The Governance Process and the Influence on Heat Islands in the City of Quevedo, Coastal Ecuador

Abstract

This article addresses the study of the governance process and the influence of urban heat islands in the city of Quevedo on the coast of Ecuador, and thus contributes to the production of technical and scientific information with a view to their mitigation. To identify the UHI pattern and visualize the temperature distribution on the soil surface, light intensity patterns on the soil surface are identified by the digital processing of the Landsat 7 ETM image. The NDVI, NDSI, and SAVI indices were also calculated, and the AQI was subsequently obtained using a weighted numerical cross-tabulation. The results show that the NDVI and SAVI indicators are correlated with each other and present a strong and positive classification with the neighborhoods and special areas in which there is a high proportion of vegetation, while the NSI and SAVI indicators are positively correlated with the areas. in which there is a greater proportion of built-up areas and roads. From a comprehensive analysis of the reviewed indicators, the authors derived an environmental quality index that explains the beneficial effects of vegetation and negatively explains the detrimental effects of a city covered in cement.

1. Introduction

In the midst of the ongoing climate crisis with corresponding global warming, the world’s metropolises are increasingly affected by the Urban Heat Island (UHI) effect, which describes the phenomenon of very high temperatures in urban areas compared with adjacent rural areas [1,2,3]. The UHI effect may result in urban temperatures of up to 4 °C higher during the day and 2.5 °C higher at nighttime [4,5,6]. The UHI effect is exacerbated by changes in land cover and the depletion of vegetation in urban areas [7,8,9,10]. This leads the built environment to absorb and retain heat, while anthropogenic activities, such as industrial processes and car use, generate waste heat, contributing additionally to the formation of UHIs [11]. Heat islands increase the vulnerability of urban areas and their inhabitants [12], damage urban environments by increasing health problems and mortality rates [13,14,15,16], worsen weather conditions, cause economic hardship, and lead to the degradation of urban infrastructure [17,18,19].

The implications of the UHI effect are largely linked to the objectives of planning livable and resilient cities [4,20,21,22]. As cities warm and experience extreme warming events, their vulnerability and hence their capacity to respond to these challenges and shocks will increase [21]. Recent studies highlight disparities in heat vulnerability, links between urban heat and other environmental outcomes [23], and the effects of planning interventions on urban heat [24]. Governments and planners increasingly understand the important role that governance plays in addressing heat in their communities [25]. Many policies and development decisions taken by local authorities have a significant impact on the formation, extent, or expansion of urban heat islands [21]. In addition to climate change, the planning and design of urban areas often make them warmer than the surrounding rural and natural areas due to the UHI effect [4].

Very few cities, such as London in England and Stuttgart in Germany, have adopted effective mitigation and sustainable management practices for UHI [26]. The practices and efforts undertaken have resulted in negligible changes [21], preventing the realization of truly climate-resilient cities [27]. However, several studies on UHI have demonstrated the potential benefits of various mitigation strategies, including urban governance and planning [26]. The development of decentralized governance and management models, enabling the formulation of common objectives and their implementation in territories [28], with tools and instruments for the implementation of public policies [29], is a viable technical approach that contributes to the mitigation of UHI [24].

Nowadays, there is an abundance of data, knowledge, and tools that help to manage decisions that transgress the urban climate [17]. Nonetheless, climate issues have not been given sufficient priority in urban planners’ decision-making so far [30]. This suggests a very wide disciplinary gap between planners trained in the humanities and fulfilling a socio-political role [31], and scientists whose interest is intimately linked to quantitative data [32]. Furthermore, the lack of integration of the physical and social sciences suggests the need for new transdisciplinary approaches to better plan, govern, and sustain cities [23]. Urban warming concerns many stakeholders, not just urban planners and designers. As a complex and multifaceted problem that contributes to a range of negative health, social [33], environmental, and economic outcomes [21], there is a need to examine the role of urban governance and planning, reflecting and responding to the complexity, uncertainty, and scale of future climate impacts [1]. Addressing the effects of heat island phenomena will require a greater emphasis on heat governance, which involves a set of actors, strategies, processes, and institutions that mitigate and manage the risks associated with urban heat [25]. Advancing urban heat governance will require interdisciplinary approaches involving urban planners, emergency managers, and public health professionals, as well as integration into comprehensive plans, risk mitigation plans, sustainability plans, heat action plans, and other policy and management documents [32,34]. This will also require large-scale co-ordination, including both short- as well as long-term approaches to effective UHI mitigation and management [23].

Heat mitigation strategies reduce heat in the built environment, including general changes in land use, urban design at the neighborhood- and site-scales, urban greenery, and reducing waste heat [25]. Heat management strategies address both chronic heat risks and extreme heat events, such as emergency preparedness, public health activities, reducing personal exposure to heat, and energy accessibility, affordability, and reliability [32]. Heat mitigation is largely the domain of the planning and design disciplines, while heat management is the domain of the public health and emergency management disciplines [35].

Recent advances in UHI mitigation through technological and community approaches are becoming a long-term sustainable alternative, which aims to reduce the environmental, social, and economic effects derived from the UHI phenomenon. This is how the adoption of domestic energy management systems has recently gained notoriety to substantially reduce energy consumption and carbon emissions [36], based on the proliferation of smart appliances and advanced communication infrastructure [37]; the emergence of energy communities from the intelligent integration of prosumers that produce and consume energy [38,39,40]; smart community energy storage [41]; and the development of multicarrier energy systems based on modern generation, conversion, and storage devices [42].

The heat island phenomenon, as already mentioned, requires a multi-level governance model in which national, sub-national, and local governments work in a coordinated way to combat it [43]. In the Latin American context, climate change is not yet high on local and national agendas, and decision-makers at the national level have not yet succeeded in putting multi-level governance into practice for the development of climate policies [44]. In the case of Ecuador, although there is a process of decentralization in accordance with the legal framework that allows local governments to organize themselves and dictate their own rules in the exercise of their exclusive and concurrent competences, and although these levels of government have developed their own initiatives in the face of climate change, considering their territorial realities, with which they wish to contribute to national policy, they continue to be subordinate to the planning and control of the central state [25,45].

Therefore, in Ecuador, several studies have been performed on heat islands, particularly in hot zones. Many cities on the Ecuadorian coast, such as Guayaquil, Esmeraldas, and Manta, have been affected by climate change [46], as they have experienced an increase in rainfall and annual temperatures, due to densification, the blocking of breezes by tall buildings and, above all, the lack of vegetation [47,48,49].

The town of Quevedo, located in the north of the province of Los Ríos in coastal Ecuador, has undergone accelerated urban expansion [28], characterized by a change in the morphology of the territory, mainly to the north and east, with growth along the three existing trunk roads [29]. This form of expansion has led to the modification of many natural areas covered in vegetation in order to consolidate urban land and make way for new settlements [50], particularly in the case of plots that continue to extend into the rural areas of the territory, resulting in variations in the surface temperature of the soil [51]. This situation has its origins in unsustainable territorial planning, due in part to the weakening of local governance, the lack of control, and the politicization of public administration.

Improving interdisciplinary co-operation and action requires a better understanding of urban heat planning and governance, where knowledge remains limited [52]. Indeed, published research on heat islands continues to increase, with more studies in the field of heat mapping and modelling, but very little focusing on planning and governance processes [23,26,53,54]. It also highlights the need for cities to take measures to combat urban warming, reinforced by the growing literature on extreme heat in large cities [55]. Subsequently, the present research, therefore, focuses on investigating the governance process and its influence on heat islands in the city of Quevedo, drawing on a wide range of knowledge and perspectives to develop more effective and sustainable solutions to the complex challenges of urban heat islands.

2. Materials and Methods

2.1. Location of the Study Area

The town of Quevedo, the capital of the canton of the same name, is located in the northern part of the province of Los Ríos, at co-ordinates 1°1′35.61′ South and 79°27′59.71′ West, as demonstrated in Figure 1, with an approximate surface area of 36.28 km². With 206,008 inhabitants (INEC, 2022), it is the most populous city in the province of Los Ríos.

2.2. Geometric and Atmospheric Correction and Digital Image Processing

The satellite image was obtained from the University of Maryland server, with a pixel size of 30 m for bands 1 to 5 and 7, 60 m for bands 6.1 and 6.2 (thermal), and 15 m for band 8 (panchromatic) [56,57,58]. Before obtaining the environmental thematic indicators needed to construct the Independent Component Analysis (ICA), the Enhanced Thematic Mapper Plus (ETM+) image was subjected to three fundamental processes, being geometric correction, conversion of the digital levels (ND) of the non-thermal spectral bands into reflectivity values, and conversion of the ND of the thermal bands into surface temperature [59,60,61,62,63,64]. The image was subjected to atmospheric correction to eliminate the effect of electromagnetic radiation scattering caused by gases and particles in suspension in the atmosphere, so that the variations in the models would be independent of atmospheric conditions [65,66,67,68,69]. For this purpose, reflectivity was calculated using Equation (1) [70,71]. For geometric correction, the image was projected using the UTM system, zone 18 North, WGS84 [72]. Vector data of urban streets and roads were used to make the geometric correction of the image, in order to allow the correct superimposition with the map [72,73,74]. This procedure included the use of a high resolution orthophoto of the city of Quevedo to obtain the land cover and at the same time characterize with a high level of detail the urban heat islands. In order to superimpose it correctly on the map of the city’s districts, the positions of the image were rectified using the map of streets and roads as a reference. Hereby, 18 control points and 4 verification points were used, which allowed it possible to obtain, by means of a second-order polynomial function, a root mean square error (RMSE) of 0.43 and 0.57 pixels in X and Y co-ordinates, respectively, an acceptable error for processes of this type [73]. To assign the ND corresponding to each pixel at the new position, the nearest neighbor method was used, which guarantees the smallest transformation of the original NDs [75]. The NDs of the non-thermal spectral bands were converted to spectral radiance at the sensor and then to reflectivity, following the steps outlined in Landsat Project Science Office (2008) [76]. All digital image processing was performed in IDRISI Selva 17.0 software [77,78].

$${\rho }_k={\displaystyle \frac{K\pi \left(L_{sen, k}-L_{a, k}\right)}{{\tau }_{k, o}\left(E_{o,k} {cos\theta }_i {\tau }_{k,i}+E_{d,k}\right)}}$$

(1)

where:

${\rho }_k=$ Reflectivity in the k-band.

K = Earth-Sun distance in astronomical units (1 AU = 1.49598- 108 Km, varies over the year between 0.983 and 1.017 AUs).

$L_{sen,k}=$ Spectral radiance received by the sensor in the k band (W m⁻² sr⁻¹ μm⁻¹).

$L_{a,k}=$ Atmospheric radiance due to k-band scattering (W m⁻² sr⁻¹ μm⁻¹).

${\tau }_{k,o}=$ Atmospheric transmittance for upward flow in the k-band.

$E_{o,k}=$ Solar irradiance at the top of the atmosphere in the k band (W m⁻² μm⁻¹).

${cos\theta }_i=$ Cosine of the zenith angle of the incident flux (complementary to the solar elevation angle).

${\tau }_{k,i}=$ Atmospheric transmissivity for downward flow in the k-band.

$E_{d,k}=$ Atmospheric diffuse irradiance due to k-band scattering (W m⁻² µm⁻¹).

The correction method applied was the default downward atmosphere transmissivity, which uses the following values in Equation (1):

$L_{a,k}=$ $a_{o,k}$ + $a_{1,k}$ NDmín. The NDmí of the histogram in the k-band.

${\tau }_{k,o}=$ cos${\theta }_o$ as ${\theta }_o$ = o; ${\tau }_{k,o}=1$.

${\tau }_{k,i}=$ 0.70, 0.78, 0.85, 0.91 for TM1, TM2, TM3 and TM4, respectively. For TM5 and TM7 the values of 0.95 and 0.97 are taken [79,80].

$E_{d,k}=$ o. Ignore diffuse irradiance.

This method uses the minimum ND of each band as a measure of the radiance due to atmospheric scattering, while the atmospheric transmissivity for the downwelling flux is extrapolated from physical experiments carried out on real atmospheres without cloud cover [81,82].

2.3. Obtaining Environmental Indicators

To obtain the surface temperature, the spectral radiance at the sensor, the thermal band was transformed into the brightness temperature of the satellite [83,84] using Equation (2).

$$TL=K2/1n\left(\right(K1/L\lambda )+1$$

(2)

where:

TL: brightness temperature in degrees Kelvin, for radiance L;

K1: calibration constant 1 e n W/(m² × sr × um);

K2: calibration constant 2 in degrees Kelvin (dimensionless);

Lλ: spectral radiance of the sensor.

This brightness temperature is that of a black body. It was therefore necessary to introduce the emissivity of the earth’s surface, considering the types of use, to obtain the kinetic surface temperature [85]. This conversion was performed using Equation (3).

$$TS=TL/1+(\lambda x TL/p)x ln\epsilon$$

(3)

where:

TS: surface temperature corrected by emissivity;

TL: satellite brightness temperature;

λ: average wavelength of the thermal band considered;

p = h × c/σ, (1.438 × 10⁻² mK), where:

h: Planck’s constant (6.626 × 10⁻³⁴ Js),

c: speed of light;

σ: Boltzmann’s constant (1.38 × 10⁻²³ J/K) [86];

ε: emissivity of the surface.

Equation (4) was used to calculate the emissivity of the coatings [87].

$$E=fv\epsilon v+\left(1-fv\right) \epsilon S$$

(4)

where:

εv and εS: emissivities of vegetation and ground cover, respectively; assumed to be 0.985 and 0.978; assumed to be 0.985 and 0.978, respectively;

fv: vegetation fraction, obtained with Equation (5) using Normalized Difference Vegetation Index (NDVI), one of the most widely used vegetation indices one of the most widely used vegetation indices [88,89,90].

$$fv=1-(1–(NDVIMAX-NDVI)/(NDVIMAX-NDVIMIN\left)\right)a$$

(5)

where:

NDVImax: maximum NDVI value for the greenest vegetation;

NDVImin: minimum NDVI present in bare soil.

Although there are negative values in the NDVI image (especially at the water and cloud surface), a minimum value of 0 was considered. The exponent a is a value that depends on the orientation of the plant leaves, and a value of 0.6 was taken [91]. Finally, the emissivity-corrected surface temperature (ST), in degrees Kelvin, was converted to degrees Celsius by subtracting the value 273 from all the pixels in the ST image. In addition to surface temperature, the other environmental indicators (NDVI, NDSI and SAVI) were obtained from the ETM+ image transformation. NDVI has been recognized as one of the most useful indicators for studying the characteristics of the terrestrial biosphere and its dynamics, at global, regional and local levels [92]. It is obtained from Equation (6).

$$NDVI=(NIR-R)/(NIR+R)$$

(6)

where NIR and R represent the near-infrared (NIR) and red (R) reflectivity of the Landsat ETM+ image, respectively. The Normalized Difference Soil Index (NDSI), highlights built-up areas and bare ground, as these surfaces are more reflective in the far infrared than in the near infrared [93]. It was obtained using Equation (7).

$$NDSI=(SWIR-NIR)/(SWIR+NIR)$$

(7)

Finally, we used the Soil-Adjusted Vegetation Index (SAVI) index, which is suitable for areas with low vegetation cover and, consequently, a high percentage of soil reflectance [94]. It was obtained by applying Equation (8).

$$SAVI=\left[\right(NIR-R)/(NIR+R+L\left)\right] (1+L)$$

(8)

L being a vegetation adjustment coefficient, equivalent to 0.5, recommended for intermediate densities [92]. High NDVI and SAVI values per pixel indicate a high presence of vegetation, which normally corresponds to a better environmental quality, given its beneficial effects on water and air purification, noise reduction, the shade and landscape value it provides, and the transfer to the atmosphere, by transpiration, of latent heat stored in the soil, leading to a reduction in surface temperature, among other things. The higher the INS values, the greater the proportion of built-up areas or asphalt (impermeable areas). Since in many cities the neighborhood is the basic spatial unit for data collection, management and planning, an average neighborhood value for each of the environmental indicators was obtained by overlaying the neighborhood map with NDVI, NDSI and SAVI raster data, using the Idrisi Selva program [95].

2.4. Pearson Correlation Analysis and Principal Component Analysis

The Pearson correlation coefficient is an indicator used to quantitatively describe the strength and direction of the relationship between two normally distributed quantitative variables and helps to determine the tendency of two variables to go together, which is also called covariance [96]. This coefficient correlation coefficient considers the covariance (sum of xy products) in the numerator and the root of the product of the sums of squares of both variables in the denominator [97], according to Equation (9).

$$r={\displaystyle \frac{\sum xy}{\sqrt{\left(\sum x^2\right)\left(\sum y^2\right)}}}$$

(9)

r = product-moment coefficient of linear correlation

$x=X-\overline{X}$; $y=Y-\overline{Y}$

Principal component analysis (PCA) is a standard statistical method for the analysis of multivariate data that allows the number of dimensions to be reduced while retaining as much of the variation in the data as possible [98,99]. The visualization and statistical analysis of these new variables, the principal components, can help to find similarities and differences between samples [100]. The procedure to follow is described below:

(a)

Standardization

First, the dataset was standardized so that each variable has a mean of 0 and a standard deviation of 1, as shown in Equation (10).

$$Z={\displaystyle \frac{X-\mu }{\sigma }}$$

(10)

Here,

μ = is the mean of the independent characteristics;

μ = {μ₁, μ₂, …, μ_m}

$\sigma$= is the standard deviation of independent characteristics;

$\sigma$ = {${\sigma }_1$, ${\sigma }_2$, …, ${\sigma }_m$}

(b)

Calculation of the covariance matrix

Then, with the original standardized variables: $Var {(x}_i)=1$ para $i=1, \dots p$, the correlation matrix was calculated, according to Equation (11):

$$Cor\left(y_j, x_i\right){\displaystyle \frac{{\lambda }_j{a}_{ij}}{1 \ast {\lambda }_j^{\frac{1}{2}}}}={\lambda }_j^{\frac{1}{2}}{a}_{ij}$$

(11)

Thus, the matrix of correlations between γ and x, according to Equation (12):

$$Cor\left(y,x\right)={\Lambda }^{1/2} A^{’}=F^{’}$$

(12)

Thus, the factorial matrix also measured the correlations between the original standardized variables and the new factors.

(c)

Calculation of the eigenvalues and eigenvectors of the correlation matrix to identify the principal components.

Let A be a square matrix nXn and X a nonzero vector, as shown in Equation (13).

$$AX=\lambda X$$

(13)

For some scalar values λ. Then λ became known as the eigenvalue of matrix A and X was identified as the eigenvector of matrix A for the corresponding eigenvalue.

It can also be written as shown in Equation (14):

$$\begin{gathered} AX-\lambda X=0 \\ \left(A-\lambda I\right)X=0 \end{gathered}$$

(14)

Wherein, the identity matrix may be of the same form as the matrix A. And the above conditions will be true only if (A − λI) are not invertible (that is, singular matrix). This is specified in Equation (15):

$$\mid A-\lambda I\mid =0$$

(15)

From the above equation, it was possible to find the eigenvalues λ and, therefore, the corresponding eigenvector was found using the equation AX = λX.

Each of the procedures used for the estimation of the Pearson correlation coefficient and the principal component analysis were developed within the statistical software STATGRAPHICS Centurion XIX.

3. Results

3.1. Thematic Environmental Indicators for the Construction of the Urban Environmental Index

The aerial orthophotography of the city of Quevedo at a resolution of 0.50 m was constructed using remote sensing and remote detection techniques from the capture of a mosaic of quadrants in the Google Earth Pro 7.3 platform. From this orthophoto, the road network and block coverage was digitized using ArcGIS 10.4 software, resulting in a vector layer in shapefile format of the urban area, illustrated in Figure 2. This road and mosaic coverage allowed geometric correction and georeferencing of the Landsat ETM 7 satellite image.

From the thermal bands 6.1 and 6.2 of the Landsat 7 ETM satellite image, a map of the ground surface temperature in degrees Celsius was obtained (°C), as shown in Figure 3a. The ground surface temperature of the city of Quevedo ranges between 17.0879 °C and 27.8451 °C for the entire urban territory. The highest temperature concentration is observed in the central area of the city, belonging to the following sectors: Playa Grande, Nicolás Infante Díaz, San Camilo, Chapulos, San Cristóbal, Venus of the Quevedo river, 24 of May, and in a generalized manner along Walter Andrade Fajardo avenue, from Special Forces Group No. 26 Cenepa to the Loreto neighborhood. These sectors are characterized by the presence of very dense urbanization and a lack of vegetation, with asphalted streets, service infrastructure, houses, buildings, and other structural elements that capture heat. The urban areas with the lowest surface soil temperatures correspond to those sectors that are far from the city center, especially in the southern area, which includes the following neighborhoods: 7 of October, Guayacán, Viva Alfaro, El Desquite, via El Empalme, Puente Sur, and Ruta Ecológica and La Salud. These neighborhoods, despite being urban, show a higher level of tree and shrub vegetation, in addition to the presence of uninhabited lots used for subsistence agriculture, which together constitute important heat dissipating elements. In this zone of the city, the urban limit is narrower and the occupation of the land by residences is reduced to very dispersed neighborhoods, with a scarce presence of houses and paved streets. This could be appreciated by superimposing the orthophoto, the raster layer of surface soil temperature and the vector layer of the road and paved network of the city of Quevedo, as shown in Figure 3b.

The levels of the Normalized Difference Vegetation Index (NDVI) range between −0.42029 and 0.741935, distributed throughout the urban area of Quevedo. The highest NDVI levels are concentrated in the downtown area, involving the following sectors: avenue 7 de Octubre, Avenue Bolívar, Avenue June Guzmán, Playa Grande, Nicolás Infante Díaz, San Camilo, San Cristóbal, Venus of Quevedo river, 24 of May, Progreso, Lotization Américas, Los Geranios, and via Valencia and Promejoras. In these sectors, urban regeneration and territorial planning developed by the local municipality has not contemplated the implementation of urban trees, so there is little or no vegetation, with paved and cement structures predominating. The sectors of the urban periphery show higher NDVI values, including: via Buena Fe, Ciudad del Norte, Special Forces Group No. 26 Cenepa, via La China, via El Empalme, Ruta de Río, via Cañalito, and Ruta Ecológica and Nuevo Amanecer. These areas mark the limits of the city’s urban development, with a scarce presence of urbanized area, predominantly extensive areas of vegetation cover, identified as primary and secondary forests, agroforestry crops, and areas of perennial agricultural crops: African palm, banana, and cocoa, as shown in Figure 4.

The Normalized Difference Soil Index (NDSI) ranges from 0 to −1 for the entire urban area of Quevedo. The central sector of the city has higher levels of NDSI due to the fact that the soil in this area is mostly covered with concrete and asphalt, by both buildings and houses, as well as roads (bare soil), is characterized by a lower reflectivity and, therefore, it concentrates more heat, resulting in a higher soil temperature. Within this zone, the following sectors were identified: downtown, San Camilo, Nicolas Infante Díaz, Ciudadela El Guayacán, San José Sur, Venus del río Quevedo, Lotización Jardines del Este, Lotización Girasol, Promejoras, Viva Alfaro, Progreso, and Isla del río. In the peripheral area of the city, especially in the south of the territory, lower levels of NDSI are observed, especially in the south and east of the city. This indicates that the ground in this area is covered by denser vegetation and has a higher reflectivity that affects the non-presence of heat islands, as illustrated in Figure 5.

The Soil-Adjusted Vegetation Index (SAVI) showed values between −0.625899 and 1.10932 for the urban area of Quevedo. This index allowed correcting the NDVI raster layer for the influences of soil brightness because this area is characterized by areas with low vegetation cover and reflectance. This shows that SAVI was higher in the area of the periphery of Quevedo, comprised by the following sectors: Baldramina, Ruta del Río, Ruta Ecológica, via El Empalme, via Buena Fe, and in the area of influence of the lateral passage (ring road). On the other hand, the central sector of the city of Quevedo has the lowest levels of SAVI, as a result of low vegetation cover and low reflectance, characterized by being moderately exposed soil, as illustrated in Figure 6.

3.2. Urban Environmental Index

The Urban Environmental Quality Index (UEQI) shows that the two urban areas lateral to the main river of Quevedo have a deficient and regular UEQI. The sectors identified in this area are: downtown, San Camilo, via Valencia, Nicolás Infante Díaz, Venus del río Quevedo, Viva Alfaro, 7 de octubre, 24 de mayo, and El Guayacán. The peripheral zones of the city presented a good and very good UEQI, and within these areas the following sectors stand out: Ruta del río, via Cañalito, Ruta Ecológica, via El Empalme, Ciudad del Norte, Lotización Nuevo Amanecer, El Desquite, La Virginia, and the area of influence of the lateral passage (ring road), as shown in Figure 7.

3.3. Principal Component Analysis and Pearson Correlation for the Validation of Relationships Between Urban Heat Islands, Governance Factors and Environmental Indicators

Pearson correlation coefficients measure the strength of the linear relationship between variables. The relationship between the variables SAVI-NDVI, NSDI-UHI, GOV-NSDI reveal positive correlations, with Pearson coefficient values close to 1.0 (0.73, 0.83), corresponding to the deep red cells, as shown in Figure 8. The relationship between SAVI-NDVI showed a Pearson coefficient of 0.83, that is, the highest in the correlogram, demonstrating a very strong linear relationship between both variables. In contrast, the relationships between the variables SAVI-UHI, SAVI-NSDI showed negative correlations close to −1.0 (−0.61, −0.69), identified in Figure 8 with a strong blue color tone. The variables UHISAVI obtained the most representative negative Pearson coefficient of the analysis −0.69, thus revealing that both variables are strongly related in a reverse direction.

The p values test the statistical significance of the estimated correlations for each combination of variables. The combination of variables with p values > 0.05 do not present statistically significant correlations at the 95% confidence level. The pairs of variables UHI-NSDI, UHI-SAVI, NDVI-SAVI, NSDI-SAVI, NSDI-SAVI, NSDI-GOV show p values < 0.05, that is, evidence significant statistical correlations, according to

Table 1. p-values for urban heat island (UHI) variables, governance factors (GOV), and environmental indicators (NDVI, NSDI, SAVI).

	UHI	NDVI	NSDI	SAVI	GOV
UHI	-	0.3065	0.0103	0.0189	0.1682
NDVI	0.3065	-	0.1354	0.0014	0.2629
NSDI	0.0103	0.1354	-	0.0466	0.0111
SAVI	0.0189	0.0014	0.0466	-	0.1865
GOV	0.1682	0.2629	0.0111	0.1865	-

.

The principal component analysis allowed us to obtain a reduced number of linear combinations of the five variables that explain most of the variability of the data. In this case, component 1 was extracted, since only this component had an eigenvalue equal to or greater than 1.0. This component represents 65.698% of the variability of the original data, as shown in

Table 2. Principal components with their eigenvalues and percentage of variation.

Component Number	Eigenvalue	Percent of Variance	Cumulative Percentage
1	3.28488	65.698	65.698
2	0.881067	17.621	83.319
3	0.591002	11.820	95.139
4	0.187954	3.759	98.898
5	0.0550931	1.102	100.000

.

In the sedimentation plot in Figure 9 a red line is marked with the value of the minimum eigenvalue, above which the principal components are chosen. The Kaiser criterion with correlations indicates that we have to retain the first components with eigenvalues greater than 1. Based on this criterion the number of principal components chosen is equal to 1, which describes 65.698% of the total variation of the data as seen in Table 2. We can then substitute the 5 original variables by the first principal component, which are uncorrelated as seen above.

Component 1 is made up of the variables GOV, NSDI and UHI; the latter two variables are very close to each other, i.e., they are associated. The NSDI presents large positive influences within this component. On the other hand, component 2 groups the variables NDVI and SAVI, which show a considerable proximity between them. In addition, it is observed that the NDVI variable contributes with positive influences within component 2, as shown in Figure 10.

4. Discussion

The variables that best explain the thermal distribution of the city of Quevedo correspond from the highest to the lowest to distance to wet sources, population density, topography of the city site, and NDVI. This is similar to the results the city of Zaragoza in Spain. However, the proximity to wetland sources is very significant in the case of the city of Quevedo, which has an important surface watercourse such as the Quevedo River, being an important element to consider in the urban and environmental planning of this city [101]. Three thematic indicators (NDVI, NDSI, and SAVI) estimated from a Landsat ETM+ satellite image were derived from this study. The results are consistent in showing that the NDVI and SAVI indicators provide important quantitative information about neighborhoods and special areas where there is a high proportion of vegetation, while the highly correlated NDSI and SAVI are positively correlated with areas that have a higher proportion of built area and roads [102]. These behaviors coincide with the reference literature, to the extent that they quantitatively explain the spatial relationship between the estimated variables and the way urban land is used in the city of Quevedo. From the integrated analysis of the indicators reviewed, an environmental quality index was derived that explains the beneficial effects of vegetation and negatively explains the adverse effects of the cement-covered city [103].

In Latin America, many countries have experienced, in the last five decades, a process of intensive urbanization that has resulted in the modification of urban space [12]. This process of urban alteration has caused significant negative impacts on the environment, mainly affecting the energy exchange processes between the earth’s surface and the atmosphere, altering the surface and subsurface hydrological system, decreasing air and water quality, and changing meso- and microclimatic conditions [1]. The factors that alter urban environmental quality and the social and economic development of cities are largely associated with the way social agents and their activities occupy and use territorial space, a situation that creates new challenges for urban planning authorities and entities.

Urban environmental quality is understood as a set of interrelated human and environmental factors among which we can highlight mainly heat islands, type, density and layout of buildings, population density, presence of green areas, and air and water quality, which have a favorable or unfavorable impact on the lives of citizens [44]. Decision-making by Quevedo’s city administrators requires the availability of data that combine in the same analytical and spatial structure indicators derived from remote sensing images, which, in contrast to data acquired through censuses, allows for the construction of more efficient and lower-cost environmental indicator systems.

The process of urbanization artificially modifies climatic conditions by raising temperatures, and decreasing air humidity and wind speed, which together favor the consolidation of UHI, which can be of two types, the UHI which corresponds to the higher temperature recorded by the air layer covering the city as a consequence of the transmission towards it of the heat accumulated by the structures and also of bodies that compose it such as the roofs and walls of buildings, streets and avenues, uncovered, industrial sites, and car parks [26].

All these surfaces are characterized by the fact that they are built with materials of the urban heat island and the UHI, on the contrary, is related to the high emission temperatures that reach the different structures and urban bodies and that are directly captured by infrared sensors [32], such as those available on earth observation satellites. The particularity of UHI is that they are not directly subject to the thermal compensations made by airflows from warmer to colder surfaces that regulate the air temperature of cities, so that more pronounced thermal features are to be expected. However, the spatial correlations between surface and air heat islands are generally high [46], therefore, it may be assumed that the spatial patterns of both are similar to a high degree, although not necessarily in terms of recorded temperature levels.

5. Conclusions

An important element of the results is the low significance found between soil temperatures and vegetation expressed as NDVI. This demonstrates the obvious lack of urban parks in the city of Quevedo, which increases the importance of planning in the city’s internal waste sites, generating parks and green areas with a predominance of wide canopy tree structures, to reduce the effects of UHI, both in terms of intensity and magnitude.

In this context, the need to determine the urban heat islands (UHI) and the urban environmental quality index (IQU) of the city of Quevedo is highlighted, given that to date this information is not available due to the high acquisition costs involved in obtaining it through census methods, with only historical environmental data and indicators that are not always adequate because they have been obtained by different methods at inappropriate scales.

In many cities in Ecuador, and especially Quevedo, urban heat islands are currently a growing problem, which is influenced by many factors, one of which is the increase in built area. The issue of the urban heat island is important because its effects are beginning to be felt, including natural degradation, health and economic losses from excessive energy use.

The results of this study can be replicated in other cities on the coast of Ecuador, such as Guayaquil and Durán, which face serious problems due to the phenomenon of urban heat islands. In these cities urban infrastructure is expanding rapidly, resulting in the change of natural land surface into impermeable and urbanized regions. Changes include replacing vegetative cover and soil with concrete and asphalt surfaces, replacing rural structures with urban composite structures, and agricultural activities in rural areas with large-scale industrial and commercial activities in urban areas.