LCZ) and Land Surface Temperature (LST) over Two Decades in Seville. Land 2024,13, 1877. https://doi.org/ 10.3390/land13111877 Academic Editors: Zongcai Wei, Yuting Liu and Andrzej Biłozor Received: 11 October 2024 Revised: 5 November 2024 Accepted: 6 November 2024 Published: 10 November 2024 Copyright: © 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). Article Thermal Footprint of the Urbanization Process: Analyzing the Heat Effects of the Urbanization Index (UI) on the Local Climate Zone (LCZ) and Land Surface Temperature (LST) over Two Decades in Seville Nadia Falah 1, Jaime Solis-Guzman 1,* and Nahid Falah 2 1 ArDiTec Research Group, Department of Architectural Constructions II, Higher Technical School of Building Engineering, Universidad de Sevilla, Av. Reina Mercedes 4-a, 41012 Seville, Spain; [email protected] 2Landkreis Harburg, Schloßplatz 6, 21423 Winsen (Luhe), Germany; [email protected] *Correspondence: [email protected] Abstract: Urbanization is a multifaceted process characterized by changes in urban areas through various means, such as sprawl, ribbon development, or infill and compact growth. This phenomenon changes the pattern of the local climate zone (LCZ) and significantly affects the climate, vegetation dynamics, energy consumption, water resources, and public health. This study aims to discern the impacts of changes in urban growth on the LCZ and land surface temperature (LST) over a two-decade period. A comprehensive methodology that integrates statistical analysis, data visualization, machine learning, and advanced techniques, such as remote sensing technology and geospatial analysis systems, is employed. ENVI, GEE, and GIS tools are utilized to collect, process, and monitor satellite data and imagery of temporal and spatial variations in intensive or diffuse urbanization processes from 2003 to 2023 to analyze and simulate land use and land cover (LULC) changes, urbanization index (UI), LCZ patterns, and LST changes over the years and to make overlapping maps of changes to recognize the relation between LULC, LCZ, and LST. This study focuses on Seville’s urban area, which has experienced rapid urbanization and a significant increase in average temperature during the last few decades. The findings of this study will provide actionable recommendations into the interplay between urban growth and climate and highlight the pivotal role of urban growth in shaping resilience and vulnerable areas based on microclimate changes. Urban planners can leverage these insights to predict alternatives for the future development of urban areas and define practical climate mitigation strategies. Keywords: urbanization process; land use and land cover (LULC); local climate zone (LCZ); climate change; land surface temperature (LST); remote sensing (RS); Google earth engine (GEE) 1. Introduction The onset of the 21st century has marked a significant acceleration in urbanization, with more than half of the world’s population now residing in urban areas. This number is anticipated to rise to nearly 60% by 2030 [ 1 , 2 ]. Rapid urban expansion profoundly affects local urban environments and has become a major catalyst for increasing LULC changes. Characterized by long-term shifts in weather patterns, climate change exacerbates temperature instability and disrupts natural cycles [3]. Urban growth and intensification patterns are driven by factors such as population shifts [ 2 ], infrastructure development, industrialization [ 4 – 6 ], historical development, economic growth [ 7 ], and various environmental, geographical, and planning conditions, which cause changes in the multifaceted nature of urban systems. All these elements shape urban morphology and texture and cause changes in LULC’s current status [ 8 – 10 ]. LULC refers to the pattern of biological and physical cover over Land 2024,13, 1877. https://doi.org/10.3390/land13111877 https://www.mdpi.com/journal/land
Land 2024,13, 1877 2 of 27 the land’s surface, including forests, water bodies, agricultural and vegetation lands, and urban areas, as well as the human activities associated with them. These changes in LULC create different types of urban growth with distinct characteristics [ 11 , 12 ], changing the pattern of microclimates over time [ 5 , 6 ]. A microclimate change is described as a long-term change in the average weather patterns that affect Earth’s local climates [1,3] , which is crucial for understanding the impact and role of urban growth patterns on climate, especially temperature variation patterns [ 3 ]. LCZs, as a standardized framework, categorize urban landscapes into different classes based on several factors, such as the physical and thermal properties of urban and natural landscapes [ 13 ]. These classes are influenced by urbanization processes, changes in LULC patterns, and climate change [ 14 , 15 ]. On one hand, the classification of LCZs involves analyzing surface structure, land cover, local climate conditions, function, and anthropogenic influence, with a particular focus on urban and peri-urban areas [ 16 , 17 ]. On the other hand, LST changes are influenced by landscape alterations, vegetation loss, LULC composition, urban materials and geometry, anthropogenic heat, natural settings, and geographic locations, the same as LCZ [ 13 , 16 , 18 ]. In many metropolitan areas, the temperature differential caused by an elevated LST, driven by a high percentage of impermeable surfaces, can escalate to around 5 ◦ C [ 19 – 21 ]. This rise in LST predominantly results from the expansion of impermeable areas, which correlate closely with urban development-induced modifications in LULC. This trend is raising significant environmental and human health concerns [ 22 – 24 ]. Over the past few decades, urban areas have seen surface temperatures rise by approximately 0.87 ◦ C, attributed to global climate change and human activities [ 4 , 19 , 20 ]. Predictions indicate the LST in urban areas will continue to increase [ 8 , 14 , 20 ]. Subsequently, evaluating the impact of LULC changes on LST is urgent for designing livable areas and creating healthy urban environments. Table 1provides an overview of previous studies, summarizing key findings related to the effects of urbanization on land cover and temperature. Table 1. Previous studies related to the LULC, LCZs, and LST. Main Point Studies Studies emphasized the correlation between climate change, land alteration, and their consequences on LST outcomes. [15,20,25–28] Studies examined the urban growth and sprawl’s impact on local climate. The effects on LST are crucial for evaluating sustainable urban development and understanding both the quantity and quality of these changes. [3,7,12,24,29–31] Studies explored the relationship between LCZs and LST, demonstrating that different LCZ types exhibit varying thermal behaviors and providing insights into how urban morphology influences local temperatures. [14,16,19,32–34] Studies defined the relationship between LULC changes and LCZ patterns and mentioned that changes in land use and cover directly influence the classification of LCZ and the integration of them allows for a more detailed analysis of urban heat islands (UHIs) by linking specific land cover types with their corresponding thermal effects. [14,21,35–37] According to Table 1, the indicators and lines of these studies include LULC, LCZs, and LST, which are defined as follows. Main types of urban growth include the following: • Compact growth or infill development: concentrating buildings within existing developed areas [8], focusing on vertical development. • Ribbon development and infrastructure-led growth: occurring alongside transportation and communication routes [5], forming linear patterns of development. • Sprawl or dispersed growth: the outward expansion of urban boundaries, often linked with suburban or exurban areas [12].
Land 2024,13, 1877 3 of 27 • Mixed development: involves varied types of development occurring over time within different parts of an urban area, reflecting a city area faced with multiple types of growth and development. • LULC change: identified various land use types, such as built-up areas (BUAs), water bodies (WBs), agricultural lands, vegetation, forest lands (A/V/F), and barren soil (BS) [ 4 , 5 , 12 ]. LULC change detection entails the comparison of images taken at various times to detect and study alterations in the landscape and to observe urban growth and deforestation [12,38]. • LCZ: This classification system, introduced by Stewart and Oke [ 14 ], categorizes urban and rural areas based on surface cover, structure, and thermal properties, providing a standardized framework for comparing different urban areas. This system is widely used to study how various urbanization forms influence microclimates, particularly in relation to LST. Different physical classifications of LCZ are explained in the methodology section. • LST: A critical factor in comprehending the thermal atmosphere and Earth’s surface in urban areas [ 16 , 39 ] is associated with different patterns of urban growth. Many factors play crucial roles in the level of LST, such as solar radiation, vegetation cover, LULC, urban growth, topography, weather and climate condition, and human activities [19–21] . LST is a critical parameter in studying Earth’s energy balance, urban heat islands, climate change, and environmental processes. Due to the interplay and connection between urbanization, LULC, LCZ, and LST changes (Figure 1), the main points mentioned based on the literature review are included: Land2024,13,18773of29 Ribbondevelopmentandinfrastructure‐ledgrowth:occurringalongsidetransportationandcommunicationroutes[5],forminglinearpatternsofdevelopment. Sprawlordispersedgrowth:theoutwardexpansionofurbanboundaries,often linkedwithsuburbanorexurbanareas[12]. Mixeddevelopment:involvesvariedtypesofdevelopmentoccurringovertime withindifferentpartsofanurbanarea,reflectingacityareafacedwithmultipletypes ofgrowthanddevelopment. LULCchange:identifiedvariouslandusetypes,suchasbuilt‐upareas(BUAs),water bodies(WBs),agriculturallands,vegetation,forestlands(A/V/F),andbarrensoil(BS) [4,5,12].LULCchangedetectionentailsthecomparisonofimagestakenatvarious timestodetectandstudyalterationsinthelandscapeandtoobserveurbangrowth anddeforestation[12,38]. LCZ:Thisclassificationsystem,introducedbyStewartandOke[14],categorizesurbanandruralareasbasedonsurfacecover,structure,andthermalproperties,providingastandardizedframeworkforcomparingdifferenturbanareas.Thissystemis widelyusedtostudyhowvariousurbanizationformsinfluencemicroclimates,particularlyinrelationtoLST.DifferentphysicalclassificationsofLCZareexplainedin themethodologysection. LST:AcriticalfactorincomprehendingthethermalatmosphereandEarth’ssurface inurbanareas[16,39]isassociatedwithdifferentpatternsofurbangrowth.Many factorsplaycrucialrolesinthelevelofLST,suchassolarradiation,vegetationcover, LULC,urbangrowth,topography,weatherandclimatecondition,andhumanactivities[19–21].LSTisacriticalparameterinstudyingEarth’senergybalance,urban heatislands,climatechange,andenvironmentalprocesses. Duetotheinterplayandconnectionbetweenurbanization,LULC,LCZ,andLST changes(Figure1),themainpointsmentionedbasedontheliteraturereviewareincluded: TheurbanizationprocessinfluencesLULCchanges(LULCCs),alteringhowlandis usedandcovered. LULCCsthenmodifyLCZ,affectingtheclassificationofdifferenturbanandrural areasbasedontheirstructureandsurfaceproperties. TheLCZimpactsLSTbydeterminingthethermalcharacteristicsofdifferentzones andLULC. BoththeLCZandLSTcontributetotheurbanheatislandeffect(UHIE),whereurban areasbecomenoticeablywarmerthantheirruralorsurroundingareasduetothe interplaybetweenurbanization,LULC,andLSTchanges(Figure1). Figure1.Relationshipbetweenurbanizationprocess,LULCC,LCZandLST. Therefore,thisresearchaimstosystematicallyreviewandevaluatetheexistingdata linkedtoLULC,LCZ,andUIandinvestigatetheadverseenvironmentalimpactsofurban growth,particularlyinlong‐termlocalclimatechangepatternsinSeville,Spain.ItexaminestwodecadesofLULCdistributionandchanges,thetypesofurbangrowth,andUI Figure 1. Relationship between urbanization process, LULCC, LCZ and LST. • The urbanization process influences LULC changes (LULCCs), altering how land is used and covered. • LULCCs then modify LCZ, affecting the classification of different urban and rural areas based on their structure and surface properties. • The LCZ impacts LST by determining the thermal characteristics of different zones and LULC. • Both the LCZ and LST contribute to the urban heat island effect (UHIE), where urban areas become noticeably warmer than their rural or surrounding areas due to the interplay between urbanization, LULC, and LST changes (Figure 1). Therefore, this research aims to systematically review and evaluate the existing data linked to LULC, LCZ, and UI and investigate the adverse environmental impacts of urban growth, particularly in long-term local climate change patterns in Seville, Spain. It examines two decades of LULC distribution and changes, the types of urban growth, and UI patterns and density and evaluates LCZ changes, population trends and flow, and their effect on LST variations from 2003 to 2023.
Land 2024,13, 1877 4 of 27 Previous research has often focused on isolated aspects of urbanization, such as land use changes or short-term temperature variations. However, this study addresses both the long-term effects of urbanization on microclimate and the interaction between multiple factors, including urban growth patterns and climate resilience strategies. This research employs advanced methodologies such as remote sensing, geospatial analysis, and machine learning, which were not widely utilized in earlier studies. The integration of these technologies allows for a more comprehensive understanding of how Seville’s urbanization has influenced its microclimate, particularly in terms of long-term LST changes. By applying multi-temporal satellite data and predictive modeling techniques, this study fills a significant gap in the literature, particularly in Mediterranean climates where urban growth patterns and climate dynamics are complex and interdependent and help to identify areas most vulnerable to temperature increases and other climatic changes due to urbanization, which is essential for developing strategies to mitigate the negative effects of urbanization and improve urban sustainability. 2. Materials and Methods The methodology for analyzing the thermal footprint of urbanization involves a mixed-method approach (descriptive and analytic) and a multi-step process combining spatial, statistical, remote sensing techniques and historical documents. Therefore, the methodology of this research consists of three main sections. The Section 2.1 introduces the process and step-by-step structure of data collection, preparation, and the methods and techniques used in Section 2.2. In Section 2.2, all the formulas related to the indices affecting the research objectives are introduced and explained. Finally, in Section 2.3, the case study is presented. 2.1. Methods The methodology is divided into two specific parts: firstly, classification, selection, and preparation that includes 4 separate sections. The first is the process of case study area and the period of time selection (mentioned in Section 2.3). The second specifies the criteria and indicators based on the literature review and the aims of this study (explained in Section 2.2). The third describes data collection and compilation and defines the sources of different types of data and the process of selection and evaluation of the data (defined in Section 2.1). The fourth outlines the collection methods for analysis (mentioned in Section 2.1). The second part, analyzing and modeling, includes 2 main sections. The first describes data pre-processing and preprocessing (defined in Sections 2.1 and 3.3). The second provides a statistical analysis and modeling, and the evaluation of current conditions is explained in the discussion part. The methodology is drawn in Figure 2, which explains these two specific parts of the method. Advanced remote sensing (RS) technology and geospatial information system (GIS and statistical methods) satellite image data, which increase the visibility of the Earth’s surface change [ 40 , 41 ], and statistics of documents of macro plans, reports, research, and projects from organizations and meteorological stations are utilized to analyze and examine the environmental characteristics. In the following sections, this paper introduces indices, models, and validation approaches that evaluate population movement, LULC, LCZ, and LST. The main dataset is multi-temporal Landsat satellite images from the USGS website [42] because this site provides data with high spatial resolution. For data preprocessing, as a crucial step in preparing the satellite products, layer stacking, mosaicking, and clipping multi-band images, the brightness temperature was adjusted and DOS1 atmospheric correction was applied using methods that are explained in the second section of Figure 2. The multi-band images of Landsat 5 TM (path/row: 164/35, collection 2), with a reflectance resolution of 30 m and thermal resolution of 120 m, were acquired from 1 to 31 July 2003, and for Landsat 8 OLI/TIRS (path/row: 164/35), the reflectance resolution was 30 m and the thermal resolution was 100 m, with images collected between 1 and 31 July 2023 used for preprocessing; the satellite images’ regions of interest
Land 2024,13, 1877 5 of 27 (ROI) were applied to ensure minimal cloud coverage (less than 10%) [ 42 ]. Supervised maximum likelihood classifier (MLC), K-mean clustering, and artificial neutral network (ANN) are the most common approaches to classify remote sensing images [ 41 , 43 , 44 ] and to calculate LULCCs; in this study, the support vector machine (SVM) system was used. To evaluate the correlation between indicators, the geographically weighted regression (GWR) equation was used in QGIS, and for assessment of the classification accuracy, overall accuracy and the kappa index were utilized, as they are mentioned in Figure 2. The images were calibrated using accurate classification methods (Section 4, part one, in Figure 2). In addition, this study utilized various software but with a focus on QGIS3.28.8, GIS 10.8.2, and ENVI 5.3 as powerful software for processing and analyzing satellite imagery, which allow for advanced image classification and change detection, and Google Earth Engine (GEE), a cloud-based platform designed for analyzing environmental data on a planetary scale using JavaScript programming language, suitable for large-scale mapping projects. Land2024,13,18775of29 Figure2.Methodology. Fordatapreprocessing,asacrucialstepinpreparingthesatelliteproducts,layer stacking,mosaicking,andclippingmulti‐bandimages,thebrightnesstemperaturewas adjustedandDOS1atmosphericcorrectionwasappliedusingmethodsthatareexplained inthesecondsectionofFigure2.Themulti‐bandimagesofLandsat5TM(path/row: 164/35,collection2),withareflectanceresolutionof30mandthermalresolutionof120 m,wereacquiredfrom1to31July2003,andforLandsat8OLI/TIRS(path/row:164/35), thereflectanceresolutionwas30mandthethermalresolutionwas100m,withimages collectedbetween1and31July2023usedforpreprocessing;thesatelliteimages’regions ofinterest(ROI)wereappliedtoensureminimalcloudcoverage(lessthan10%)[42].Supervisedmaximumlikelihoodclassifier(MLC),K‐meanclustering,andartificialneutral network(ANN)arethemostcommonapproachestoclassifyremotesensingimages [41,43,44]andtocalculateLULCCs;inthisstudy,thesupportvectormachine(SVM)systemwasused.Toevaluatethecorrelationbetweenindicators,thegeographically weightedregression(GWR)equationwasusedinQGIS,andforassessmentoftheclassificationaccuracy,overallaccuracyandthekappaindexwereutilized,astheyarementionedinFigure2.Theimageswerecalibratedusingaccurateclassificationmethods(Section4,partone,inFigure2).Inaddition,thisstudyutilizedvarioussoftwarebutwitha focusonQGIS3.28.8,GIS10.8.2,andENVI5.3aspowerfulsoftwareforprocessingand analyzingsatelliteimagery,whichallowforadvancedimageclassificationandchange detection,andGoogleEarthEngine(GEE),acloud‐basedplatformdesignedforanalyzing environmentaldataonaplanetaryscaleusingJavaScriptprogramminglanguage,suitable forlarge‐scalemappingprojects. Figure 2. Methodology. 2.2. Define Indicators As mentioned in the introduction, there are different types of indicators that can be used to establish a correlation between LULCC, LCZ, and LST. Based on the initial definitions and the relationship between the components, the following indicators have been examined as the main influencing factors: normalized difference vegetation index (NDVI), normalized difference built-up index (NDBI), and UI. NDBI identifies areas with built-up surfaces like roads and buildings, while NDVI indicates areas with healthy vegetation. The formulas employed and method explanations for them are explained in the following sections. • LULCC: Change detection entails comparing two or more images acquired at different times to identify and analyze changes in the landscape [ 12 , 38 ] and the context of the
Land 2024,13, 1877 6 of 27 urban area, which is identified by different colors for the BUA levels for cities, villages, roads, industrial areas; WB levels for rivers, canals, and swamps, A/V/F area; and BS [ 4 , 9 , 12 ]. One common method for LULC change detection involves computing the LULC for each period and then subtracting one from the other [ 8 ]. For a given land cover type (e.g., urban), the area can be calculated using the formula in Equation (1). Area_(Type,Time) = Σ(Pixels classified as Type) ×Area per Pixel (1) The change in a specific land cover type between two time periods (Time1 (2003) and Time2 (2023)) can be calculated as in Equation (2): ∆Area_Type = Area_(Type,Time2) −Area_(Type,Time1) (2) • UI: This clarifies the amount of urban areas in case studies [ 3 , 38 ]. The Shannon entropy model was employed to evaluate urbanization trends and study how the city expands and deforestation occurs [ 11 , 45 ] by measuring the extent of BUAs relative to natural landscapes or other land uses for comparing urbanization levels across regions or tracking urban growth. The structure of the model is as follows, in Equation (3): H=∑n i=1Pi ∗ln(pi)(3) In the formula above, the variables are defined as follows: H = Shannon entropy value; Pi = ratio of BUA; i: area of the total BUA; N = total area [46]. The value of the entropy is from zero to Ln, where (n) represents the amount of sprawl urban growth. A greater value of entropy indicates more dispersed urban growth [ 47 ]. • NDBI: A remote sensing index used to identify changes in BUA density or urban extent cover [ 4 , 8 , 12 ]. The formula for NDBI is typically derived from the reflectance values in the near-infrared and shortwave infrared of the land surface bands of the satellite data (Landsat) and is expressed in Equation (4). NDBI =(SWIR −NIR) (SWIR +NIR)(4) In this formula, the variables are defined as follows: SWIR is the shortwave infrared band; NIR is the near-infrared band. The resulting NDBI values range from − 1 to 1. Positive values indicate a higher likelihood of BUAs, while negative values suggest non-BUAs, such as vegetation or water bodies. NDBI values closer to − 1 typically represent areas with little to no built-up features, such as vegetation or water bodies [ 4 ]. An increase in NDBI values (moving toward 0) suggests an increase in BUAs or urbanization. Conversely, a decrease in NDBI values (moving away from 0 toward − 1) indicates a decrease in BUAs or urbanization. It is crucial to note that the effectiveness of the NDBI for mapping BUAs may depend on the characteristics of the study area, the spatial resolution of the satellite imagery, temporal resolution, and the presence of factors such as shadows or mixed land cover types [8]. • NDVI: This is used to estimate vegetation health, density, and coverage [ 48 ], as well as the intensity of greenness and surface radiant temperature of the area. It is calculated using the reflectance values from the NIR and RED bands surface reflectance data of Landsat 5 TM (bands 4 and 3) and Landsat 8 OLI/TIRS satellite sensor (bands 5 and 4). NDVI operates on the concept that vibrant, healthy vegetation primarily
Land 2024,13, 1877 7 of 27 absorbs most of the visible light (specifically red) and reflects a significant amount of the near-infrared light [4]. Equation (5) is as follows: NDVI =(NIR −RED) (NIR +RED)(5) NIR: Near-infrared band; RED: Red band. The index value of NDVI ranges from −1 to 1. Positive and high NDVI values closer to +1 indicate a higher likelihood of denser and healthier vegetation, while low NDVI values closer to 0 and negative values suggest sparse or unhealthy vegetation, and negative NDVI values closer to − 1 typically indicate non-vegetated surfaces, such as BS. A decrease in NDVI values, particularly in urban areas, indicates a reduction in the density of green spaces, likely due to increased urbanization [48]. • LCZ: This system was developed to provide a standardized method for describing and comparing the physical characteristics of different areas, particularly in the context of urban climate research [ 36 , 37 ]. The classification of LCZ defined by Stewart and Oke [ 14 ] is based on qualitative assessments of urban and rural landscapes. The 17 standard LCZ types are categorized into two main groups: urban LCZ types and natural LCZ types, as follows in Figure 3: Land2024,13,18777of29 bodies.NDBIvaluescloserto−1typicallyrepresentareaswithlittletonobuilt‐upfeatures,suchasvegetationorwaterbodies[4].AnincreaseinNDBIvalues(movingtoward 0)suggestsanincreaseinBUAsorurbanization.Conversely,adecreaseinNDBIvalues (movingawayfrom0toward−1)indicatesadecreaseinBUAsorurbanization.Itiscrucial tonotethattheeffectivenessoftheNDBIformappingBUAsmaydependonthecharacteristicsofthestudyarea,thespatialresolutionofthesatelliteimagery,temporalresolution,andthepresenceoffactorssuchasshadowsormixedlandcovertypes[8]. NDVI:Thisisusedtoestimatevegetationhealth,density,andcoverage[48],aswell astheintensityofgreennessandsurfaceradianttemperatureofthearea.ItiscalculatedusingthereflectancevaluesfromtheNIRandREDbandssurfacereflectance dataofLandsat5TM(bands4and3)andLandsat8OLI/TIRSsatellitesensor(bands 5and4).NDVIoperatesontheconceptthatvibrant,healthyvegetationprimarily absorbsmostofthevisiblelight(specificallyred)andreflectsasignificantamountof thenear‐infraredlight[4].Equation(5)isasfollows: 𝑁𝐷𝑉𝐼 𝑁𝐼𝑅𝑅𝐸𝐷 𝑁𝐼𝑅𝑅𝐸𝐷(5) NIR:Near‐infraredband; RED:Redband. TheindexvalueofNDVIrangesfrom−1to1.PositiveandhighNDVIvaluescloser to+1indicateahigherlikelihoodofdenserandhealthiervegetation,whilelowNDVIvaluescloserto0andnegativevaluessuggestsparseorunhealthyvegetation,andnegative NDVIvaluescloserto−1typicallyindicatenon‐vegetatedsurfaces,suchasBS.Adecrease inNDVIvalues,particularlyinurbanareas,indicatesareductioninthedensityofgreen spaces,likelyduetoincreasedurbanization[48]. LCZ:Thissystemwasdevelopedtoprovideastandardizedmethodfordescribing andcomparingthephysicalcharacteristicsofdifferentareas,particularlyinthecontextofurbanclimateresearch[36,37].TheclassificationofLCZdefinedbyStewart andOke[14]isbasedonqualitativeassessmentsofurbanandrurallandscapes.The 17standardLCZtypesarecategorizedintotwomaingroups:urbanLCZtypesand naturalLCZtypes,asfollowsinFigure3: Figure3.UrbanandnaturalLCZtypesbasedonStewartandOke[14]. Figure 3. Urban and natural LCZ types based on Stewart and Oke [14]. There are mainly three types of LCZ classification methods: the remote-sensing-based method, including WUDAPT L0, LCZ Generator, and GEE; the GIS-based method; and the combined method [ 17 , 19 , 36 ]. Generally, the GIS-based method is applied for urban areas, estimating parameters such as building height, density, greening rate, and other spatial metrics, while the remote-sensing-based method is more suitable for natural coverage areas, calculating indices like surface cover, vegetation fraction, and albedo [ 21 , 49 ]. In this research, the primary LCZ classification process was carried out using the LCZ Generator, which provides templates for use with GIS during the sampling process. These templates were also used as auxiliary data to enhance the overall classification accuracy. The LCZ Generator follows WUDAPT’s (World Urban Database and Access Portal Tools) default protocol [ 35 , 36 ], which has been sequentially improved over time to facilitate single-city LCZ mapping [ 21 , 49 ]. For the classification, WUDAPT_L0_Training_template [ 50 ] was used to generate L0 data. After collecting samples for the 17 LCZ classes, the remaining
Land 2024,13, 1877 8 of 27 steps were performed using the LCZ Generator according to the submission form on the website [ 17 , 36 , 50 ]. To ensure accuracy, the random forest (RF) algorithm was applied. This method classifies each pixel into an LCZ type based on decision trees and requires minimal training data. To validate the results, the classification process was performed using both the LCZ Generator and the GIS template for consistency, following the WUDAPT_L0_Training_template [ 50 ]. By comparing the outputs of both systems, any potential errors were identified and addressed. The RF algorithm demonstrated high accuracy, and the classification results were confirmed to be consistent with minimal discrepancies. • LST: This can be derived from thermal infrared data from satellite sensors [ 20 , 51 ]. The formula to convert the brightness temperature to LST uses the thermal band data from Landsat. This formula corrects for emissivity and atmospheric effects to estimate the actual LST [ 19 , 21 ]. The formula involves several steps, including the use of thermal infrared (TIR) imagery and the application of Planck’s law and atmospheric correction models. Equation (6) is as follows: LST(◦C)=TB 1+λ×TB ρ×ln ε −273.15 (6) TB: At-satellite brightness temperature in Kelvin, and the final LST is in Celsius; λ: Wavelength of emitted radiance; ρ: Constant (h*c/σ= 1.438 ×10−2mK); ε: Surface emissivity. High LST values suggest that the surface might occur in urban areas, deserts, or other regions with little vegetation or high heat-absorbing materials, like asphalt and concrete. Low LST values indicate a surface typically found in areas with dense vegetation, water bodies, or regions with a higher moisture content. 2.3. Study Area The study area of this research is the city of Seville and its urban settlement area. Seville, in Spain, is situated at 37.3891 ◦ N, 5.9845 ◦ W [ 51 , 52 ]. Spread over an area of 141.28 km 2 , with an average elevation of 6 m above sea level, Seville experiences a Mediterranean climate characterized by hot, dry summers and mild, rainy winters. The average annual temperature in 2020 was around 19 ◦ C, with summer highs often exceeding 35 ◦ C, and every year, 40 ◦ C is surpassed on several occasions [ 8 , 53 ]. Winters are mild, and the city receives about 550 mm of rainfall annually, mostly between October and April. This climate, along with rapid urbanization, influences the city’s microclimate and UHIE. Seville is a city with a rich history and cultural heritage dating back to Roman times, and due to being inhabited for centuries, Seville, like many other cities worldwide, underwent significant rapid urban development and substantial urban expansion during the first decade of the 21st century. This period saw a strong boost from public and private investments, which translated into numerous infrastructure projects, urban developments, and economic initiatives [ 8 , 54 – 56 ]. 3. Results In this section, the main indicators that were mentioned in the methodology section are evaluated and overlapped for providing an understanding of the condition of the study area over a period of 20 years. The parameters include an evaluation of the history of the urban growth trend, population trend, UI trend, and LULCC, NDBI and NDVI, and LCZ and LST patterns from 2003 to 2023. In the first step, the history of urbanization in Seville was studied to find the main causes and reasons for change, and then the process of changing over time was evaluated with the help of previous plans, research studies, and historical documents.
Land 2024,13, 1877 9 of 27 3.1. History of Urban Growth Trend This study meticulously examines the various facets of urban growth in Seville, incorporating insights from contemporary urban studies and previous plans, programs, and projects to provide a comprehensive overview of Seville’s urban growth trends and their climatic ramifications from 2003 to 2023. Based on findings from previous research and under the new General Urban Development Plans (1988, 2006, and 2018), the process of urban growth in the previous century was influenced by various internal and external factors [ 5 , 8 ]. Some World Exhibitions and national and international events and projects, such as the Ibero-American Exposition 1929–1930 (from 1015 to 1029, 36.19% increase in BUA) and the World Exhibition—EXPO’92 (from 1981 to 1992, 40.79% increase in BUA) [ 53 ], had a huge effect of shaping the context of the city in 21 centuries. By examining the city’s structure, Seville shows an ecological design with a central urban area and radial growth, in which newly developed areas are gradually integrated over several years [57]. In Figure 4, the map of urban growth from 1918 to 1995 (before the period of time that was evaluated in this research) displays rapid and varied types of urbanization and shows the direction of growth in the city of Seville. Initially, growth was circular and compact and involved infill development, but over the decades, it transitioned to ribbon and sprawl [8,55,56]. Land2024,13,187710of29 Figure4.VisualrepresentationofurbangrowthtrendsandchangesinthecityofSevilleoverthe decadesbasedonthemasterplanofthecity,2006[8,54,55,57]. ThesedataillustrateandprovethephysicalgrowthofthecityofSevillebefore2003, highlightingtheimportanceofstudyingthishistoricaldataduetotherapidurbanization andhighpopulationgrowthrateinthepreviouscentury,theeffectsofwhicharestill evidentyearslater.InFigure4,theamountofeachtypeofurbangrowthisevaluatedand classifiedfromcompacttomediumcompact,andfrommediumdispersedtototallydispersedorsprawl[8,52–54].Itisnotedthattherewasalimitationinaccessingthedata; therefore,forthepreviouscentury,onlytheperiodfrom1975to2023ismentioned(fora betterunderstandingoftheprocessofchanging,thetimeofevaluationwasextended). Additionally,thepreprocessingandprocessingimages(Figure5)clarifythatthe trendofdispersedorsprawlgrowthinthecitywasdrivenbytheneedtoconnecttosettlementsaroundthemaincity,resultingfromrapidurbanization.Now,inthesatellite imagerelatedto2023,wearefacedwithanurbanarea,notjustacity,basedonthedirectionandpatternofsprawlandchangesinurbanizationinthestudyarea.Thesechanges demonstrateanetworkofconnectionsbetweenthecityandsurroundingurbanareas. Therefore,inthisstudy,thecityofSevilleanditsfirstlayerofsettlementwillbeevaluated; inthesatelliteimages,abufferof10Kmfromthecenterofthecityisanalyzed. Figure 4. Visual representation of urban growth trends and changes in the city of Seville over the decades based on the master plan of the city, 2006 [8,54,55,57]. These data illustrate and prove the physical growth of the city of Seville before 2003, highlighting the importance of studying this historical data due to the rapid urbanization and high population growth rate in the previous century, the effects of which are still evident years later. In Figure 4, the amount of each type of urban growth is evaluated and classified from compact to medium compact, and from medium dispersed to totally dispersed or sprawl [ 8 , 52 – 54 ]. It is noted that there was a limitation in accessing the data;
Land 2024,13, 1877 16 of 27 Land2024,13,187717of29 reflectinfrastructureexpansionorcitydevelopmentpoliciespushingurbansprawlpattern.BSchanges(yellow)areevident,especiallyinthelatterdecade(2013–2023),suggestingeitherlandclearance,deforestation,orareaspreparedforfutureconstructionordevelopment.A/V/Fchangesrepresentfluctuationsinagriculture,forests,orvineyards. Whiletherearepocketsofthesechanges,theyseemlessdominantintheoveralltransition comparedtoBUAs,suggestingmorefocusonurbangrowththanagriculturalexpansion orreforestation.WBchangesindicatewaterbodychanges,buttheminimalrepresentation ofthiscategorysuggeststhatwater‐relatedchangeswerelesssignificantinthis timeframe. Figure12.ComparisonoftypesofchangesinLULCovertime. Ingeneral,thenorthernandsouthernzonesofthehighlightedregioninSevilleappeartohaveexperiencedthemostintensechanges.Inparticular,theexpansionofBUAs isevidentinthesouthernperipheries,potentiallyindicatingsuburbanizationorthedevelopmentofnewresidentialorindustrialzones.Innercityareas,particularlythose withinthemarkedboundary,showlesschange,suggestingthaturbandensificationor expansionhasprimarilyoccurredintheoutskirts,whilethecentralcityarearemainsstable. Thesechanges,particularlythesignificantriseinBUAs,likelypointtourbandevelopmentandlandpreparationforfutureprojects.Therefore,thereisaneedforevaluation ofcityreportsanddevelopmentplans.Oneofthemaindatapointsthathelpscategorize thechangesisthecitydatafromopen‐accessresourcesandcitydocuments.Thecity’s strategicplanswerestudied,andtwomapsweremadedemonstratingwhatthedevelopmentprocesswouldbelikeinthefuture(Figure13).Thesetwomapsaredevelopment planningandclassificationofurbandevelopableandnon‐developableareas,basedon onlineGISdataofSeville[60].ThemapofdevelopmentandUImapfromsatelliteimages showthesamepattern,meaningthenumberofchangesinUIisbasedonpredictedplans andprojectsofthecitymasterplan,respectinglegal,environmental,andhistoricalopportunitiesandconstraints,whichisbecauseofurbanplanninglawsandregulationsthat haveinfluencedlandusechangesandthedirectionofgrowth. Figure 12. Comparison of types of changes in LULC over time. These changes, particularly the significant rise in BUAs, likely point to urban development and land preparation for future projects. Therefore, there is a need for evaluation of city reports and development plans. One of the main datapoints that helps categorize the changes is the city data from open-access resources and city documents. The city’s strategic plans were studied, and two maps were made demonstrating what the development process would be like in the future (Figure 13). These two maps are development planning and classification of urban developable and non-developable areas, based on online GIS data of Seville [60]. The map of development and UI map from satellite images show the same pattern, meaning the number of changes in UI is based on predicted plans and projects of the city master plan, respecting legal, environmental, and historical opportunities and constraints, which is because of urban planning laws and regulations that have influenced land use changes and the direction of growth. The development planning map (Figure 13A) shows different stages of urban planning across the city, including areas that are still being processed, those that have received final approval (pink), and areas where development has been annulled due to legal reasons (purple). It highlights zones of significant interest where urban development is either ongoing or proposed. The classification map (B) provides a detailed categorization of land into developable and non-developable zones based on its potential for urban development and its conservation status [ 61 , 62 ]. Developable areas are shown in colors like yellow and orange, indicating land that can be urbanized, while green areas represent non-developable zones, such as protected natural or rural areas. There is a clear distinction between urban expansion zones (developable land) and protected or rural zones (non-developable land). These maps, together, illustrate the planning status and how urban planning is structured in terms of both the approval process and land categorization, helping to visualize which areas will undergo future development and which are preserved. These maps also provide a comprehensive view of urban development strategy, highlighting the complexities of managing growth and showing which areas are prioritized for urbanization and which are protected from it.
Land 2024,13, 1877 17 of 27 The combination of Figures 12 and 13 indicates that the areas designated as “development zones” in the proposed plan have experienced the highest levels of change in both BS and BUAs. Notably, these changes became more pronounced during the second decade of the period under review. These maps are based on current data from the city of Seville, obtained through up-to-date sources and modeled in conjunction with satellite imagery. Land2024,13,187718of29 Thedevelopmentplanningmap(Figure13A)showsdifferentstagesofurbanplanningacrossthecity,includingareasthatarestillbeingprocessed,thosethathavereceived finalapproval(pink),andareaswheredevelopmenthasbeenannulledduetolegalreasons(purple).Ithighlightszonesofsignificantinterestwhereurbandevelopmentiseither ongoingorproposed.Theclassificationmap(B)providesadetailedcategorizationofland intodevelopableandnon‐developablezonesbasedonitspotentialforurbandevelopmentanditsconservationstatus[61,62].Developableareasareshownincolorslikeyellow andorange,indicatinglandthatcanbeurbanized,whilegreenareasrepresentnon‐de‐ velopablezones,suchasprotectednaturalorruralareas. Figure13.Mapofdevelopmentplanningandclassificationofurbandevelopabletypes[60]. Thereisacleardistinctionbetweenurbanexpansionzones(developableland)and protectedorruralzones(non‐developableland).Thesemaps,together,illustratetheplanningstatusandhowurbanplanningisstructuredintermsofboththeapprovalprocess andlandcategorization,helpingtovisualizewhichareaswillundergofuturedevelopmentandwhicharepreserved.Thesemapsalsoprovideacomprehensiveviewofurban developmentstrategy,highlightingthecomplexitiesofmanaginggrowthandshowing whichareasareprioritizedforurbanizationandwhichareprotectedfromit. ThecombinationofFigures12and13indicatesthattheareasdesignatedas“developmentzones”intheproposedplanhaveexperiencedthehighestlevelsofchangeinboth BSandBUAs.Notably,thesechangesbecamemorepronouncedduringtheseconddecadeoftheperiodunderreview.Thesemapsarebasedoncurrentdatafromthecityof Seville,obtainedthroughup‐to‐datesourcesandmodeledinconjunctionwithsatellite imagery. 3.3.4.LCZ Aswasmentionedinthemethodologysection,tocomparetheLCZsofSevillebetween2003and2023,severalfactorsshouldbeconsidered,includingchangesinurbanization,vegetationcover,industrialexpansion,andinfrastructuraldevelopments [14,35,36].Sevillehasbecomemoreurbanized,especiallywiththeriseofmid‐riseand high‐risebuildings.Newareashavebeendeveloped,andsuburbanzoneshavebecome morecompact.In2003,thedenselybuiltenvironmentinthecitycenterandlackofgreenerywasobvious,andfocusonclimatemitigationinurbandesignwaslesspronounced, Figure 13. Map of development planning and classification of urban developable types [60]. 3.3.4. LCZ As was mentioned in the methodology section, to compare the LCZs of Seville between 2003 and 2023, several factors should be considered, including changes in urbanization, vegetation cover, industrial expansion, and infrastructural developments [ 14 , 35 , 36 ]. Seville has become more urbanized, especially with the rise of mid-rise and high-rise buildings. New areas have been developed, and suburban zones have become more compact. In 2003, the densely built environment in the city center and lack of greenery was obvious, and focus on climate mitigation in urban design was less pronounced, but over the past 20 years, the city has introduced efforts to introduce more green spaces through tree planting, construct sustainable buildings, and incorporate smart infrastructure to control climate changes, to expand LCZs A and B (Figure 3), and to limit excessive temperature rise in LCZs 1–3. The general data of the LCZ maps is defined by the LCZ Generator [50] (Table 5): Table 5. General data of LCZ map; LCZ Generator [36,50]. Submission Information Submission Date Representative Date Seville LCZ map 23 July 2024 12 December 2023 16 July 2024 12 September 2003 Figure 14 shows the LCZ distribution in Seville in 2003 and 2023. The colors represent different LCZ classes, such as urban areas, vegetation, bare soil, water bodies, etc. In 2003, the urban region was more concentrated in a central location with less surrounding urban sprawl, but over 20 years, there are visible changes in the land cover, with a likely transition of some natural zones into more developed or urbanized regions and a visible increase in red zones (LCZs 1 and 2), indicating a significant rise in built-up or dense urban areas between 2003 and 2023.
Land 2024,13, 1877 18 of 27 Land2024,13,187719of29 butoverthepast20years,thecityhasintroducedeffortstointroducemoregreenspaces throughtreeplanting,constructsustainablebuildings,andincorporatesmartinfrastructuretocontrolclimatechanges,toexpandLCZsAandB(Figure3),andtolimitexcessive temperatureriseinLCZs1–3.ThegeneraldataoftheLCZmapsisdefinedbytheLCZ Generator[50](Table5): Table5.GeneraldataofLCZmap;LCZGenerator[36,50]. SubmissionInformationSubmissionDateRepresentativeDate SevilleLCZmap23July202412December2023 16July202412September2003 Figure14showstheLCZdistributioninSevillein2003and2023.Thecolorsrepresent differentLCZclasses,suchasurbanareas,vegetation,baresoil,waterbodies,etc.In2003, theurbanregionwasmoreconcentratedinacentrallocationwithlesssurroundingurban sprawl,butover20years,therearevisiblechangesinthelandcover,withalikelytransitionofsomenaturalzonesintomoredevelopedorurbanizedregionsandavisibleincreaseinredzones(LCZs1and2),indicatingasignificantriseinbuilt‐upordenseurban areasbetween2003and2023. Figure14.LCZmap;analyzedinQGISandLCZGenerator[42,50,58]. Theareaandclasseswiththemostsignificantchangesbasedonthefindingsofthe LCZmapsandmapsofUIandurbandevelopmentandtheresultsofthemasterandstrategicplansofSevilleareshowninTable6. Figure 14. LCZ map; analyzed in QGIS and LCZ Generator [42,50,58]. The area and classes with the most significant changes based on the findings of the LCZ maps and maps of UI and urban development and the results of the master and strategic plans of Seville are shown in Table 6. Table 6. Main changes in LCZ classes in Seville urbanization and built environment (2003 vs. 2023). LCZ Classes 2003 2023 LCZ 1 (compact high-rise) Limited to a few business districts. Further urban development and modernization added areas like Torre Sevilla and new commercial centers. LCZ 2 (compact mid-rise) Central part of Seville, especially historic areas like Santa Cruz and El Arenal, retained their mid-rise, compact structures. The central part has remained largely unchanged, as historical preservation laws protect this area. LCZ 8 (large low-rise) Industrial areas were still in expansion, but significant growth was underway. Large low-rise zones have expanded significantly due to the growth of the Seville industrial park and commercial warehousing sectors LCZ 9 (sparsely built) Areas on the outskirts, with large plots of land, are still undeveloped. Urban expansion and developments have reduced sparsely built areas. LCZ 10 (heavy industry) Industrial areas were concentrated in specific zones (on the outskirts). Industrial expansion, in the Zona Franca and areas near the port, reflect Seville’s growing industrial economy. LCZ A (dense trees) Parks, such as Parque de María Luisa, were well-maintained. The expansion of green spaces and urban forestry projects slightly expanded LCZ A. LCZ B (scattered trees) Tress were commonly found in residential neighborhoods and near the Guadalquivir River. Expansion of urban gardens and street trees as part of Seville’s environmental policies increases this type of LCZ.
Land 2024,13, 1877 19 of 27 The box plots on the right side show the accuracy and F1 metrics for different LCZ classes (Figure 14). Each color in the box plot corresponds to a different LCZ class, as shown in the map legend. The higher the box plot, the more accurately the LCZ class has been predicted or classified. It can be seen that some LCZ classes, like urban areas, might have higher accuracy compared to others. This metric combines precision and recall, providing an overall measure of classification performance. High F1 values in the box plots mean that certain LCZ classes, such as those in purple and blue, have been classified with greater consistency. 3.3.5. LST LST, as a primary indicator for the evaluation of climate change and environmental conditions, is classified based on a period of 20 years. Figure 15 indicates that the entire urban area has experienced an excessive increase in mean temperature because of urban expansion, LCZ changes, and a dense population. The maps in Figure 15 show how different types of land cover absorb and retain heat. Land2024,13,187721of29 Figure15.AnalysisofLST(°C)patternovertimeandchangesinLSTmean. Mostlybuiltareasarehotterthanotherareas,andagriculturelandincreasesthermal comfort,butthenumberofchangesintheA/F/Vareaissignificantlynoticeable.ThecentralpartofSeville,markedbygreenareas,andthehistoricalareaindicatelowertemperaturescomparedtothesurroundingregions.Thissuggeststhatthecoreurbanarea,despitebeingdenselybuilt‐up,hasmanagedtoavoidthesameleveloftemperatureincrease astheoutlyingareas. Togainaclearerinsightintothepatternofchanges,datatrendsfrom1993to2023 wereevaluatedandaredisplayedinFigure16.Thegraphshowsaconsistentupward trendinthemeanLSTinJuly,from30.13°Cin1993to38.37°Cin2023[61,62].Thishighlightsanear8°Criseinsurfacetemperaturesover30years.Intotal,theannualaverage increaseratefrom2003to2013is0.198°Cperyear;from2013to2023is0.228°Cperyear; andtheoverallrate(from1993to2023)is0.275°Cperyear.Thecompoundannualgrowth rate(CAGR)iscalculatedbasedEquation(7): 𝐶𝐴𝐺𝑅 𝑉𝑎𝑙𝑢𝑒 𝑎𝑡 𝑡ℎ𝑒 𝑒𝑛𝑑 𝑉𝑎𝑙𝑢𝑒 𝑎𝑡 𝑡ℎ𝑒 𝑏𝑒𝑔𝑖𝑛𝑖𝑛𝑔 1(7) Valueattheend:temperaturein2023; Valueatthebeginning:temperaturein2003; NumberofYears:between2003and2023(20years). TheCAGRfortheLSTinSevillefrom2003to2023is0.59%peryear;thatishigher thantheglobalaverageforurbanareas[61,62].TheglobalurbanLSThasincreasedby approximately0.38%peryearduringasimilartimeframe[61]. Figure16.MeanofLST(°C)andLULCCinurbanareainJuly2003and2023. Figure 15. Analysis of LST (◦C) pattern over time and changes in LST mean. The LST maps were calculated using RS-based Landsat 5 TM, Collection 2 (band 6) for 2003 and Landsat 8 TIRS (band 10) for 2023, with temperatures expressed in Celsius. The data were obtained from satellite images collected between 1 July and 31 July for both years. The final LST map is the result of combining multiple satellite images captured over this time period to create a comprehensive representation of typical summer conditions. In ENVI, standard pre-processing techniques, including atmospheric correction and layer stacking, were applied to ensure accurate and consistent LST calculations [42]. Mostly built areas are hotter than other areas, and agriculture land increases thermal comfort, but the number of changes in the A/F/V area is significantly noticeable. The central part of Seville, marked by green areas, and the historical area indicate lower temperatures compared to the surrounding regions. This suggests that the core urban area, despite being densely built-up, has managed to avoid the same level of temperature increase as the outlying areas. To gain a clearer insight into the pattern of changes, data trends from 1993 to 2023 were evaluated and are displayed in Figure 16. The graph shows a consistent upward trend in the mean LST in July, from 30.13 ◦ C in 1993 to 38.37 ◦ C in 2023 [ 61 , 62 ]. This highlights a near 8 ◦ C rise in surface temperatures over 30 years. In total, the annual average increase rate from 2003 to 2013 is 0.198 ◦ C per year; from 2013 to 2023 is 0.228 ◦ C per year; and the overall rate (from 1993 to 2023) is 0.275 ◦ C per year. The compound annual growth rate (CAGR) is calculated based Equation (7):
Land 2024,13, 1877 20 of 27 CAGR =Value at the end Value at the begining 1 Number o f Years −1 (7) Value at the end: temperature in 2023; Value at the beginning: temperature in 2003; Number of Years: between 2003 and 2023 (20 years). Land2024,13,187721of29 Figure15.AnalysisofLST(°C)patternovertimeandchangesinLSTmean. Mostlybuiltareasarehotterthanotherareas,andagriculturelandincreasesthermal comfort,butthenumberofchangesintheA/F/Vareaissignificantlynoticeable.ThecentralpartofSeville,markedbygreenareas,andthehistoricalareaindicatelowertemperaturescomparedtothesurroundingregions.Thissuggeststhatthecoreurbanarea,despitebeingdenselybuilt‐up,hasmanagedtoavoidthesameleveloftemperatureincrease astheoutlyingareas. Togainaclearerinsightintothepatternofchanges,datatrendsfrom1993to2023 wereevaluatedandaredisplayedinFigure16.Thegraphshowsaconsistentupward trendinthemeanLSTinJuly,from30.13°Cin1993to38.37°Cin2023[61,62].Thishighlightsanear8°Criseinsurfacetemperaturesover30years.Intotal,theannualaverage increaseratefrom2003to2013is0.198°Cperyear;from2013to2023is0.228°Cperyear; andtheoverallrate(from1993to2023)is0.275°Cperyear.Thecompoundannualgrowth rate(CAGR)iscalculatedbasedEquation(7): 𝐶𝐴𝐺𝑅 𝑉𝑎𝑙𝑢𝑒 𝑎𝑡 𝑡ℎ𝑒 𝑒𝑛𝑑 𝑉𝑎𝑙𝑢𝑒 𝑎𝑡 𝑡ℎ𝑒 𝑏𝑒𝑔𝑖𝑛𝑖𝑛𝑔 1(7) Valueattheend:temperaturein2023; Valueatthebeginning:temperaturein2003; NumberofYears:between2003and2023(20years). TheCAGRfortheLSTinSevillefrom2003to2023is0.59%peryear;thatishigher thantheglobalaverageforurbanareas[61,62].TheglobalurbanLSThasincreasedby approximately0.38%peryearduringasimilartimeframe[61]. Figure16.MeanofLST(°C)andLULCCinurbanareainJuly2003and2023. Figure 16. Mean of LST (◦C) and LULCC in urban area in July 2003 and 2023. The CAGR for the LST in Seville from 2003 to 2023 is 0.59% per year; that is higher than the global average for urban areas [ 61 , 62 ]. The global urban LST has increased by approximately 0.38% per year during a similar timeframe [61]. In Figure 16, each category shows an increase in the LST. The BS category experienced the highest increase, with the LST rising from 34 ◦ C in 2003 to 40.5 ◦ C in 2023, a 6.5 ◦ C increase, followed by BUA; the locations of these hotspot areas are specified in Figure 15. As the LST increases, the NDVI declines, leading to more extensive areas of stress and possibly desertification or reduced agricultural productivity over time. This relationship highlights the environmental changes happening in Seville, possibly due to climate change, urbanization, or other anthropogenic factors, and clarifies that Seville is experiencing a relatively rapid increase in LST due to rapid urbanization process in past decades. In connection with Figures 11 and 13, that means this amount of change and increase in hotspot areas is the result of the proposed city plans from previous years [60]. 4. Discussion This section uses overlapping maps and thermal images to illustrate the direct correlation between urban expansion, the LCZ, and increased LST. These visuals highlight that urbanized (BUAs) and barren soil (BS) areas tend to be warmer, while vegetated and WB zones are cooler. Figure 15 specifically shows that in 2023, BS areas experienced the most significant temperature increase over the last 20 years, with all the LULC categories generally showing rising LSTs. These findings correlate urban expansion with temperature increases, in comparison with the 20-year LCZ maps, and, particularly, an increase in density in the central (not the historical part of the city) part and expanding outwards corresponds with reductions in BS areas. It is evident that areas that have undergone changes in the density patterns of their urban fabric, as well as those where classification or LULC has shifted based on the land context, have experienced the most significant changes in LST. These changes are illustrated in Figure 14 in the LCZ section. The city center of Seville, with its compact mid-rise (LCZ 2) and low-rise (LCZ 3) buildings, was already experiencing higher temperatures compared to the surrounding areas, especially in summer. New developments, particularly the increase in high-rise buildings (LCZ 1) and expansion in industrial zones (LCZ 10), have contributed to higher LSTs in these areas. The addition of more high-rise buildings has led to significant increases in LST in areas like Torre Sevilla and the Triana district, where tall buildings trap heat. The LST in
Land 2024,13, 1877 21 of 27 these areas is significantly higher compared to low-rise or suburban zones due to reduced vegetation and the heat-retention properties of concrete and asphalt (Figure 15). While the LST in low-rise zones (LCZ 3 and LCZ 6) has remained high, urban greening efforts and changes in construction materials have slightly mitigated temperature increases in some neighborhoods. Industrial expansion (LCZ 8 and LCZ 10) has contributed to an increase in LST. These areas are now some of the hottest zones in Seville, with large heat-retaining surfaces and minimal vegetation (e.g., asphalt, concrete). In addition, the LST of urban areas, based on UI (Figures 11 and 12), was compared with the LST of non-urban areas to understand the impact of urbanization on local temperatures. This is reflected in the UI maps (Figure 11) and LST maps (Figure 15), showing more red areas (BUAs) in 2023 compared to 2003, indicating an increase in the temperature. Therefore, there is a need for an urbanization pattern map based on overlapping UI maps over time (Figure 17). This map shows Seville’s urban growth trends, pattern directions, and limitations, and reveals significant urban growth through sprawl and ribbon development, particularly at the city’s borders and in suburban areas, with dense infill development in central urban areas. Land2024,13,187723of29 Figure17.Thepatternofurbanizationprocessanddirectionofurbangrowth. 4.1.ThePatternoftheUrbanizationProcess Figure17highlightstheroleofLULCpatternsinshapinglocaltemperatureregimes andindicatesatrendtowardamixeddevelopmentpatternovera20‐yearperiod.The mapalsoshowsanincreasingofurbandispersion,basedonthecharacteristicsofBUAs, andareductioninBS,influencedbyfactorssuchasinfrastructureandnetworkenhancements,whichhaveshiftedresidentialpatternsfromthecitytomoreaffordablesuburban areas,andhistoricalconservation,whoseeffortshavelimitedchangesinSeville’shistoricalcore,preservingitscharacter[53,54]. Thecentral,eastern,andnorthernpartsofSevillearethemainhotspotsexperiencing themostsignificanttemperatureincreases,whilethehistoricalcenter,areasneartheriver, andthewesternandsouthernregionsalsoseewarming,buttoalesserextentcompared tothecoreurbanareas.ComparingmapsoftheLSTandLULCCshowthatthestudyarea hasexperiencedrapidurbanization,andthehighestaveragetemperaturechangeshappen inthesementionedareas.Theseareasareshownbasedontheirdistancesfromthecenter ofthecity,witharadianceof10Km(Figure18),whichprovesthatthepartsoftheurban areathathadsignificantincreasesinLSTaretheareaswithsignificantLULCC,NDBI,and NDVIchanges.Forinstance,areaswithlowNDVIvaluesarepredictedtooftenexhibit higherLSTvaluesduetothelackofvegetationcoverandincreasedheatabsorptionby built‐upsurfaces.Anotherpointthatmustbementionedisrelatedtothepopulationtrend (Figure6).ThecityofSevillefacedanalmostzerogrowthrateinrecentyears,whilethe urbanareahashadasignificantgrowthrate;therefore,thepatternofchangearoundthe citybasedonthegrowthrateintheurbanareaandtheresultofthecity’sdevelopment strategiesisacceptable. Figure 17. The pattern of urbanization process and direction of urban growth. 4.1. The Pattern of the Urbanization Process Figure 17 highlights the role of LULC patterns in shaping local temperature regimes and indicates a trend toward a mixed development pattern over a 20-year period. The map also shows an increasing of urban dispersion, based on the characteristics of BUAs, and a reduction in BS, influenced by factors such as infrastructure and network enhancements, which have shifted residential patterns from the city to more affordable suburban areas, and historical conservation, whose efforts have limited changes in Seville’s historical core, preserving its character [53,54]. The central, eastern, and northern parts of Seville are the main hotspots experiencing the most significant temperature increases, while the historical center, areas near the river, and the western and southern regions also see warming, but to a lesser extent compared to
Land 2024,13, 1877 22 of 27 the core urban areas. Comparing maps of the LST and LULCC show that the study area has experienced rapid urbanization, and the highest average temperature changes happen in these mentioned areas. These areas are shown based on their distances from the center of the city, with a radiance of 10 Km (Figure 18), which proves that the parts of the urban area that had significant increases in LST are the areas with significant LULCC, NDBI, and NDVI changes. For instance, areas with low NDVI values are predicted to often exhibit higher LST values due to the lack of vegetation cover and increased heat absorption by built-up surfaces. Another point that must be mentioned is related to the population trend (Figure 6). The city of Seville faced an almost zero growth rate in recent years, while the urban area has had a significant growth rate; therefore, the pattern of change around the city based on the growth rate in the urban area and the result of the city’s development strategies is acceptable. Land2024,13,187724of29 Figure18.AreawithsignificantchangesinLST(2003–2023). Figure17providesavisualrepresentationofdifferenttypesofurbangrowthwithin thecityandsurroundingarea,basedonhistoricaltrendsandLCZchangesandthefindingsofFigure13,thatoverlap,usingGISandrasterdatabases.Thegrowthdirections showthemaindirectionsinwhichthecity’surbanexpansionisoccurring,pointingtowardtheoutskirtsorlessdevelopedareasoftheprovincealongmajortransportroutes. Thishighlightsalinearexpansionpatternthatcaninfluencetransportationandhousing policies.Thepresenceofinfillgrowthincentralareassuggestsattemptstoutilizeavailablespaceefficiently,whereasthesprawlgrowtharoundtheperipheryindicatesongoing expansion,possiblyatthecostofgreenfieldoragriculturalland. EvaluatingFigure17withFigure13,thesemainpointsmustbementioned: Areas(city)markedas“inprocess”or“finalproposal”inmap12.Aalignwithregionsshowinginfillorsprawlgrowthinthecurrentmap,indicatingwherefuture developmentsaretargetedorplanned. Areasmarkedas“annulledbysentence”couldcorrelatewithareasthatarenotexperiencingsignificantgrowthinthecurrentmap,suggestingthatregulatoryrestrictionshavehalteddevelopments. Thedistinctionbetweendevelopableandnon‐developablelandintheurbanization process(map13.B)isdirectlylinkedtothegrowthtypesindicatedhere.Forexample, areasmarkednon‐developableduetoenvironmentalprotectionmightbeadjacentto butdonotoverlapwithsprawlgrowthareas. The“consolidatedurbanland”regionsfrommap13.Blikelycorrespondtothe“urbanarea”and“infillgrowth”sectionsinthismap. Thesemapscollectivelyprovideamulti‐facetedviewofurbandevelopmentdynamics,understandingthetypesofgrowth,thestagesofdevelopment,andtheclassification oflanduse.Suchdetailedmappingaidsinassessingthebalancebetweendevelopment needsandenvironmentalconservation,crucialforcraftingpoliciesthatguideurbanexpansioninamannerthatisbotheconomicallyviableandecologicallyresponsible.Inessence,thesemapsformafoundationaltoolfordecision‐makerstovisualizeandstrategize Figure 18. Area with significant changes in LST (2003–2023). Figure 17 provides a visual representation of different types of urban growth within the city and surrounding area, based on historical trends and LCZ changes and the findings of Figure 13, that overlap, using GIS and raster databases. The growth directions show the main directions in which the city’s urban expansion is occurring, pointing toward the outskirts or less developed areas of the province along major transport routes. This highlights a linear expansion pattern that can influence transportation and housing policies. The presence of infill growth in central areas suggests attempts to utilize available space efficiently, whereas the sprawl growth around the periphery indicates ongoing expansion, possibly at the cost of greenfield or agricultural land. Evaluating Figure 17 with Figure 13, these main points must be mentioned: • Areas (city) marked as “in process” or “final proposal” in map 12.A align with regions showing infill or sprawl growth in the current map, indicating where future developments are targeted or planned. • Areas marked as “annulled by sentence” could correlate with areas that are not experiencing significant growth in the current map, suggesting that regulatory restrictions have halted developments.
Land 2024,13, 1877 23 of 27 • The distinction between developable and non-developable land in the urbanization process (map 13.B) is directly linked to the growth types indicated here. For example, areas marked non-developable due to environmental protection might be adjacent to but do not overlap with sprawl growth areas. • The “consolidated urban land” regions from map 13.B likely correspond to the “urban area” and “infill growth” sections in this map. These maps collectively provide a multi-faceted view of urban development dynamics, understanding the types of growth, the stages of development, and the classification of land use. Such detailed mapping aids in assessing the balance between development needs and environmental conservation, crucial for crafting policies that guide urban expansion in a manner that is both economically viable and ecologically responsible. In essence, these maps form a foundational tool for decision-makers to visualize and strategize urban growth, ensuring that development is aligned with broader sustainability and resilience goals. 4.2. The Pattern of LST Changes over 20 Years These urbanization trends correlate with changes in the LST and highlight the impact of urban development on local climates. Based on Figure 12 that was obtained from the overlay of the UI, with the aim of making a more precise comparison with LST change patterns, a map of the LST differences over time was created by overlaying the maps of Figure 15 via programming in GEE platform. Figure 18 is a comprehensive map that shows which part of the city is faced with a significant amount of changes in temperature. This does not mean each area has high or low temperature; it identifies the areas that have experienced the most significant temperature changes due to urban development. This map shows the areas that are at risk, sensitive, and vulnerable in the future in terms of temperature changes, particularly the northern and southeastern peripheries, where urban sprawl and industrial growth have led to significant increases in LST. These areas are most at risk due to the lack of green infrastructure and the dominance of impervious surfaces. Seville has experienced an overall increase in the average LST over the past 20 years, partly due to urban expansion and climate change. The city’s growing compact built areas, particularly in high-rise and mid-rise zones, have led to higher temperatures in these areas. The expansion of compact BUAs, classified as LCZ 1, LCZ 2, and LCZ 10 (high-rise, midrise, and industrial zones) now show the highest LST values compared to 2003 (Figure 14). These zones experience hotter surface temperatures, especially during the summer. The expansion of green zones (LCZ A and B) in 2023 and urban greening policies have slightly mitigated LST increases in some areas. In 2003, suburban areas of Seville had lower LSTs compared to the city center. By 2023, due to suburban expansion (LCZ 6 and LCZ 9), these areas also showed a rise in LST. New developments in areas like Sevilla Este have led to a mix of LST trends, where built environments contribute to increased heat, but new green spaces help offset some of the warming. All these indicators, compared to the trend of urban growth in the period before the timeframe studied (Figure 4), demonstrate the direct effects of urban sprawl on changes in LCZs and LST. The correlation between urbanization and LST is evident throughout this study, demonstrating how unplanned urban expansion can lead to undesirable temperature increases and environmental degradation. To achieve sustainable urban growth, these findings underscore the importance of incorporating climate-responsive planning strategies. For example, encouraging infill development and mixed-use zoning can limit urban sprawl while optimizing land use. Additionally, integrating renewable energy sources, improving public transportation systems, increasing the proportion of green buildings, limiting industrial expansion near residential zones, and promoting compact vertical growth instead of horizontal sprawl can significantly reduce Seville’s overall carbon footprint and will help maintain thermal balance. In summary, this paper confirms the results of previous research on urbanization and surface temperature changes. Although no prior study has comprehensively addressed the simultaneous analysis of LST, LCZ, and the urbanization index, this research builds
Land 2024,13, 1877 24 of 27 upon the findings from studies on UHI, LCZ, and urban growth patterns. Given Seville’s specific climatic and temperature conditions, numerous studies have examined its thermal characteristics, and this paper integrates these components while further confirming the trends in the city’s strategic development plans. This research highlights the links between urbanization and surface temperature, advancing the insights from previous studies. 5. Conclusions This study examines the thermal impacts of urbanization on LCZs and LSTs in Seville over the past two decades. The findings emphasize that urban planning must integrate green infrastructure and sustainable land use to mitigate the environmental impacts of urban growth. The key recommendations include the expansion of green spaces, especially in vulnerable northern and southeastern areas, where urban sprawl and industrial development have led to significant temperature increases. Additionally, enforcing stricter zoning laws to limit sprawl and encourage compact, vertical growth will help reduce the long-term effects of UHI. This study also highlights the need for continuous evaluation of urban plans and strategies to understand their long-term impacts on LCZ and LST patterns. The results confirm that changes in land use, particularly in built-up areas (BUA), have led to significant temperature rises, with an average increase of 6.4 ◦ C in these areas. Predictive models for LCZ and LST changes should be developed to help urban planners to implement climate-resilient policies for managing sustainable growth. Seville’s experience provides a model for global urban environmental management, demonstrating how cities can adopt land-use policies that address the negative effects of urbanization and promote sustainable development. Regular assessment of city plans and strategic actions is necessary to understand their impact on the local climate and LST. Enhancing green infrastructure and promoting sustainable development in sensitive areas are essential measures to counteract the negative consequences of rapid urbanization. This research also concludes that, in addition to global climate change, urban development and the shift from compact to sprawling growth patterns have significantly contributed to the rise in surface temperatures in Seville. Therefore, it is crucial to review and adjust the city’s strategic plans to mitigate these environmental impacts. Future research should focus on developing predictive models to assess the direct and indirect effects of LULCCs on LST and guide urban planning in vulnerable areas by adopting climate-resilient infrastructure. The analysis of Seville’s urbanization process underscores the importance of understanding how land use and cover changes can influence local climate dynamics. This study’s results indicate that the increase in temperature within BUAs is more significant than in other land types, with the total temperature rise in Seville being 4.26 ◦C , and 6.4 ◦C specifically in BUAs. This rise in temperature is closely linked to urban development trends and the transition from compact growth to dispersed urban growth. In conclusion, Seville’s urban development trends, while aligned with global climate patterns, also highlight the role of local urban planning decisions in shaping temperature increases. To effectively manage these challenges, cities must regularly revise their strategic plans to ensure they are compatible with environmental sustainability goals. Future research should focus on further refining the models used in this study and applying them to other cities to enhance global urban resilience against climate change. Author Contributions: Conceptualization, N.F. (Nadia Falah), N.F. (Nahid Falah) and J.S.-G.; methodology, N.F. (Nadia Falah) and N.F. (Nahid Falah); software, N.F. (Nadia Falah) and N.F. (Nahid Falah); validation, N.F. (Nadia Falah) and J.S.-G.; formal analysis, N.F. (Nadia Falah) and J.S.-G.; investigation, N.F. (Nadia Falah) and N.F. (Nahid Falah); resources, N.F. (Nadia Falah), N.F. (Nahid Falah) and J.S.-G.; data curation, N.F. (Nadia Falah) and J.S.-G.; writing—original draft preparation N.F. (Nadia Falah), N.F. (Nahid Falah) and J.S.-G.; writing—review and editing, N.F. (Nadia Falah) and J.S.-G.; visualisation, N.F. (Nadia Falah) and N.F. (Nahid Falah); supervision, J.S.-G. All authors have read and agreed to the published version of the manuscript.
Land 2024,13, 1877 25 of 27 Funding: This research received no external funding. Data Availability Statement: The original contributions presented in this study are included in the article; further inquiries can be directed to the corresponding author. Conflicts of Interest: The authors declare no conflicts of interest. Abbreviations LST Land surface temperature LCZ Local climate zones LULC Land use and land cover UI Urbanization index RS Remote sensing GEE Google Earth Engine GIS Geographic information system ENVI Environmental visual imaging software BUA Built-up area WB Water body A/V/F Agricultural/vegetation/forest BS Barren soil MLC Maximum likelihood classifier SVM Support vector machine GWR Geographically weighted regression NDVI Normalized difference vegetation index NDBI Normalized difference built-up index ANN Artificial neural network TIR Thermal infrared ROI Region of interest TA Training area WUDAPT World Urban Database and Access Portal Tools CAGR Compound annual growth rate SPSS Statistical Package for the Social Sciences UHIE Urban heat island effect USGS United States Geological Survey References 1. Almulhim, A.I.; Bibri, S.E.; Sharifi, A.; Ahmad, S.; Almatar, K.M. Emerging Trends and Knowledge Structures of Urbanization and Environmental Sustainability: A Regional Perspective. Sustainability 2022,14, 13195. [CrossRef] 2. United Nations World Urbanization Prospects—Population Division—United Nations. Available online: https://population.un. org/wup/Country-Profiles/ (accessed on 4 December 2023). 3. Yan, H.; Liu, Z. A New Perspective on the Evaluation of Urbanization Sustainability: Urban Health Examination. Sustainability 2023,15, 9338. [CrossRef] 4. Zheng, Y.; Tang, L.; Wang, H. An Improved Approach for Monitoring Urban Built-up Areas by Combining NPP-VIIRS Nighttime Light, NDVI, NDWI, and NDBI. J. Clean. Prod. 2021,328, 129488. [CrossRef] 5. Lopez-Carreiro, I.; Monzon, A. Evaluating Sustainability and Innovation of Mobility Patterns in Spanish Cities. Analysis by Size and Urban Typology. Sustain. Cities Soc. 2018,38, 684–696. [CrossRef] 6. World Bank. Urban Development Overview. Available online: https://www.worldbank.org/en/topic/urbandevelopment/ overview (accessed on 5 December 2023). 7. Feng, Y.; Wang, R.; Tong, X.; Zhai, S. Comparison of Change and Static State as the Dependent Variable for Modeling Urban Growth. Geocarto Int. 2021,37, 6975–6998. [CrossRef] 8. Gago, E.J.; Berrizbeitia, S.E.; Torres, R.P.; Muneer, T. Effect of Land Use/Cover Changes on Urban Cool Island Phenomenon in Seville, Spain. Energies 2020,13, 3040. [CrossRef] 9. López-Casado, D.; Fernández-Salinas, V. The Expression of Illegal Urbanism in the Urban Morphology and Landscape: The Case of the Metropolitan Area of Seville (Spain). Land 2023,12, 2108. [CrossRef] 10. Hewitt, R.J.; Shadman Roodposhti, M.; Bryan, B.A. There’s No Best Model! Addressing Limitations of Land-Use Scenario Modelling through Multi-Model Ensembles. Int. J. Geogr. Inf. Sci. 2022,36, 2352–2385. [CrossRef] 11. Liu, Y.; Hu, Y.; Long, S.; Liu, L.; Liu, X. Analysis of the Effectiveness of Urban Land-Use-Change Models Based on the Measurement of Spatio-Temporal, Dynamic Urban Growth: A Cellular Automata Case Study. Sustainability 2017,9, 796. [CrossRef]