Sage Journals: Discover world-class research

Abstract

The relationship between urban heat island (UHI) and land use/land cover (LULC), and local climate zone (LCZ) is apparent and takes rising attention in the current literature. This study aimed to investigate the relationship between meteorological data collected from 30 stations between 2016 and 2022 and Istanbul’s LULCs and LCZs. Several notable findings were uncovered, providing light on the UHI phenomenon and its consequences for the city’s characteristics. The stations in urban areas (typology A) had higher temperatures than stations in rural/suburban (typology B) and forested landscape (typology C). Those yearly values were ∼1°C for monthly mean temperatures and ∼1.5°C for monthly minimum temperatures. Moreover, urban areas possessed +4 and +2 hot days (35°C and above) for typologies B and C, respectively. Another remarkable result was that stations situated close to water surfaces exhibited a lower tendency to exceed temperatures of 35°C. Furthermore, built-type LCZs wind velocity achieved a lower value than land cover type, and humidity in typology A was 5% and 10% less than the typologies B and C, respectively. Consequently, the southern part of Istanbul emerged as the most vulnerable location to the UHI phenomena, suffering greater temperatures.

Keywords

Urban heat island Land use/land cover Local climate zones Meteorological data Istanbul

Introduction

Climate change is a remarkable reality of our time. The theory of systematic scientific evaluations conducted since the 1970s has progressed from theory to established acceptance. The most recent and concrete pieces of evidence of climate change are a 1.1°C increase in global mean temperature from 1880 to 2023, a carbon dioxide ratio in the atmosphere reaching 422 ppm, a 12.6% decrease in glacier area since 1979 and a 10 cm rise in sea level.¹ These alterations influence both natural and manmade processes, including changes in precipitation patterns, extreme weather occurrences, phenological shifts and varied habitats such as forest and rural and urban environment.^2,3

Nowadays, the world’s population is about 8 billion, and is predicted to increase to 8.5 billion by 2030, 9.7 billion by 2050 and 10.4 billion by 2100. Cities now accommodate 55% of the world’s population. By 2050, it is expected to grow to 68%.⁴ Cities, covering approximately 3% of the Earth’s surface, account for 75% of global energy consumption by their inhabitants.⁵ Due to various difficulties such as rising population, rapid urbanization and excessive pressure on natural habitats, many NGOs and public sectors offer a variety of conservation and implementation suggestions to reduce the impact of climate change by addressing the complex interplay between population dynamics, urbanization and environmental sustainability.^6–8

Many problems are currently awaiting resolution in cities. One of the most obvious issues is urban heat island (UHI). The UHI is an increased temperature occurrence in a city compared to its immediate surroundings,⁹ and is caused by a variety of reasons. Contrary to common perception, it cannot only be attributed to a lack of green space in the city. The excessive use of heat-absorbing materials, dense population and traffic, constructing a built environment without considering local climate characteristics, and inappropriate land use decisions increase heat retention and create localized hotspots. This may result in raised temperatures, heat-related health risks, increased energy consumption for cooling and compromised air quality, and may be specified as components that cause the establishment of an urban heat island (UHI).^10,11

Another serious problem to date has been the change in land use/land cover (LULC).¹² Climate change, LULC and energy are all interlinked.¹³ LULC change are recognized as important interconnections that are impacted by human activities like population growth, rapid urbanization and increasing pressure on ecosystem resources, in addition to changing climatic conditions.¹⁴ The interplay between LULC and climate change is not given enough attention, even though effective interventions are necessary in land-based ecosystems.¹⁵ Changes in the LULC have an impact on the climate, either directly or indirectly.^16,17 There are many studies examining the relationship between LULC and climate, revealing the importance of this issue at various scales such as global, regional and local.^18–20 Additionally, some governments have been focusing on LULC changes and their consequences on climate.^21,22

Urbanization has a strong correlation with LULC dynamics, which in turn leads to changes in UHI intensity. Each LULC type effects UHI differently as the environment changes; therefore, understanding this impact and link is crucial.²³ As such, the density and occurrence of UHI are highly related to LULC patterns,^24–26 which have a significant impact on land-surface climatic interactions, altering heat exchange, moisture, trace gas fluxes and albedo.²⁷

The geographical monitoring and mapping of variation in LULC assist local governments and decision-makers in developing comprehensive and sustainable plans for future urbanization and city sustainability.²⁸ Through this, dense meteorological data has taken on an important role.^29,30 For example, Okumus and Terzi (2021)³¹ determined the surface temperatures in urban areas to be approximately 4°C higher than in rural areas of Istanbul. Likewise, Şekertekin and Marangoz (2019)³² found a similar temperature difference for the city of Zonguldak. Moreover, Toy and Yilmaz (2010)³³ studied Erzincan City, which is a well-planned, unindustrialized and medium-sized Turkish town. The urban environment is 2°C warmer than the rural for minimum temperatures. Similar results have been observed for the above-mentioned studies, although the study areas are different.

Climate parameters are critical for cities, especially when UHI is present. Understanding and monitoring the interaction between climate parameters and LULC is critical for covering the scope and impact of UHI in cities. LULC changes can impact local climate variables such as temperature patterns, wind velocity and humidity. Wind is critical for cooling the city and introducing fresh air into the city voids. Urban environments tend to cut the wind velocity and to be drier than their surroundings.^34,35 Humidity is another significant parameter that regulates perceived temperature, meaning high humidity levels can make the air feel warmer than the actual temperature. To deeply investigate the effect of LULC on climate, Stewart and Oke (2012)³⁶ introduced a detailed classification method called local climate zone (LCZ). The LCZ offers researchers to classify the land with 17 different definitions and provides a protocol to compare the land use type all over the world.

UHI is one of the most important problems for metropolitan cities like Istanbul³⁷ and the importance of the LULC/LCZ and their effect on climate is opulent. Therefore, this research investigated the relationship between climate data and LULC by looking at a variety of climatic factors such as monthly mean temperature (°C), monthly minimum mean temperature (°C), monthly maximum mean temperature (°C), frequency of days exceeding 35°C, monthly mean wind velocity (m/s) and monthly mean humidity (%). Through this approach, the aim was to shed light on the influence of LULC on climatic conditions by evaluating specific factors, notably the UHI effect and its interaction with LULC and LCZ. This investigation addressed the following questions: (1) How does LULC influence meteorological data? and (2) do LCZ types exhibit variations amongst themselves? Through our study, we contribute to a deeper understanding of how UHI variability interacts with LULC and LCZ dynamics, giving useful insights for landscape and urban planning, climate mitigation methods and sustainable urban development.

Material

The study area covered Istanbul Province (Figure 1). Under the material title, the physical, demographic and climatic characteristics of the study area are described.

Figure 1.

Location of the study area (a and b), selected meteorological stations (c) and Istanbul topography (d).

Istanbul

Istanbul is a metropolitan city with a population of approximately 16 million and consists of 39 districts. Istanbul borders the Black Sea to the north and the Marmara Sea to the south. It has a unique geographical position with the strait connecting these two seas (Figure 1(c)). The altitude of the city varies between 0 and 537 m. Its highest point is Aydos Hill, which is adjacent to Kartal, Pendik, Sultanbeyli and Sancaktepe districts on the Anatolian side of Istanbul. Aydos Hill is followed by two hills with an altitude of 478 m on the Şile–Kocaeli Border and 442 m on the Çatalca–Tekirdağ Border. In addition, Alemdağ is located within the borders of Çekmeköy at 442 m, and Kayışdağı is located within the borders of Ataşehir with an altitude of 438 m, which are amongst the high hills in Istanbul (Figure 1(d)).

Istanbul has been under the influence of rapid urbanization since 1980. Some of the physical steps that led to increased urbanization are the opening of the Bosphorus Bridge (now the 15 July Sehitler Bridge) in 1973 and the Fatih Sultan Mehmet Bridge in 1988. The planning and positioning of many industrial zones on the peripheries of Istanbul accelerated the migration from Anatolia to Istanbul, which caused a significant change in Istanbul’s LULC and LCZ, and the city expanded rapidly in the east-west direction.³⁸

The forests of Istanbul are amongst the natural areas that are most heavily damaged due to human-based pressures.³⁹ While there are moist-temperate leafy forests in the north of Istanbul, there are maquis and Pinus brutia Ten. forests in the inner and southern parts. This area constitutes 44% of the total area of Istanbul (238,000 ha). The primary tree species in these forests are: Quercus spp., Fagus spp., Carpinus spp. and Castanea spp. The total forest area has decreased from 270 thousand hectares to 238 thousand hectares in the last 50 years. Structural pressures on the forest and wrong urbanization policies are the most important reasons for this negative situation.⁴⁰

In addition, the vertical architectural policy adopted after the 2000s has increased the anthropogenic effects in the urban area. In addition, the construction of Yavuz Sultan Selim Bridge and roads in the north of the province may cause the city to expand towards the north in the near future.^41,42 Due to such effects, Istanbul has become one of the most populated regions in the world. The most populated districts of Istanbul are the southern districts on the coast of the Marmara Sea. The northern parts of the province have forest cover.⁴³ When these main findings are put together, it is very important to understand how the urban areas of Istanbul show the temperature difference according to its LULCs and LCZs. Due to wrong urbanization, most places were formed haphazardly, and this affected the urban climate.

Istanbul climate

Istanbul’s climate is transitional between Mediterranean and Black Sea types with cold and wet in winter and hot and humid in summer.⁴⁴ The southern parts of Istanbul generally represent the Mediterranean climate type and are warmer than other regions. However, due to the winds blowing from the north, the city is cooler and rainier than other Mediterranean regions further south.⁴⁵ In addition, the climate is mostly affected by the Marmara and Black Sea and the presence of the Bosphorus passing through the city.⁴⁶

The annual temperature is about 14.7°C between 1980 and 2022, derived from Sarıyer-Kumköy, Kadıköy-Rıhtım, Şile, Florya and Sarıyer stations, and the mean annual maximum and minimum temperatures are 18.6°C and 11.1°C, respectively. The annual precipitation is about 760 mm, and the northern part receives slightly more precipitation than the southern part. Lastly, the prevailing wind directions of the city are on the northeast and southwest.

Methods

This section consists of two parts: preliminary and main research. Preliminary research includes the processing of raw data, and the main research involved the evaluation of the processed data (Figure 2).

Figure 2.

Method flow chart.

The preliminary part investigated the long-term annual mean temperatures (°C), the annual mean temperatures (°C) between 1980 and 2022, the meteorological stations’ selection procedure and avoiding missing data.

The main part included the extraction and analysis of the data. In this context, between 2016 and 2022, monthly mean temperature (°C), monthly minimum mean temperature (°C), monthly maximum mean temperature (°C), day frequencies with 35°C and above temperature, monthly mean wind velocity (m/s) and monthly mean relative humidity (%) data were examined.

First, statistical procedures were applied. Each data set was explained by descriptive statistical methods. Next, the data were subjected to normality test. If normality test results were positive, then statistical procedures continued with ANOVA and Scheffe multi-comparison tests. If normality was negative, the data was left with descriptive statistics and visualization. The purpose of ANOVA and Scheffe multi-comparison tests was to explore the land use/land cover (LULC) typologies influence the data or not.

Finally, the regression test was applied to correct the latitude–longitude effect depending on the data set to increase the data purity and accuracy. For LULC typology, the concepts of urban and rural may not be sufficient to classify areas alone.¹⁰ For this reason, the station and its surroundings are supported by LCZ typologies. Then the data were evaluated with various visualization techniques.

Long-term annual mean temperature (°C) time series

First, a long-term time series was plotted with data from six different stations over annual mean temperatures (°C). The stations were selected for the criteria of having the oldest data in order to present the long-term data properly. These stations were Atatürk Airport (1980–2022), Florya (1937–2022), Kadıköy-Rıhtım (1929–2022), Sarıyer (1949–2022), Sarıyer-Kumköy (1951–2022) and Şile (1940–2022). Thus, an increasing trend was determined in annual mean temperatures in 1980 and later (Figure 3, left graph). Then, another time series was drawn for the data between 1980 and 2022, and after 2010, it was seen that the stations located in the dense urban texture (Atatürk Airport and Kadıköy-Rıhtım) began to separate from other stations. In other words, the range of 2010–2022 was found to be suitable for evaluations (Figure 3, right graph). After this stage, the selection procedure for meteorological stations was started.

Figure 3.

Graph of long-term mean annual temperature data (°C) (left) and graph of annual mean temperature data (°C) from 1980 and later (right).

The selection procedure of meteorological stations and the determination of the year range

First of all, the number of recording stations was determined by going back to 2022, and as of 2016, 30 stations were found to be recording. This determination revealed the year range, which constitutes the main pillar of the study, as 2016–2022. Statistical procedures were effective in determining the maximum number of stations. These details are explained below.

Data analysis and accuracy

Data analysis and accuracy included several statistical procedures: Shapiro–Wilk normality test, ANOVA test, Scheffe multi-comparison test and linear regression test. This procedure was applied to monthly mean temperature (°C), monthly mean minimum temperature (°C), monthly mean maximum temperature (°C), monthly mean wind velocity (m/s) and monthly mean relative humidity (%).

In statistics, the assumption of normality is important for correctly analyzing data. The assumption of normality is one of the first steps for the validity of hypothesis tests.⁴⁷ First, the Shapiro–Wilk test of normality, which can be applied to observations with a sample size of less than 50, was applied to each data set. With this, we can decide whether the data can be evaluated with a parametric or non-parametric test.⁴⁸

After determining the normality, the relationship between all data from 30 stations and the LULC was examined with a one-way ANOVA (Analysis of Variance) test, which was considered significant if at least one of the variables was different but did not show which two variables were different.⁴⁹ To reveal this detail, Scheffe’s test, which is a multi-comparison test, was performed and the evaluation phase was started. All statistics were calculated by SPPS 22.0 and R Studio v. 4.12.⁵⁰

Determining the spatial land use/land cover and local climate zones of the stations

There are several methods to define LULC. The most well-known is the CORINE categorization form, which is typically referred to as urban, rural and forest. CORINE 2018 Land Cover divides land into three main classifications: urban, rural/suburban and forest patterns. For urban classification, each class has a continuous urban fabric, a discontinuous urban fabric, industrial or commercial units, road and rail networks, associated land, etc. It has detailed sub-classifications such as non-irrigated arable land, continuously irrigated land, complex cultivation patterns, etc., for rural/suburban classification; and broadleaf forest, mixed forest, transitional woodland-shrub, etc., for forest classification.⁵¹

The other well-known classification method is the local climate zone (LCZ) classification. The LCZ provides a basic standardization for UHI studies and allows comparing urban temperature observations in different areas over these standards. LCZ types are divided into 17 distinct classifications which are grouped under two broad categories. The first primary title expresses a type of built-type, which are compact high-rise (1), compact mid-rise (2), compact low-rise (3), open high-rise (4), open mid-rise (5), open low-rise (6), lightweight low-rise (7), large low-rise (8), sparsely built (9) and heavy industry (10), while the second indicates the land cover type, which are dense trees (A), scattered trees (B), bush-scrub (C), low plants (D), bare rock and paved (E), bare soil and sand (F) and water (G).³⁶

CORINE 2018 Land Cover was used for LULC detection of stations⁵¹; nonetheless, we benefited from Google Earth aerial photographs to make LULC clear. As such, we decided on the LULC classification with urban (typology A), rural/suburban (typology B) and forested landscape (typology C). Typology A was for those who remained inside the urban macro-form; typology B was for stations between the urban and forested landscape; and typology C was assigned to the stations inside the forest texture. In this case, seventeen (17) urban, seven (7) rural/suburban, and six (6) forested landscape typologies were determined as LULC typologies.

To analyze the LULC distributions in detail, the most dominant local climate zone (LCZ) was determined with the help of a 500 m × 500 m grid covering the station and its immediate surroundings (Figure 4). Yang et al. (2023)⁵² also used the same measures to assess the LCZ types. World Urban Database and Access Portal Tools (WUDAPT) database was used to determine LCZ types. WUDAPT is an international community-based initiative to obtain and disseminate climate-related data on the physical geographies of cities for modelling and analysis.⁵³ In this study, an LCZ map created for Istanbul with an accuracy of 0.79 was used.⁵⁴ While determining the LCZ types, the most dominant LCZ type in the 500 m × 500 m grid was taken as a basis, and the presence of water in the grid was coded between 0 and 1.0 (zero) indicates no or little effect, and 1 indicates that the presence of water is very effective.

Figure 4.

Local climate zone patterns of selected sites.

In this context, one compact mid-rise built-type, one low-rise built-type, five openly arranged mid-rise built-type, three openly arranged low-rise built-type, six scattered built-type, one heavy industry built-type, two densely wooded land cover type, five sparsely wooded land cover type, four ground cover plants land cover type and 2 bare rock land cover type were determined (Table 1).

Table 1.

Information on selected stations.

Code	Stations	Station numbers	Latitude	Longitude	Altitude (m)	LULC^a	LCZ^b
S1	Arnavutkoy-Terkosbarajı	18,734	41.3364	28.6175	16	3	B
S2	Arnavutköy	18,402	41.2203	28.7075	140	2	D
S3	Atatürk Havalimanı	17,060	40.9819	28.8208	33	1	E
S4	Beykoz	18,396	41.1417	29.0739	5	1	9
S5	Beykoz-Anadolufeneri	18,421	41.2233	29.1658	58	3	B
S6	Büyükçekmece	18,099	41.0453	28.5901	20	2	D
S7	Çekmekoy-Ömerli	18,397	41.0783	29.3256	80	3	A
S8	Eyüp	18,101	41.1028	28.9242	54	3	B
S9	Fatih-Deniz Bilimleri Enstitüsü	17,603	41.0055	28.9601	10	1	3
S10	Florya	17,636	40.9758	28.7865	37	1	9
S11	Güngören-Davutpaşa	17,814	41.0266	28.8853	68	1	5
S12	Istanbul-Bölge	17,064	40.9113	29.1558	16	1	5
S13	Kadıkoy-Rıhtım	17,062	40.9883	29.0191	5	1	6
S14	Kadıköy-Göztepe	17,813	40.9794	29.0571	41	1	6
S15	Pendik-Ömerli	18,736	41.0083	29.3419	70	3	B
S16	Sabiha Gökçen	17,063	40.8977	29.3033	91	1	E
S17	Samandıra-Havaalanı	17,065	40.9866	29.2135	123	1	D
S18	Sancaktepe	18,399	41.0086	29.2822	110	2	D
S19	Sarıyer	17,061	41.1464	29.0502	59	1	9
S20	Sarıyer-Kumkoy	17,059	41.2505	29.0384	38	2	6
S21	Silivri	18,400	41.1846	28.1575	195	2	A
S22	Silivri-Anamendirek	17,438	41.0731	28.2394	9	2	5
S23	Şile	17,610	41.1688	29.6007	83	2	9
S24	Şile-İsaköy<	18,735	41.1083	29.8283	6	3	B
S25	Şişli	18,401	41.0547	28.9683	60	1	2
S26	Tuzla	18,100	40.8278	29.2931	3	1	9
S27	Tuzla Aydınlı Liman Feneri	17,437	40.8514	29.2717	8	1	10
S28	Tuzla İTÜ Güney Mendirek	17,448	40.8128	29.2978	13	1	5
S29	Ümraniye	18,403	41.0293	29.1383	191	1	9
S30	Üsküdar	18,404	41.0322	29.0464	75	1	5

^a1: urban, 2: rural/suburban, 3: forested landscape.

^bLocal climate zone typologies are explained in the Method section.

Data interpretation and visualization

All data was visualized by Quantum Geographic Information System (QGIS) for maps and R Studio for the figures of time series. On QGIS, inverse distance weighted (IDW) interpolation was used in a monthly scale, and a topographical correction on monthly mean temperature (°C), monthly minimum mean temperature (°C) and monthly maximum mean temperature (°C) to correctly visualize the data. By doing this, a 0.65°C constant value was used for every 100 m altitude.

Findings

Under this heading, monthly mean temperature (°C), monthly minimum mean temperature (°C), monthly maximum mean temperature (°C), temperature frequencies of 35°C and above, monthly mean wind velocity (m/s) and monthly mean relative humidity (%) data were evaluated.

Monthly mean temperatures (°C)

The mean monthly temperatures (°C) for the years 2016–2022 were as follows: 6.20, 7.73, 8.73, 12.59, 17.51, 22.18, 24.38, 24.98, 21.40, 16.61, 12.98 and 8.98. January recorded the lowest temperature, while August registered the highest.

Shapiro–Wilk’s test of normality shows that mean temperature data for all months showed normal distribution (Table 2). This shows the suitability of analysis of all tests with parametric statistical methods. According to the one-way ANOVA result, there is a statistical difference between the LULC and mean temperature values of all months at the 95% confidence level (p < 0.05) (Table 3). This difference means that at least one LULC data is statistically different from the others. However, since the ANOVA test alone did not show which two LULCs differed, the assistance of the Scheffe multi-comparison test was needed.

Table 2.

Results of Shapiro–Wilk test on monthly mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.506	0.158	0.370	0.305	0.495	0.527	0.746	0.677	0.522	0.448	0.207	0.259

Table 3.

Results of ANOVA test on monthly mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.001	0.001	0.000	0.000	0.000	0.000	0.000	0.002	0.001	0.000	0.000	0.001

Table 4 shows the main outputs of monthly mean temperatures (°C) that there are statistically significant differences between typologies A and B in all months except August and September. Temperature values ranged between 0.77°C and 1.11°C. When we looked at the differences between typologies A and C, there were statistically significant values for all months. The differences in these values varied between 0.60°C and 1.38°C. In other words, typology A was warmer than typologies B and C, respectively, in all months. In addition, no statistical difference was found between typologies B and C.

Table 4.

Results of Scheffe’s multi-comparison test on monthly mean temperature (°C).

LULC		Months
LULC		Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Urban (A)	Rural/suburban (B)	0.04	0.003	0.001	0.001	0.001	0.004	0.029	-	-	0.022	0.011	0.004
Urban (A)	Forest (C)	0.029	0.036	0.005	0.002	0.002	0.001	0.001	0.004	0.002	0.001	0.002	0.01
Average differences (°C)^a	A-B	1.0	0.78	0.85	1.11	1.05	0.96	0.86	-	-	0.77	0.89	1.0
Average differences (°C)^a	A-C	0.8	0.60	0.77	1.22	1.07	1.13	1.38	1.15	1.20	1.18	1.13	1.0

^aOnly statistically significant values shown: A-C and A-B.

Since there are significant distances between station locations, linear regression was established for the effect of latitude and longitude on monthly mean temperatures, and the effect of both latitude and longitude was minimal (Tables 5 and 6). Despite this, a regression equation was established on the data, and the effect of latitude and longitude was eliminated, and statistical procedures were repeated.

Table 5.

Linear regression results of latitude and monthly mean temperatures (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Multiple R-squared	0.217	0.237	0.327	0.430	0.369	0.269	0.178	0.09	0.082	0.116	0.137	0.162
Adjusted R-squared	0.189	0.210	0.303	0.414	0.346	0.243	0.149	0.057	0.049	0.084	0.106	0.132
p-value	0.009	0.006	0.001	0.001	0.001	0.003	0.02	0.108	0.123	0.065	0.044	0.027

Table 6.

Linear regression results of longitude and monthly mean temperatures (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Multiple R-squared	0.056	0.091	0.025	0.002	0.001	0.003	0.053	0.047	0.06	0.03	0.008	0.011
Adjusted R-squared	0.022	0.058	−0.001	−0.033	−0.035	−0.032	0.019	0.013	0.026	−0.003	−0.026	−0.02
p-value	0.20	0.10	0.40	0.80	0.91	0.76	0.22	0.24	0.19	0.35	0.62	0.57

Table 7 shows that all values were normally distributed for normalized monthly mean temperature values (°C). According to Table 8, significant differences were found in all months. Table 9 explains which two LULC typologies differ. Accordingly, there were 0.38°C and 0.56°C temperature differences between typologies A and B, and 0.38°C to 0.84°C between typologies A and C. This secondary control case proved once again that typology A is warmer than other LULC typologies. In addition, no statistical difference was found between typologies B and C.

Table 7.

Results of Shapiro–Wilk test on normalized monthly mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.626	0.676	0.826	0.270	0.350	0.175	0.063	0.109	0.155	0.052	0.059	0.626

Table 8.

Results of ANOVA test on normalized monthly mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.002	0.002	0.001	0.001	0.001	0.001	0.003	0.005	0.006	0.002	0.001	0.001

Table 9.

Results of Scheffe’s multi-comparison test on normalized monthly mean temperature (°C).

LULC		Months
LULC		Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Urban (A)	Rural/suburban (B)	0.006	0.006	0.007	0.032	0.025	0.040	-	-	-	-	-	0.007
Urban (A)	Forest (C)	0.029	-	0.006	0.002	0.002	0.002	0.003	0.005	0.006	0.003	0.002	0.006
Average differences (°C)^a	A-B	0.46	0.38	0.42	0.56	0.50	0.42	-	-	-	-	-	0.40
Average differences (°C)^a	A-C	0.38	-	0.45	0.84	0.73	0.66	0.74	0.51	0.52	0.51	0.50	0.43

^aOnly statistically significant values shown: A-C and A-B.

Another type of classification that defines study areas was LCZ. Monthly mean temperature data (°C) showed that there was a temperature difference between 0.7°C and 1.13°C for LCZs with built-type (Table 10). As for normalized data, differences were found between 0.12°C and 0.30°C (Table 11). In both cases, built-type LCZ was found to be warmer than land cover type.

Table 10.

Monthly mean temperature data (°C) of LCZ typologies.

LCZ type	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2	6.84	8.33	9.41	13.61	18.46	23.04	25.14	25.66	22.29	17.47	13.71	9.56
3	6.74	8.19	9.3	13.4	18.3	23.07	25.37	25.77	22.07	17.2	13.57	9.67
5	6.62	8.02	8.97	13.00	17.77	22.41	24.83	25.39	21.91	17.12	13.51	9.48
6	6.92	8.31	9.30	13.13	18.09	22.74	25.05	25.57	21.86	17.27	13.68	9.98
9	6.16	7.61	8.61	12.39	17.39	22.16	24.40	25.03	21.40	16.61	12.98	8.96
10	7.43	8.47	9.83	12.97	17.76	22.2	24.37	24.93	21.84	17.21	14.04	9.59
A	4.97	6.93	8.04	12.09	16.76	21.14	23.26	23.88	20.31	15.61	11.82	7.56
B	5.83	7.42	8.30	11.81	16.88	21.49	23.50	24.24	20.70	15.92	12.33	8.47
D	5.54	7.29	8.31	12.28	17.24	22.08	24.09	24.70	21.00	16.04	12.37	8.34
E	6.49	7.99	9.13	13.31	18.17	22.89	25.20	25.69	22.05	17.18	13.40	9.25
BT-LCT^a	1.08	0.75	0.79	0.71	0.70	0.70	0.85	0.76	0.88	0.96	1.10	1.13

^aBT: built-type, LCT: land cover type.

Table 11.

Normalized monthly mean temperature data (°C) of LCZ typologies.

LCZ type	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2	6.17	7.69	8.72	12.6	17.52	22.19	24.42	25.02	21.44	16.64	13	8.96
3	6.29	7.79	8.86	12.84	17.72	22.38	24.62	25.15	21.57	16.77	13.13	9.09
5	6.37	7.85	8.95	13.02	17.88	22.52	24.78	25.26	21.69	16.89	13.24	9.18
6	6.14	7.68	8.67	12.47	17.41	22.09	24.29	24.92	21.34	16.55	12.91	8.91
9	6.23	7.76	8.73	12.52	17.45	22.12	24.26	24.89	21.30	16.53	12.92	8.97
10	6.76	8.18	9.27	13.37	18.2	22.77	24.89	25.3	21.7	16.97	13.4	9.48
A	5.93	7.49	8.51	12.38	17.32	22.04	24.36	25.00	21.44	16.59	12.90	8.77
B	5.96	7.55	8.44	12.00	17.00	21.72	23.84	24.60	21.01	16.24	12.62	8.69
D	6.14	7.67	8.69	12.57	17.49	22.17	24.40	25.01	21.43	16.62	12.98	8.94
E	6.49	7.95	9.03	13.09	17.94	22.57	24.77	25.24	21.66	16.88	13.26	9.26
BT-LCT^a	0.20	0.16	0.20	0.30	0.26	0.22	0.20	0.13	0.12	0.14	0.16	0.18

^aBT: built-type, LCT: land cover type.

January was the coldest month at 6.20°C. In January, the higher parts of the city with the natural formation were important cold area clusters. These examples were seen in Silivri and Çatalca districts (left-rectangular), Aydos and Kayışdağı regions (right-rectangular in Figure 5). In August, the mean temperature was 24.98°C. Similar to January, there were cool islands in the same areas. However, temperature differences were narrow compared to January values (Figure 6).

Figure 5.

Interpolation map of monthly mean temperatures (°C) of January by the inverse weighted distance. Topography correction has been made on the maps and the raw data has been reduced by 0.65°C for every 100 m of elevation. 25 m resolution EarthExplorer satellite DEM data was used for topography correction.

Figure 6.

Interpolation map of monthly mean temperatures (°C) of August by the inverse weighted distance. Topography correction has been made on the maps and the raw data has been reduced by 0.65°C for every 100 m of elevation. 25 m resolution EarthExplorer satellite DEM data was used for topography correction.

Another component of the study was the existing water surface at the station and its immediate surroundings (station-centred 500 × 500 m grid). According to Table 12, stations with no or little water surface effect (code 0) were between 0.16°C and 0.45°C cooler in relatively cold months. This was the opposite with 0.17°C to 0.57°C for April, May, June and July. In other words, while the water surface had a cooling effect in relatively hot months, it possessed a heat-trapping feature in cool and cold months (Table 12).

Table 12.

The effect of water surfaces on the station’s monthly mean temperatures (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
0	6.05	7.67	8.73	12.78	17.65	22.30	24.43	24.98	21.34	16.52	12.83	8.83
1	6.50	7.84	8.74	12.21	17.23	21.94	24.26	24.98	21.50	16.80	13.28	9.26
Difference	−0.45	−0.17	−0.01	0.57	0.42	0.36	0.17	−0.01	−0.16	−0.28	−0.45	−0.42

0: no or little water surface effect, 1: high water surface effect.

Monthly minimum mean temperatures (°C)

The monthly minimum mean temperatures for the specified period were as follows: 3.37, 4.57, 5.25, 8.19, 13.27, 18.04, 20.12, 21.16, 17.47, 13.15, 9.92 and 6.25. According to the monthly minimum mean temperature (°C) between 2016 and 2022, the coldest month and the warmest month are January at 3.37°C and August at 21.16°C. Similar to monthly mean temperature (°C), urbanized area of Istanbul was much warmer than the other typologies (Figures 7 and 8).

Figure 7.

Interpolation map of monthly minimum mean temperatures (°C) of January by the inverse weighted distance. Topography correction has been made on the maps and the raw data has been reduced by 0.65°C for every 100 m of elevation. 25 m resolution EarthExplorer satellite DEM data was used for topography correction.

Figure 8.

Interpolation map of monthly minimum mean temperatures (°C) of August by the inverse weighted distance. Topography correction has been made on the maps and the raw data has been reduced by 0.65°C for every 100 m of elevation. 25 m resolution EarthExplorer satellite DEM data was used for topography correction.

When the normal distribution conditions of the data were examined, all values showed normal distribution except for April, October and November months (Table 13). Therefore, those months were not interpreted. According to the ANOVA test, significant differences were found in all months for monthly mean minimum temperatures (°C) (Table 14).

Table 13.

Results of Shapiro–Wilk test on monthly minimum mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.174	0.106	0.356	0.027	0.102	0.138	0.147	0.106	0.060	0.048	0.046	0.445

Table 14.

Results of ANOVA test on monthly minimum mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.001	0.001	0.001	-	0.001	0.001	0.002	0.009	0.002	-	-	0.001

The main differences emerged between typologies A and B, and typologies A and C. These monthly minimum mean temperature differences ranged from 1.10 to 1.40°C for typologies A and B, and 1.35 to 1.90°C for typologies A and C. In addition, no statistical difference was found between typologies B and C (Table 15). These findings showed parallelism with the monthly mean temperature values.

Table 15.

Results of Scheffe’s multi-comparison test on monthly minimum mean temperature (°C).

LULC		Months
LULC		Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Urban (A)	Rural/suburban (B)	0.002	0.005	0.007	-	0.002	0.003	0.044	-	0.043	-	-	0.005
Urban (A)	Forested L. (C)	0.002	0.001	0.002	-	0.001	0.001	0.008	0.020	0.005	-	-	0.001
Average differences (°C)^a	A-B	1.40	1.10	1.16	-	1.35	1.35	1.26	-	1.26	-	-	1.37
Average differences (°C)^a	A-C	1.52	1.35	1.44	-	1.50	1.54	1.72	1.51	1.81	-	-	1.90

^aOnly statistically significant values shown: A-C and A-B.

Furthermore, linear regression was established to normalize the data. The effect of both latitude and longitude was minimal (Tables 16 and 17). Despite this, a regression equation was established on the data, and the effect of latitude and longitude was eliminated, and statistical procedures were repeated.

Table 16.

Linear regression results of latitude and monthly minimum mean temperatures (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Multiple R-squared	0.0006	0.0001	0.006	-	0.016	0.02	0.08	0.1	-	-	0.05	0.01
Adjusted R-squared	−0.035	−0.036	−0.03	-	−0.01	−0.02	0.05	0.06	-	-	0.01	−0.02
p-value	0.9	0.99	0.68	-	0.51	0.41	0.13	0.1	-	-	0.24	0.59

Table 17.

Linear regression results of longitude and monthly minimum mean temperatures (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Multiple R-squared	0.14	0.15	0.14	-	0.15	0.13	0.03	0.0008	-	-	0.04	0.12
Adjusted R-squared	0.11	0.12	0.1	-	0.12	0.1	−0.008	−0.0035	-	-	0.003	0.09
p-value	0.04	0.036	0.045	-	0.033	0.052	0.39	0.99	-	-	0.3	0.057

Table 18 shows the data normality for normalized monthly minimum mean temperature (°C). As indicated before, the months of April, October and November were not interpreted. In addition, the values of June obtained no normality. Therefore, the data for June were not evaluated as well.

Table 18.

Results of Shapiro–Wilk test on normalized monthly minimum mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.585	0.426	0.062	-	0.052	0.046	0.539	0.099	0.590	-	-	0.057

Table 19 shows some significant differences in the normalized monthly minimum mean temperature values (°C) for 7 months, which were January, February, March, May, July, September and December. Table 20 explains which two LULC typologies these differences were between. Accordingly, two values were obtained in January and February with 0.45°C and 0.38°C, respectively. Additionally, there were 0.54°C to 0.73°C differences between typologies A and C. This secondary control case proved once again that the urban environment is warmer than typology C. In addition, no statistical difference was found between the typologies B and C.

Table 19.

Results of ANOVA test on normalized monthly minimum mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.001	0.001	0.001	-	0.001	0.002	-	-	0.015	0.022	0.010	0.001

Table 20.

Results of Scheffe’s multi-comparison test on normalized monthly minimum mean temperature (°C).

LULC		Months
LULC		Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Urban (A)	Rural/suburban (B)	0.015	0.021	-	-	-	-	-	-	-	-	-	-
Urban (A)	Forest (C)	0.002	0.002	0.002	-	0.002	-	0.024	-	0.015	-	-	0.002
Average differences (°C)^a	A-B	0.45	0.38	-	-	-	-	-	-	-	-	-	-
Average differences (°C)^a	A-C	0.59	0.54	0.62	-	0.72	-	0.62	-	0.57	-	-	0.73

^aOnly statistically significant values shown: A-C and A-B.

Moreover, there was a temperature difference between 1.28°C and 1.59°C for built-type LCZ (Table 21). As for normalized data, the range was found to be 0.10°C and 0.23°C (Table 22). In both cases, built-type LCZs were found to be warmer than the land cover type. When the water surface effect was examined, stations whose no or little water surface effect (code 0) in all months were 0.24°C to 0.60°C cooler than the high water surface effect (code 1) (Table 23).

Table 21.

Monthly mean minimum temperature data (°C) of LCZ typologies.

LCZ type	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2	4.34	5.49	6.26	-	14.27	19.04	21.23	22.19	18.6	-	-	7.2
3	4.44	5.51	6.27	-	14.21	18.99	21.09	21.96	18.29	-	-	7.29
5	3.95	5.13	5.79	-	13.88	18.64	20.85	21.81	18.28	-	-	6.90
6	4.49	5.37	6.04	-	13.95	18.75	21.03	22.17	18.56	-	-	7.55
9	3.43	4.59	5.24	-	13.24	18.02	20.15	21.24	17.54	-	-	6.38
10	4.88	5.83	6.95	-	14.66	19.26	21.07	21.61	18.81	-	-	7.52
A	1.82	3.48	4.14	-	12.06	16.54	18.40	19.55	15.83	-	-	4.66
B	2.56	3.77	4.44	-	12.51	17.27	19.28	20.37	16.54	-	-	5.12
D	2.31	3.73	4.38	-	12.37	17.28	19.16	20.34	16.35	-	-	5.25
E	3.97	5.13	5.80	-	14.09	18.92	21.00	21.82	18.12	-	-	7.05
BT-LCT^a	1.59	1.29	1.41	-	1.28	1.28	1.45	1.31	1.64	-	-	1.62

^aBT: built-type, LCT: land cover type.

Table 22.

Normalized monthly mean minimum temperature data (°C) of LCZ typologies.

LCZ Type	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2	3.37	4.58	5.27	-	13.3	18.08	20.19	-	17.53	-	-	6.28
3	3.54	4.73	5.44	-	13.5	18.27	20.34	-	17.67	-	-	6.48
5	3.67	4.84	5.58	-	13.66	18.42	20.48	-	17.80	-	-	6.64
6	3.29	4.50	5.17	-	13.18	17.95	20.05	-	17.41	-	-	6.15
9	3.34	4.53	5.19	-	13.18	17.94	19.96	-	17.34	-	-	6.16
10	3.97	5.09	5.78	-	13.84	18.57	20.35	-	17.72	-	-	6.84
A	3.18	4.42	5.15	-	13.19	17.99	20.29	-	17.59	-	-	6.15
B	2.96	4.20	4.81	-	12.75	17.53	19.64	-	17.03	-	-	5.72
D	3.34	4.55	5.25	-	13.28	18.06	20.19	-	17.53	-	-	6.26
E	3.74	4.89	5.61	-	13.67	18.42	20.39	-	17.73	-	-	6.66
BT-LCT^a	0.23	0.20	0.20	-	0.22	0.21	0.10	-	0.11	-	-	0.23

^aBT: built-type, LCT: land cover type.

Table 23.

The effect of water surfaces on the station’s monthly minimum mean temperatures (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
0	3.21	4.46	5.13	-	13.19	17.94	19.93	20.96	17.27	-	-	6.10
1	3.69	4.79	5.49	-	13.43	18.24	20.48	21.55	17.87	-	-	6.54
Difference	−0.48	−0.32	−0.36	-	−0.24	−0.30	−0.55	−0.59	−0.60	-	-	−0.43

0: no or little water surface effect, 1: high water surface effect.

Monthly maximum mean temperatures (°C)

Throughout the given period, the highest monthly mean temperatures (°C) were recorded as follows: 10.51, 12.79, 14.31, 19.26, 23.81, 27.95, 29.88, 30.23, 27.16, 22.07, 17.78 and 13.26. According to the monthly maximum mean temperature (°C), the coldest and the warmest month were January and August, respectively. Figure 9 is the map of the monthly maximum mean temperature data for all months. Accordingly, some stations obtained more monthly maximum mean temperature values (°C) than others. These were Fatih-Deniz Bil. Enst. (S9), Kadıköy-Rıhtım (S13), Sabiha Gökçen (S16), Samandıra (S17) and Sarıyer-Kumköy (S20) stations (Figure 9).

Figure 9.

Interpolation map of monthly maximum mean temperatures (°C) of all months by the inverse weighted distance. Topography correction has been made on the maps and the raw data has been reduced by 0.65°C for every 100 m of elevation. 25 m resolution EarthExplorer satellite DEM data was used for topography correction.

Shapiro–Wilk normality test was performed on monthly maximum mean temperature (°C) data and the data did not show normal distribution (Table 24). At this stage, ANOVA, Scheffe multiple comparison test and linear regression analysis could not be performed. Descriptive statistical methods were conducted and maps were interpreted. Table 25 shows the water surface effect on monthly maximum mean temperature (°C). As such, values were higher in cases where there was no or little water surface effect (except for January and December). This can be explained by the cooling effect of water.

Table 24.

Results of Shapiro−Wilk test on monthly maximum mean temperature (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.000	0.000	0.000	0.000	0.000	0.000	0.003	0.014	0.000	0.000	0.000	0.000

Table 25.

The effect of water surfaces on the station’s monthly maximum mean temperatures (°C).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
0	10.41	12.83	14.41	19.60	24.16	28.30	30.28	30.61	27.44	22.25	17.84	13.13
1	10.71	12.72	14.12	18.56	23.11	27.24	29.08	29.47	26.60	21.70	17.67	13.51
Difference	−0.30	0.11	0.29	1.04	1.05	1.06	1.19	1.14	0.84	0.55	0.17	−0.37

0: no or little water surface effect, 1: high water surface effect.

Frequency of days 35°C and above for selected stations

Increasing temperatures can cause heat-related diseases,^55,56 and the values of 30°C–35°C and above are critical.^57,58 For this reason, we defined 35°C as the threshold point. The days above this threshold value are expressed with the concept of a hot day.^59–61

Table 26 demonstrates 35°C and above values. Typologies A, B and C were 11.16 (212/17), 9.17 (55/7) and 6.67 (40/6) day frequencies, respectively. In other words, typology A has an average of +2 hot-day frequencies compared to typology B stations, and +4 hot-day frequencies compared to typology C. These results coincide with the monthly mean and monthly minimum mean temperature (°C).

Table 26.

Years and total values of 35°C and above.

Codes	Stations	2016	2017	2018	2019	2020	2021	2022	Total values
S1	Arnavutkoy-Terkosbarajı	0	0	0	0	0	2	0	2
S2	Arnavutköy	1	5	1	0	0	3	1	11
S3	Atatürk Havalimanı	2	8	2	0	0	6	0	18
S4	Beykoz	0	3	0	0	0	3	0	6
S5	Beykoz-Anadolufeneri	0	0	0	0	0	0	0	0
S6	Büyükçekmece	1	6	0	0	0	8	0	15
S7	Çekmekoy-Ömerli	0	5	0	0	2	5	0	12
S8	Eyüp	1	6	0	1	1	8	0	17
S9	Fatih-Deniz Bilimleri Enstitüsü	0	4	0	0	0	11	1	16
S10	Florya	1	5	1	0	0	6	0	13
S11	Güngören-Davutpaşa	0	4	0	0	1	6	0	11
S12	Istanbul-Bölge	2	10	1	1	1	9	0	24
S13	Kadıkoy-Rıhtım	0	5	0	0	0	3	0	8
S14	Kadıköy-Göztepe	2	3	1	0	0	4	0	10
S15	Pendik-Ömerli	0	2	0	0	2	3	0	7
S16	Sabiha Gökçen	1	4	0	0	0	7	0	12
S17	Samandıra-Havaalanı	0	4	0	0	0	7	0	11
S18	Sancaktepe	0	6	0	0	1	8	0	15
S19	Sarıyer	0	2	0	0	0	1	0	3
S20	Sarıyer-Kumkoy	0	2	0	0	0	1	0	3
S21	Silivri	0	0	0	0	0	4	0	4
S22	Silivri-Anamendirek	0	0	0	0	0	0	0	0
S23	Şile	0	1	1	0	2	1	2	7
S24	Şile-İsaköy	0	0	0	1	0	1	0	2
S25	Şişli	0	4	0	0	0	10	0	14
S26	Tuzla	5	7	8	11	3	12	2	48
S27	Tuzla Aydınlı Liman Feneri	0	1	0	0	0	0	0	1
S28	Tuzla İTÜ Güney Mendirek	0	0	0	0	0	0	0	0
S29	Ümraniye	0	5	1	0	0	3	0	9
S30	Üsküdar	0	4	0	0	0	4	0	8

In addition, the tendency to exceed 35°C was lower at stations close to the sea. This highlights the fact that the areas on the south coast were warmer despite being cooled by the effects of the sea. However, this effect was not seen in the interior of the south coast due to the built environment that cuts the wind.

Between 2016 and 2022, the frequency of 35°C and above values at Tuzla station (S26) was determined as 48. These extreme heat measurements are remarkable. Five of the values were in 2016, 7 in 2017, 8 in 2018, 11 in 2019, 3 in 2020, 12 in 2021 and 2 in 2022. In addition, temperature values of 35°C and above were not detected at stations S5 (Beykoz Anadolu-Feneri), S22 (Silivri Ana-Mendirek) and S28 (Tuzla ITU Güney Mendirek) (Table 25). Those were located next to the sea.

S28 was Tuzla İTÜ Güney Mendirek station, which was very close to S26. However, the extreme values of 35°C and above detected in S28 were not detected. The reason why no extreme temperature was observed can be explained by the cooling effect of the winds coming from the sea. Lastly, July and August were the months with the highest temperatures of 35°C and above.

Monthly mean wind velocity (m/s)

The monthly mean wind velocity values (m/s) for the years 2016–2022 were as follows: 3.38, 3.39, 3.19, 2.92, 2.80, 2.76, 3.20, 3.39, 3.12, 2.83, 2.98 and 3.24. The three stations with the least wind, in general, were Istanbul-Bölge (1.46 m/s), Kadıköy-Göztepe (1.61 m/s) and Beykoz (2 m/s). The stations with the highest monthly mean wind velocity (m/s) were Beykoz-Anadolufeneri (4.84 m/s), Silivri Ana-Mendirek (4.57 m/s) and Atatürk Airport (4.54 m/s).

Wind data for Arnavutköy-Terkos, Pendik-Ömerli and Şile-İsaköy stations were not recorded. Therefore, those could not be included in the sampling. Data from statistical analyses were evaluated on 27 station data. Mean values were calculated by taking these considerations into account.

Statistical procedures were also performed on the monthly mean wind velocity. The data were normally distributed (Table 27). However, no significant value was observed in the ANOVA test results (Table 28). Therefore, this part was only evaluated by descriptive statistics and visualization.

Table 27.

Results of Shapiro–Wilk test on monthly mean wind velocity (m/s).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.610	0.848	0.714	0.194	0.120	0.170	0.171	0.071	0.221	0.339	0.512	0.683

Table 28.

Results of ANOVA test on monthly mean wind velocity (m/s).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.610	0.848	0.714	0.194	0.120	0.170	0.171	0.071	0.221	0.339	0.512	0.683

The highest and lowest windy months were February and June with 3.39 (m/s) and 2.76 (m/s), respectively. In addition, the wind values of the stations where the water surface effect is high were found to be higher with the range of 0.11 to 1.19 m/s (Table 29). This is explained by the fact that the water surface has a low roughness length and generally does not contain any obstacles to block the wind. Moreover, the opposite situation was observed in January and December. This issue can be shaped by which direction the wind blows and the elements that can create a wind curtain in that direction.

Table 29.

The effect of water surfaces on the station’s monthly mean wind velocity (m/s).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
0	3.08	3.16	3.02	2.81	2.69	2.66	3.08	3.24	2.92	2.59	2.73	2.97
1	3.00	4.11	3.95	3.62	3.22	3.09	3.04	3.58	3.83	3.64	3.43	3.58
Difference	0.08	−0.96	−0.92	−0.81	−0.53	−0.43	0.04	−0.34	−0.91	−1.05	−0.70	−0.61

0: no or little water surface effect, 1: high water surface effect.

Monthly mean relative humidity (%)

The monthly mean relative humidity percentages for the period under consideration were as follows: 80.16, 79.71, 77.27, 72.70, 75.06, 75.96, 73.16, 74.69, 72.80, 78.76, 80.14 and 80.20. The highest humid month (%) was December with 80.2%, and the lowest humidity was in April with 72.7% for the years 2016–2022.

Table 30 shows that monthly mean relative humidity (%) achieved a normal distribution in March, June, July and August. Therefore, additional tests were only interpreted over the relevant months (Table 31). For March and June, there was a humidity difference of 4.05% and 5.83% between typologies A and B, and this result was statistically significant. In other words, the humidity content of typology B was approximately 5% higher.

Table 30.

Results of Shapiro–Wilk test on monthly mean humidity (%).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	0.000	0.038	0.095	0.000	0.000	0.444	0.255	0.078	0.009	0.000	0.000	0.000

Table 31.

Results of ANOVA test on monthly mean humidity (%).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Sig. (p < 0.05)	-	-	0.001	-	-	0.001	0.001	0.001	-	-	-	-

Moreover, there were 3.88% to 11.83% humidity differences between typologies A and C for the months of March, June, July and August. A similar situation occurred with differences of 6% to 11.8% between typologies B and C. As a result, typology C appears more humid than the other two LULC typologies (Table 32). Table 33 demonstrates the effect of water surfaces. Around 5% higher humidity was observed in stations where the water surface effect is high, especially in summer, which is quite natural.

Table 32.

Results of Scheffe’s multi-comparison test on monthly mean humidity (%).

LULC		Months
LULC		Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Urban (A)	Rural/suburban (B)	-	-	0.015	-	-	0.016	-	-	-	-	-	-
Urban (A)	Forested L. (C)	-	-	0.001	-	-	0.001	0.001	0.001	-	-	-	-
Rural/suburban (B)	Forested L. (C)	-	-	-	-	-	0.037	0.049	0.031	-	-	-	-
Average differences (°C)^a	A-B	-	-	−4.05	-	-	−5.83	-	-	-	-	-	-
	A-C	-	-	−7.51	-	-	−11.83	−5.33	−3.88	-	-	-	-
	B-C	-	-	-	-	-	−6.00	−11.8	−10.68	-	-	-	-

^aOnly statistically significant values shown: A-C, A-B and B-C.

Table 33.

The effect of water surfaces on the station’s monthly mean humidity (%).

Months	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
0	80.07	79.28	76.45	70.96	73.61	74.28	71.81	73.32	71.70	78.22	79.69	79.95
1	80.29	80.36	78.72	76.03	78.06	79.56	76.24	77.73	75.03	79.76	81.02	80.60
Difference	−0.22	−1.08	−2.27	−5.07	−4.44	−5.28	−4.42	−4.42	−3.33	−1.54	−1.33	−0.66

0: no or little water surface effect, 1: high water surface effect.

Discussion

Istanbul Metropolitan Municipality (IMM) stated in its 2050 vision report that it is working on issues such as fighting the climate crisis, improving the quality of life, reducing environmental pollution and greenhouse gas emissions, and providing climate justice in Istanbul. All these efforts are important to make Istanbul holistically more resilient and sustainable, and directly corresponding the United Nation Sustainable Development Goals: Sustainable Cities and Communities (SDG 11) and Climate Action (SDG 13).⁶² The report highlights that Istanbul’s UHI effect appears to be concentrated in areas defined as densely built environments.

Considering the two main elements of the research, LULC and LCZ, the effects of these on meteorological data are of vital importance in the formation of the UHI. This situation can be discussed in various ways. For example, mean temperatures are the parameter that shows the difference between day and night in the most balanced way. Detecting the urban heat island effect within mean temperatures can be challenging. However, some significant findings emerged: there was a difference between typology A, which was 0.89°C warmer than typology B, and typology C, which was 1.05°C warmer than typology B, on a yearly basis. Furthermore, built-type LCZs were observed to be 0.75°C warmer than land cover-type LCZs, annually. This proves that the UHI was clearly seen on the monthly mean temperature values, and those values were overlapped with the reports of Istanbul Metropolitan Municipality.^63,64

Using the cooling effects of water bodies as a type of helpful service from nature is an important way to reduce urban heat in cities during the daytime.⁶⁵ For example, Yang et al.⁶⁶ showed that sea breezes could alleviate the urban heat island effect in Shenzhen, China. Moreover, Guo et al.⁶⁷ explored the cooling capacity of the sea in Dalian City of China and noted that it can cool the inner city area between 2.5 and 9.2 km. However, the dense urban development in the southern part of Istanbul removes any cooling influence from the Sea of Marmara.⁶⁸ The main source of these temperature differences is thought to originate from the dense urbanization located in the southern part of the city. Despite this situation, natural areas within the urban macro-form work like cool air clusters. The most prominent of these areas can be listed as Aydos Hill, Maltepe, Kayışdağı and Alemdağ, respectively. The fact that all these areas are located on the Anatolian side of Istanbul is an important finding.

The monthly minimum mean temperature (°C) values have been pointed out as a highly significant indicator of the UHI effect in this study. The urban heat island is mainly a nighttime phenomenon.⁶⁹ This is attributed to the fact that urban materials absorb and retain heat throughout the day, releasing it gradually during the night. In contrast, the cooling process is more rapid in rural areas. Consequently, minimum temperatures have the potential to be higher in urban environments due to the delayed release of accumulated heat, contributing to an elevation in minimum temperatures in urban areas.⁷⁰ Accordingly, temperature differences were evident between typologies A and B, and typologies A and C. Typology A was 1.23°C warmer than typology B and was 1.62°C warmer than typology C. In other words, the urban environment was found to be warmer than both rural/suburban and forested landscape, and these differences were statistically significant (p < 0.05). Similarly, Baykara (2023)⁷¹ found the urbanized area of Istanbul to be warmer than rural. Additionally, built-type LCZs were 1.30°C warmer than land cover-type LCZs. When the water surface effect is discussed, the stations with no or less water surface effect in all months were determined to be 0.43°C cooler. This is explained by the slow heating and late cooling of the water surface.

There was a significant output that monthly maximum mean temperature values (°C) of five stations recorded higher values than the others: Fatih-Deniz Bilimleri Enstitüsü (S9), Kadıköy-Rıhtım (S13), Sabiha Gökçen (S16), Samandıra (S17) and Sarıyer-Kumköy (S20) stations. S9 was located in Fatih district, and S13 was located in Kadıköy district. Those are one of the densest districts in terms of building stock. S16 and S17 were in land uses with an airport, that is, the amount of large impermeable concrete surface is high. These stations are likely to stand out in terms of monthly maximum value. S20 was located in the northern part of Istanbul, outside the urban macro-form. Although there is no dense construction in its immediate surroundings, interestingly, it recorded extreme values as well. This phenomenon may be attributed to sea water temperature variations. The Black Sea 24.3°C water temperature was approximately in August between 1970 and 2022.⁷² This underscores the significant impact of water as a contributing factor. Furthermore, the monthly maximum mean temperatures were higher in cases where the water surface effect was low or absent (code 0).

Urban form affects the cooling capacity of the sea by affecting the surface temperature and the fluidity of the wind.⁶⁷ When the days with 35°C and above values were examined, typologies A, B and C obtained a frequency of 11.16, 9.17 and 6.67 days, respectively. In other words, typology A obtained an average of +2 hot-day frequency compared to typology B and +4 hot-day frequency compared to typology C. Similarly, Wei et al. (2022)⁷³ discovered in their study that regions characterized by dense built environments exhibited an increased frequency of days with temperatures of 35°C and above. Moreover, stations close to the water surface have a lower tendency to exceed 35°C. This highlights the fact that the areas on the south coast were still warm, despite being cooled by the effects of the sea. However, this effect was not seen in the inner parts of the south coast due to the built environment that cuts the wind. Furthermore, the frequency of 35°C and above values at Tuzla station (S26) between 2016 and 2022 was determined as 48. This situation can be explained by the closeness of the station to the Tuzla Wastewater Treatment Plant and ISTON (Istanbul Concrete Elements and Ready Mixed Concrete Factory).

The wind is a critical parameter for UHI because increasing wind velocity would decrease the UHI effect.^74,75 Milelli et al. (2023)⁷⁶ indicated that urban areas have less wind velocity than rural. Moreover, Aksak et al. (2023)⁷⁷ found a decreasing trend of wind velocity in Istanbul after the 2000s with the effect of urbanization. Our results demonstrated similar outputs. The stations with less wind were seen in urban areas, meaning that typology A obtained 0.33 m/s less wind velocity than typology B, and achieved 0.31 m/s less wind velocity than typology C, yearly. Furthermore, the three stations with the least winds are Istanbul-Bölge (1.46 m/s), Kadıköy-Göztepe (1.61 m/s) and Beykoz (2 m/s). This situation can be explained by the effect of elements that have the potential to block the wind. The stations with higher monthly mean wind velocity (m/s) values were determined at the stations located in open areas: Beykoz Anadolu-Feneri (4.84 m/s), Silivri-Anamendirek (4.57 m/s) and Atatürk Airport (4.54 m/s). The immediate surroundings of these three stations are quite open, and the elements with the potential to block the wind are limited.

The LULC changes shape of the urban microclimate, including UHI effects regarding urban temperature and humidity.⁷⁸ Urban areas contain less humidity than their surroundings.⁷⁹ This issue can alter the thermal sense of people in urban voids. Because thermal sense of people depends on the temperature and also humidity. Excessive humidity may enhance the latent heat exposure, and this can create a heat-trapping effect especially in the earlier hours in Istanbul. Additionally, Hass et al. (2016)⁸⁰ found that relative humidity were found to vary significantly between locations’ vegetation structure. Our results support the findings of these researchers. Typology A achieved 10% less than typology C and 5% less humidity than typology B. In particular, stations such as Istanbul-Bölge and Atatürk Airport of highly urbanized areas recorded the lowest August mean humidity levels, at 64.97% and 64.74%, respectively. These environments are characterized by a higher prevalence of impervious surfaces, and under the absence of a water effect. The monthly mean relative humidity (%) was affected by water surfaces. Especially in the summer months, stations that have the water surface effect high (code 1) achieved around 5% humidity more. In winter, this situation was close to stable.

Conclusion and suggestions

Understanding and mitigating the UHI is crucial for building resilient cities that align with UN Sustainable Development Goals (SDGs). There is no doubt that UHI is strongly related to the sustainable development (SDG 11) and climate action (SDG 13). The significance of climate data in analyzing the UHI has been clearly demonstrated in this study. This study concluded that LULC and LCZ types are influential in formation of UHI in Istanbul with following details:

• Typology A (urban) indicated higher temperatures compared to stations located in typologies B (rural/suburban) and C (forested landscape).

• The southern part of the city emerges as the most vulnerable to the UHI phenomenon because of the densely built stocks.

• Stations located close to water surfaces exhibit a lower tendency to exceed 35°C.

• Cool air clusters within the urban macro-form are more diverse on the Anatolian side of Istanbul compared to the European side, where such clusters are relatively limited.

• Typology A was found to be less windy than other typologies.

• Typology C exhibited higher humidity compared to typologies A and B.

Incorporating data from 30 stations underscores the precision of the results obtained. Overall, despite the cooling influence of the sea, urban areas in Istanbul exhibit higher temperatures than other spatial types. By fostering collaboration and implementing appropriate strategies, it is feasible to cultivate a more sustainable and resilient urban environment that emphasizes the well-being and quality of life of Istanbul’s inhabitants.

In future studies, it is essential to develop a comprehensive local-based urban heat island mitigation strategy that aligns with landscape planning and design principles. Wind emerges as the most significant climate parameter for dispersing the urban heat island effect. Specifically, cooling winds originating from the northeast direction can significantly enhance their effectiveness within the city when channelled appropriately. Thus, advocating for strategies like establishing green corridors oriented in the northeast-southwest direction becomes imperative. These strategies should be strengthened by integrating them with existing cool island clusters within the city and where green corridors are impractical, creating local cool islands. Furthermore, this entails the creation of conducive pathways to harness the morning sea breeze from the south, a consequence of land–sea breezes. Implementing these strategies offers locally appropriate and effective solutions to mitigate the heat island effect in the Istanbul metropolitan area. This study provides valuable insights and serves as a foundation for informed decision-making in landscape planning and design, enabling the development of effective strategies to mitigate the adverse effects of UHI.

Footnotes

Acknowledgements

This article was developed as part of the doctoral thesis of the corresponding author. We extend our gratitude to Prof. Ömer Lütfi Şen (PhD) and Assoc. Prof. Mert Ekşi (PhD) for their invaluable contributions as members of the thesis monitoring committee. We also thank the referees for their insightful and constructive feedback on the article. Additionally, we express our appreciation to the Istanbul Regional Directorate of Meteorology, 1st Region, for providing the data used in this study.

Author contributions

HO and MEK designed the study and developed the methodology. HO collected the secondary data and performed the analysis. MEK contributed to the theoretical framing of the paper. HO and MEK jointly produced the manuscript text, and HO performed later manuscript revisions.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Hüseyin Ögçe

References

NASA . Global climate change: vital signs of the planet, https://climate.nasa.gov/ (2023, accessed 9 September 2023).

UN-Habitat . Planning for climate change: a strategic, values-based approach for urban planners. Nairobi: UNON, 2014.

Mosca

Canepa

Perini

. Strategies for adaptation to and mitigation of climate change: key performance indicators to assess nature-based solutions performances. Urban Clim 2023; 49: 101580.

UN-DESA . Population division. New York, NY: United Nations Publication, 2022.

UN Environment Programme . GOAL 11: sustainable cities and communities, https://www.unep.org/explore-topics/sustainable-development-goals/why-do-sustainable-development-goals-matter/goal-11 (2023, accessed 24 January 2024).

City of London . Climate action strategy 2020-2027. London: City of London, 2022, https://www.cityoflondon.gov.uk/assets/Services-Environment/climate-action-strategy-2020-2027-20-10-20.pdf

UN . United nations secreteriat climate action plan 2020-2030, 2019, https://www.un.org/management/sites/www.un.org.management/files/united-nations-secretariat-climate-action-plan.pdf

World Bank . World bank group climate action plan 2021-2025. Washington, DC: World Bank, 2021, https://openknowledge.worldbank.org/server/api/core/bitstreams/19f8b285-7c5b-5312-8acd-d9628bac9e8e/content

Oke

. The urban energy balance. Prog Phys Geogr 1988; 12: 471–508.

10.

Stewart

. Redefining the urban heat island, 2011, https://open.library.ubc.ca/collections/24/items/1.0072360

11.

Zhang

. Cooling effect and control factors of common shrubs on the urban heat island effect in a southern city in China. Sci Rep 2020; 10: 17317.

12.

Dihkan

Karsli

Guneroglu

. Evaluation of urban heat island effect in Turkey. Arabian J Geosci 2018; 11: 186. DOI: 10.1007/s12517-018-3533-3.

13.

Dale

Efroymson

Kline

. The land use–climate change–energy nexus. Landsc Ecol 2011; 26: 755–773.

14.

Froese

Schilling

. The nexus of climate change, land use, and conflicts. Curr Clim Change Rep 2019; 5: 24–35.

15.

Bosma

Hein

. The climate and land use change nexus: implications for designing adaptation and conservation investment strategies in Sub-Saharan Africa. Sustain Dev 2023; 31: 3811–3830.

16.

Pielke

. Influence of the spatial distribution of vegetation and soils on the prediction of cumulus Convective rainfall. Rev Geophys 2001; 39: 151–177.

17.

Pitman

. The evolution of, and revolution in, land surface schemes designed for climate models. Int J Climatol 2003; 23: 479–510.

18.

Schulte to Bühne

Pettorelli

. Perspectives: predicting the effects of climate change on ancient woodlands when it interacts with pressures from surrounding land use/land cover. Ecol Manag 2023; 544: 121236.

19.

Huang

Leng

Cai

Wang

Ren

. Facilitating urban climate forecasts in rapidly urbanizing regions with land-use change modeling. Urban Clim 2021; 36: 100806.

20.

Zhang

Hong

. Land surface temperature retrieval from Landsat 8 OLI/TIRS images based on back-propagation neural network. Indoor Built Environ 2019; 30: 22–38.

21.

Government of Ireland . Climate action plan 2023, 2023, https://www.gov.ie/en/publication/7bd8c-climate-action-plan-2023/, https://www.gov.ie/pdf/?file=https://assets.gov.ie/256997/b5da0446-8d81-4fb5-991e-65dd807bb257.pdf#page=null

22.

Republic of Turkey . Republic of Turkey climate change action plan 2011 - 2023, https://webdosya.csb.gov.tr/db/iklim/editordosya/iklim_degisikligi_eylem_plani_EN_2014.pdf (2012, accessed 20 November 2023).

23.

Rao

Tassinari

Torreggiani

. Exploring the land-use urban heat island nexus under climate change conditions using machine learning approach: a spatio-temporal analysis of remotely sensed data. Heliyon 2023; 9: e18423.

24.

Cevik Degerli

Cetin

. Evaluation of UTFVI index effect on climate change in terms of urbanization. Environ Sci Pollut Res 2023; 30: 75273–75280.

25.

Bokaie

Zarkesh

Arasteh

Hosseini

. Assessment of urban heat island based on the relationship between land surface temperature and land use/land cover in Tehran. Sustain Cities Soc 2016; 23: 94–104.

26.

Cai

Zhang

Zheng

Pan

. Quantifying the impact of land use/land cover changes on the urban heat island: a case study of the natural Wetlands distribution area of Fuzhou City, China. Wetlands 2016; 36: 285–298.

27.

Alemayehu

Minale

Legesse

. Land-use and land-cover dynamics nexus to local climate variability in Suha watershed, upper Blue Nile basin, Northwest Ethiopia. Sustain Environ 2023; 9: 2278828.

28.

Srivastava

Shukla

Singh

Jha

. Spatio-temporal dynamics of land use/cover and land surface temperature in Prayagraj city, India. Indoor Built Environ 2023; 32: 1250–1268.

29.

Shi

Huang

Wang

Yang

Shi

. Influence of urbanization on the thermal environment of meteorological station: satellite-observed evidence. Adv Clim Change Res 2015; 6: 7–15.

30.

Cecilia

Casasanta

Petenko

Conidi

Argentini

. Measuring the urban heat island of Rome through a dense weather station network and remote sensing imperviousness data. Urban Clim 2023; 47: 101355.

31.

Erdem Okumus

Terzi

. Evaluating the role of urban fabric on surface urban heat island: the case of Istanbul. Sustain Cities Soc 2021; 73: 103128. DOI: 10.1016/j.scs.2021.103128.

32.

Şekertekin

Marangoz

. Zonguldak metropolitan alanındaki arazi kullanımı arazi örtüsünün yer yüzey sıcaklığına etkisi. Geomatik 2019; 4: 101–111.

33.

Toy

Yilmaz

. Evaluation of 10-year temperature differences between urban and rural areas of a well-planned, unindustrialized, and medium-sized Turkish town, Erzincan. J Urban Plann Dev 2010; 136: 349–356.

34.

Meili

Paschalis

Manoli

Fatichi

. Diurnal and seasonal patterns of global urban dry islands. Environ Res Lett 2022; 17: 054044.

35.

Lokoshchenko

. Urban heat island and urban dry island in Moscow and their centennial changes. J Appl Meteorol Climatol 2017; 56: 2729–2745.

36.

Stewart

Oke

. Local climate zones for urban temperature studies. Bull Am Meteorol Soc 2012; 93: 1879–1900.

37.

Kuşçu Şimşek

Şengezer

. Istanbul metropoliten alanında kentsel isınmanın azaltılmasında yeşil alanların önemi. MEGARON 2012; 7: 116–128.

38.

Karaburun

Demirci

Suen

I-S

. Impacts of urban growth on forest cover in Istanbul (1987–2007). Environ Monit Assess 2010; 166: 267–277.

39.

Khorrami

Heidarlou

Feizizadeh

. Evaluation of the environmental impacts of urbanization from the viewpoint of increased skin temperatures: a case study from Istanbul, Turkey. Appl Geomat 2021; 13: 311–324.

40.

Akkemik

. Ekosistem, iklim ve kentsel büyüme perspektifinden İstanbul ve Kuzey Ormanları. İstanbul: Türkiye Ormancılar Derneği (TOD), 2020.

41.

Ayazli

Kilic

Lauf

Demir

Kleinschmit

. Simulating urban growth driven by transportation networks: a case study of the Istanbul third bridge. Land Use Pol 2015; 49: 332–340.

42.

Aysev

. Urbanization processes of northern Istanbul in the 2000’s: Yavuz Sultan Selim Bridge and the Northern Marmara Highway. Metu J Fac Archit 2022; 39: 137–163.

43.

Akın

Sunar

Berberoğlu

. Urban change analysis and future growth of Istanbul. Environ Monit Assess 2015; 187: 506.

44.

Ünal

Sonuç

Incecik

Topcu

Diren-Üstün

Temizöz

. Investigating urban heat island intensity in Istanbul. Theor Appl Climatol 2020; 139: 175–190.

45.

Ezber

Sen

L Kindap

Karaca

. Climatic effects of urbanization in Istanbul: a statistical and modeling analysis. Int J Climatol. 2007; 27: 667–679.

46.

Karaca

Tayanç

Toros

. Effects of urbanization on climate of İstanbul and Ankara. Atmos Environ 1995; 29: 3411–3421.

47.

Sheather

. A modern approach to regression with R. 1st ed. New York, NY: Springer, 2009. DOI: 10.1007/978-0-387-09608-7.

48.

de Sá

JPM

. Applied statistics using SPSS, STATISTICA, MATLAB and R. 2nd ed. Berlin, Heidelberg: Springer, 2007. DOI: 10.1007/978-3-540-71972-4.

49.

Bivand

Pebesma

Gómez-Rubio

. Applied spatial data analysis with R. New York, NY: Springer, 2008. DOI: 10.1007/978-0-387-78171-6.

50.

R Core Team . R: a language for statistical computing. Vienna, Austria: RFoundation for Statistical Computing, 2022, https://www.r-project.org/ (accessed 24 June 2023).

51.

CORINE . CORINE land cover, https://land.copernicus.eu/pan-european/corine-land-cover/clc2018 (2018, accessed 6 February 2022).

52.

Yang

Peng

LLH

Chen

Yao

. General air temperature and humidity features of local climate zones: a multi-city observational study in eastern China. Urban Clim 2023; 51: 101652.

53.

Ching

Mills

Bechtel

See

Feddema

Wang

Ren

Brousse

Martilli

Neophytou

Mouzourides

Stewart

Hanna

Foley

Alexander

Aliaga

Niyogi

Shreevastava

Bhalachandran

Masson

Hidalgo

Fung

Andrade

Baklanov

Dai

Milcinski

Demuzere

Brunsell

Pesaresi

Miao

Chen

Theeuwes

. WUDAPT: an urban weather, climate, and environmental modeling infrastructure for the anthropocene. Bull Am Meteorol Soc 2018; 99: 1907–1924.

54.

. WUDAPT level 0 training data for Istanbul (Turkey, republic of), submitted to the LCZ generator, https://lcz-generator.rub.de/factsheets/603eada77a39c456eb9d62d83172bc4ac545dad4/603eada77a39c456eb9d62d83172bc4ac545dad4_factsheet.html (2022, accessed 4 September 2023).

55.

Faurie

Varghese

Liu

. Association between high temperature and heatwaves with heat-related illnesses: a systematic review and meta-analysis. Sci Total Environ 2022; 852: 158332.

56.

Gillerot

Landuyt

Chow

Haluza

Ponette

Jactel

Bruelheide

Jaroszewicz

Scherer-Lorenzen

De Frenne

Muys

Verheyen

. Forest structure and composition alleviate human thermal stress. Global Change Biol 2022; 28: 7340–7352.

57.

Giannaros

Agathangelidis

Papavasileiou

Galanaki

Kotroni

Lagouvardos

Giannaros

Cartalis

Matzarakis

. The extreme heat wave of July-August 2021 in the Athens urban area (Greece): atmospheric and human-biometeorological analysis exploiting ultra-high resolution numerical modeling and the local climate zone framework. Sci Total Environ 2023; 857: 159300.

58.

Trájer

Sebestyén

Domokos

Abonyi

. Indicators for climate change-driven urban health impact assessment. J Environ Manag 2022; 323: 116165.

59.

Liu

Lian

Dai

Lai

. Comparing the effects of sun and wind on outdoor thermal comfort: a case study based on longitudinal subject tests in cold climate region. Sci Total Environ 2022; 825: 154009.

60.

Nguyen

Tran

T-N-D

Grodzka-Łukaszewska

Sinicyn

Lakshmi

. Assessment of urbanization-induced land-use change and its impact on temperature, evaporation, and humidity in Central Vietnam. Water 2022; 14(21): 3367. DOI: 10.3390/w14213367.

61.

Rahman

Franceschi

Pattnaik

Moser-Reischl

Hartmann

Paeth

Pretzsch

Rötzer

Pauleit

. Spatial and temporal changes of outdoor thermal stress: influence of urban land cover types. Sci Rep 2022; 12: 671.

62.

İBB . Istanbul 2050 vision. Istanbul, https://vizyon2050.istanbul/upload/content/vizyon2050_strateji_belgesi_120623comp2_2023612_1352369.pdf (2023, accessed 26 January 2024).

63.

İBB . Istanbul iklim değişikliği eylem planı: final raporu. Istanbul, https://cevre.ibb.istanbul/wp-content/uploads/2022/05/Final_Raporu.pdf (2018, accessed 4 March 2023).

64.

İBB . İstanbul iklim değişikliği eylem planı. Istanbul, https://cevre.ibb.istanbul/wp-content/uploads/2022/09/istanbul-iklim-degisikligi-eylem-plani.pdf (2022, accessed 1 April 2023).

65.

Zhang

. Water bodies’ cooling effects on urban land daytime surface temperature: ecosystem service reducing heat island effect. Sustainability 2019; 11(3): 787. DOI: 10.3390/su11030787.

66.

Yang

Xin

Zhang

Xiao

Xia

. Contributions of sea–land breeze and local climate zones to daytime and nighttime heat island intensity. Npj Urban Sustain 2022; 2: 12.

67.

Guo

Zhao

Zhang

Dong

Zhu

Lau

SSY

. Effects of urban form on sea cooling capacity under the heatwave. Sustain Cities Soc 2023; 88: 104271.

68.

Erdem Okumus

Terzi

. Reconsidering urban densification for microclimatic improvement: planning and design strategies for Istanbul. Int J Archit Plan 2022; 10: 660–687.

69.

Parker

. Urban heat island effects on estimates of observed climate change. WIREs Clim Chang 2010; 1: 123–133.

70.

Ögçe

. Kentsel arazi yüzey değişikliklerinin kentsel iklime olan etkisi: Trabzon örneği. Trabzon: Karadeniz Teknik Üniversitesi Fen Bilimleri Enstitüsü, 2023.

71.

Baykara

. An assessment of long-term urban heat island impact on Istanbul’s climate. Int J Environ Geoinformatics 2023; 10: 40–47.

72.

MGM . 2022 yılı karadeniz Deniz suyu sıcaklığı. Ankara, https://www.mgm.gov.tr/FILES/resmi-istatistikler/denizSuyu/Karadeniz-Deniz-Suyu-Sicakligi-Analizi-2022.pdf (2022, accessed 1 August 2023).

73.

Wei

Chen

Blaschke

Peng

Xue

. Synergies between urban heat island and urban heat wave effects in 9 global mega-regions from 2003 to 2020. Rem Sens 2021; 14: 70.

74.

Oke

. The energetic basis of the urban heat island. Q J R Meteorol Soc 1982; 108: 1–24.

75.

Al-Obaidi

Rayburg

Półrolniczak

Neave

. Assessing the impact of wind conditions on urban heat islands in large Australian cities. J Ecol Eng 2021; 22: 1–15.

76.

Milelli

Bassani

Garbero

Poggi

von Hardenberg

Ridolfi

. Characterization of the urban heat and dry island effects in the Turin metropolitan area. Urban Clim 2023; 47: 101397.

77.

Aksak

Öztürk

ŞK

Ünsal

. Kentsel isı adasının iklim parametreleri ve uzaktan algılama üzerinden incelenmesi: İstanbul kenti örneği. Ege Coğrafya Derg. 2023; 32: 151–171.

78.

Huang

Song

. Urban moisture and dry islands: spatiotemporal variation patterns and mechanisms of urban air humidity changes across the globe. Environ Res Lett 2023; 18: 103003.

79.

Çağlak

Özlü

Toy

. Urban – rural climatic difference under marine effect, in the sample of Samsun city. Iğdır Üniversitesi Fen Bilim Enstitüsü Derg 2019; 9: 330–338.

80.

Hass

Ellis

Mason

Hathaway

Howe

. Heat and humidity in the city: neighborhood heat index variability in a mid-sized city in the southeastern United States. Int J Environ Res Publ Health. 2016; 13(1): 117. doi:10.3390/ijerph13010117.

Urban heat island phenomenon in Istanbul: A comprehensive analysis of land use/land cover and local climate zone effect

Abstract

Keywords

Introduction

Material

Istanbul

Istanbul climate

Methods

Long-term annual mean temperature (°C) time series

The selection procedure of meteorological stations and the determination of the year range

Data analysis and accuracy

Determining the spatial land use/land cover and local climate zones of the stations

Data interpretation and visualization

Findings

Monthly mean temperatures (°C)

Monthly minimum mean temperatures (°C)

Monthly maximum mean temperatures (°C)

Frequency of days 35°C and above for selected stations

Monthly mean wind velocity (m/s)

Monthly mean relative humidity (%)

Discussion

Conclusion and suggestions

Footnotes

Acknowledgements

Author contributions

Declaration of conflicting interests

Funding

ORCID iD

References