ASSESSMENT OF THE IMPACT OF URBANIZATION GROWTH ON THE CLIMATE OF BAGHDAD PROVINCE USING REMOTE SENSING TECHNIQUES.

This study was aimed to recognize the spatial and temporal features of urban development and its impact on the climate of Baghdad City. The analytical method of the study relies on changes in Land Use/Land Cover (LULC), Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index (NDBI), and Land Surface Temperature(LST); GIS technology was used to measure these statistics. Landsat (5,8) and Sentinel2A imagery were used to detect the change of urbanization growth, vegetation change, and land surface temperature during the study period from 2001 to 2020, whereas used the supervised classification technique for determination for LULC variations. The results showed significant changes among the LULC classes in the studied periods, as most of the LULC changes were caused by human activities. The most prominent changes in LULC were the urban expansion on agricultural land, continuously in all years, which led to the decline of vegetation resulting from land degradation. The Building area increased from around 863 km2 in 2001 to 1469 km2 in 2020 concentrated mainly in the center, the northeastern and southeastern part of the city. Moderate plants decrease from around 88.5 km2 in 2001 to 59 km2 in 2020 and dense plants decreased from 0.0252 km2 in 2001 to 0.0135 km2 in 2020. This has led to negative effects on the climate where temperature rates increased from (26-47) degrees Celsius in 2001 to (32-56) degrees Celsius in the last year of the study, the highest temperatures were recorded in urban growth areas and areas without vegetation.


INTRODUCTION
In most of the major cities in the world, urbanization is a major concern as cities are expanding at an alarming rate because of the migration from rural to urban areas for better job opportunities and living conditions. In 1950, only 30 percent of the global population lived in cities. However, 66 percent of the global population are expected to live in cities in 2050(27). This double rise will lead to many urban problems (12). During the last five decades, urban spatial areas have grown at an accelerated rate, and urban population growth rates are higher than the overall growth in several countries because urban areas are the locus of economic activity and transport nodes (21). The conversion of natural lands into impermeable lands can have a significant impact on the environment. The hydrological system, biodiversity, and local climate may contribute to bad aspects, such as the case of the urban heat island. a status in which the predominant surface temperature in an urban zone is commonly rising than the surface temperature in a neighboring rural zone, with a large temperature variation during nighttime than within midday. The urban spatial growth study and the resulting phenomenon of the urban heat island often require detailed data on urban areas such as scale, shape, and spatial context: therefore, a method is needed to quickly disclose the data. For urban planners and decision-makers, the timely accessibility of data is of great significance. Fortunately, remote sensing satellite technology provides a substantial promise to reach this requirement. Satellites observations can provide global accuracy and repetitive measures of the conditions of the earth's surface with distinct spatial and spectral resolutions (4,31). To distinguish between urban land and non-urban land, several researchers have used remote sensing imagery. The traditional multispectral classification began with a common method for the description of urban areas. However, due to spectral confusion of the heterogeneous Urban land-class built-up, this does not produce satisfactory precision, usually less than 80 percent. Therefore, many studies have not only used a single method of classification to extract the built-up urban lands but have also combined numerous approaches to improve extraction (21). Remote Sensing is a very active and useful tool that can be used to detect land cover types (6). In recent decades, due to developments in applications for remote sensing and Geographic Information Systems GIS, the usage of has replaced the traditional methods of field surveys for identifying builtup urban land due to advances in remote sensing technology and GIS. In general, satellite imaging techniques for extracting the built-up region can be classified into two groups: Techniques that are based on traditional classification of multispectral images like supervised, object-based, unsupervised, or classification for deep learning (29). Alternate to the classification of images, several researchers have tried normalized indices of difference using particular spectral bands for automatic satellite imagery extraction of built-up ground (31). The built-up land image was produced using the NDBI(32). Significant studies have tried to study land use and shifts in land cover using GIS and remote sensing methods. Anthropogenic behaviors are found to significantly affect the urban climate (3,9,20). Therefore, more efforts are needed to track transitions in LU/LC in urban areas(28). Furthermore, following the replacement of vegetation by urbanized area, climate parameters would also be changed (13, 19,30,33). Due to changes in land cover in the US, both the minimum and maximum temperatures rose (17). In Mahmudiyah district, the change in the land cover and land use was found due to human activities in the period 1990-2007, which led to land degradation (7). considered land use and land cover change of the central components for managing natural resources and monitoring environmental changes. In Egypt, the results were urban growth had brought serious losses of agricultural land and water bodies. Urban growth was responsible for a variety of urban environmental issues like decreased air quality, increased runoff, subsequent flooding, increased local temperature, deterioration of water quality, et. (15). The image taken by Landsat 8 of the OLI sensor was used to analyze the NDVI with Arc GIS 10.3. NDVI result showed that it was ranging between (0--0.6) in no vegetative cover areas such as water surfaces and bare soils areas at 2.13% of the total area of Al-Zawra park area in Baghdad city. It was also ranging between (0.4 -0.3) in areas heavy with large trees planted intensively in that area of 1.25% of the park. Other values ranged between (0.2 -0.3) of areas planted with less intensive trees them previous of 12.4% of the total area of the park. NDVI ranged (0.1 -0.2) in areas planted separately with plants where this area formed 60.92% of the total park area. Last category ranged between (0 -0.1) where it includes light vegetation cover of pale colors including weeds and grasses up to parking lots and empty spaces of (0) NDVI values at 23.30% of the total area of the park (2)The city of Baquba and its outskirts in Diyala province, central Iraq, between latitudes 44º 42ʹ 31.78ʺ ‫ــ‬ 44 º33ʹ 14.99ʺ and 33º41ʹ 46.66ʺ ‫ــ‬ 33 º 48ʹ 23.18ʺ an area of 180,835 km2 has been studied. In order to classify the earth covers, it was relied on the field survey to determine the grounding points, used two satellite data from Landsat 8, the first one on 23/3/2014, the second on 21/3/2019, and the production of (NDVI), (NDWI) and (NDBI) maps. The results of the survey were showed five varieties are vegetation cover, agricultural land, water, buildings and barren land. They were identified and compared with the 75 land control points, The accuracy of the classification was calculated using Kappa It was 89% , and purely concluded that the use of manuals NDVI, NDWI and NDBI was useful for classifying Land coverings and detecting changes as they are considered an easy and fast method (18). According to the literature review above, it is obvious that population growth and subsequent land use and land cover changes have significant adverse effects on the local environment. In general, current studies have concentrated on exploring bivariate relationships between urbanization and changes in LULC (8,26) , urbanization, and temperature changes (11). Hence, the study aimed to find a relationship between the changes that occurred in the land use/land cover and urbanization, and their effects on climate change, to develop a good strategy to prevent climate fluctuations in the city of Baghdad in the future. MATERIALS AND METHODS Description of the study area: Baghdad is located in the central part of Iraq between latitudes 33.452° and 33.184° N, longitudes 44.189° and 44.576° E Figure1, located at an altitude of about 32 meters above sea level, within the flat sedimentary plain sector. Baghdad province includes ten districts and thirty sub-districts, and the borders of the Baghdad Municipality encompass fourteen administrative units, eight in Rusafa (east of the Tigris River) and six in Karkh (west of the Tigris river), with a total coverage area of 870 km 2 . The main characteristics of the study area are large temperature extremism, Low precipitation, low relative humidity, and the sun's high brightness. According to government estimates, the population of Baghdad is more than 8 million (22) and the population is concentrated in the city center. In summer, the region is characterized by a dry and hot climate and a mild climate to cool in spring and winter. The average rainfall is about 150 mm, with the maximum temperature near 40 °C in July. Within Baghdad city, there are many industrial projects such as the food industry, paper industry, etc. The city of Baghdad has also busy residential and commercial areas (16).

Figure 1. Location of the study area Data acquisition
Satellite (Landsat 5 with sensor TM (Path/Row: 168 / 037, 169 / 037), Landsat 8 with sensor OLI (Path/Row: 168/ 37, 169/ 37)) data and Sentinel2A with sensor (MSI) have been downloaded from the " USGS website (https://earthexplorer.usgs.gov)", to Identifying changes in the vegetation cover, (LU/LC), buildings, and land surface temperature. For two separate years, satellite images were chosen, noting that Sentinel 2 does not have a thermal band, so Landsat 8 images were used to calculating the earth's surface temperature for 2020. The first image is from Landsat 5 in 2001 and the second image in 2020 from Sentinel 2. The data from satellites used in the study is shown in Table 1.
After the images were collected, some digital processing of the satellite images was conducted using GIS 10.7 software to prepare visualizations for the LULC classification, buildings, vegetation, and surface temperature analysis of the land. They were also carried out to transform digital values (DN) into radiance values and reflectance values. Pre-processing images Mosaic process were used to merge of two or more Landsat or sentinel images, Baghdad is situated in three areas to scan the Sentinel satellite2 and two areas to scan the Landsat satellite. Study area extraction was applied after completing the mosaic and making as one capture. This method is used to extract the boundaries of the study area Figure 2  A) B)

Figure 3. Mosaic images and extraction of Study area images for C,D) sentinel 2 Post-processing images
The changes in Land Use/Land Cover (LULC), Normalized Difference Built-up Index (NDBI), Normalized Difference Vegetation Index (NDVI), and Land Surface Temperature(LST) were used in this study.

Normalized difference built-up index (NDBI)
The built-up land image was generated using the NDBI with the equation below (32).

Supervised classification
Land-cover map generation is one of the most common applications in the study of remote sensing images and it is typically accomplished by supervised classification techniques. These techniques include the availability of accurate ground reference samples to be used in the classification algorithm's learning process. The accuracy of the labeled training samples depends on the quantity and consistency of the availability of the sample (14). There are several methods of supervised classification, and one of these methods has been used in this research called Maximum Likelihood Classification this method is based on Bay's rule if we consider of the phenomena on the surface of the earth with the letter C: C=Ci, Cj, Ck........Cnc ……………………. (2) Where nc total number of classes, if we take a pixel with vector X to a gray level, the probability of belonging vector x to class Ci given P (Ci| x) …………… ….…………. (3) If the probability is known to each class, it is possible to determine class Which belongs to pixel with vector X by comparing the possibilities, this can be expressed mathematically by equation: P (Ci | x) > P (Cj | x) ……. …. (4) The x belongs to class Ci , Since the probability value cannot be directly identified, it can be applied 'Bay rule' provides for P(Ci), is the probability of class Ci and P(x) is the proportion of affiliation pixel to class Ci thus, the equation becomes as follows: P(x)= ∑ p (Ci | x) * P(Ci) ……………… (5) Where i number of land cover class from 1 to nc, in this equation we need valuable knowledge P (Ci| x) P (Ci| x) = 1/ [ (2π) 1/2 δi ] * exp[ -(x -µi)2 / (2 δi2)]…………… (6) In this equation we need to find only two coefficients for each class: μi which is the mean for the Ci and δi which is the standard deviation of Ci class data (24). we choose training samples through field visits and depending on the visual interpretation with Taking into account that the training samples are homogeneous like figures 4 below: The classification accuracy can also be identified using the Kappa coefficient (k), it is a coefficient that measures the extent of the correlation between the actual 27 agreements and the expected agreement by chance. Where if the value of the Kappa coefficient is greater than 0.8, it represents a good accuracy, if it is between 0.8-0.4, a medium accuracy and if it is less than 0.4, the classification process is poor accuracy This procedure uses all the elements in the error matrix, not just the diagonal (23).

Kappa coefficient
Where (observed accuracy) the number of test points matches the real use and the classified map and represents the proportion of total diagonal elements to the number of test points in the error matrix, and (chance agreement) expected by accident and it can be calculated by the equation: Chance agreement =Σ ( * ) …(9)

Normalized difference vegetation index (NDVI)
To detect the changes related to vegetation, vegetation indices measured from satellite images can be used. The "normalized difference vegetation index (NDVI)" is developed for estimating vegetation cover from the reflective bands of satellite data, and it's considered as one of the most important geotechnical applications and a means to monitor and interpret changes occurring in the land cover. The NDVI differentiation technique showed the best detection of vegetation change, where plants with high reflectivity in the near-infrared wavelength range and low reflectivity in the red wavelength range are determined by this indicator, The NDVI data layer is defined as(2,25) :

NDVI (Landsat 5) = B4 -B3 / B4 + B3 NDVI (Landsat 8) = B5-B4 / B5+B4 NDVI (sentinel 2A) = B8-B4 / B8+ B4 Calculate the land surface temperature
To extract the (LST) from a satellite image, Band 6 that represented the thermal band for (Landsat 5 with sensor TM) and band 10 for (Landsat 8 with sensor OLI) are used. Converting the digital satellite image values to values for spectral radiation is the first step in measuring the surface temperature of the land. Depending upon the values of spectral radiation (min and max) values from the satellite sensor from the metadata register, the radiation values are then translated into temperatures (10). …. (11) Where: L λ : spectral radiation in [mW cm -2 sr -1 mm -1 ] . Lmax l -Lmin l : spectrum radiation for each range, As its values, are extracted from the Metadata file with the satellite image DN: Is the pixel DN value Q calmax -Q calmin : The value of a digital number of sensor spectral bands from the Metadata. The DN converted to spectral radiance using the equation: Lλ = ML*Qcal + AL…. (12) where: Lλ = Spectral radiance in [mW cm-2sr-1mm-1]. ML=Radiance multiplicative scaling factor for the band (RADIANCE_MULT_BAND from the metadata) Qcal = pixel value in DN. AL=Radiance additive scaling factor for the band (RADIANCE_ADD_BAND_ from the metadata). to convert radiation values to temperatures by using Planck's Radiance Function.

RESULTS AND DISCUSSION
The ArcGIS Version 10.7 program was used for remote sensing images of Baghdad for the years 2001 and 2020 in July (where July is one of the very hot summer months), the built-up category was more accurately isolated from the rest of the land use/land cover categories from the Landsat 5 and Sentinel 2A images using NDBI equation 1 to detect the status of Spatio-temporal variation over the years in a buildings area in Baghdad city (2001,2020). During the study period, the blue and nearinfrared bands were used to display the difference for building blocks. The results are show in Figure 5 and Table 2 These results corresponded significantly to the reality of the annual rate of population increase in the city of Baghdad, where variation was observed from time to time due to human economic conditions successively affecting Iraq during the study, which had a major effect on urban expansion in the study region as well as on agricultural areas   Figure 7. To ensure the accuracy of the supervised classification results, the following spectral index values NDVI, NDWI, BSI, and NDIBI were calculated using ArcGIS version 10.7 for satellite images Landsat5 and sentinel 2A and the years 2001 and 2020. Table 4 shows the land cover/land use for 2001 and 2020.     Figure 8  displays NDVI values for the 2001 and 2020  images of the study region while Table 7 shows the NDVI at its maximum and minimum. Unsupervised classification, as seen in Figure 9, was applied to the images. For each image, a vegetation cover area is calculated as seen in    Figure 10 to calculate the effect of the negative change of the vegetation and to increase buildings on LST. The vegetation degradation process is part of the different degradation cycles faced by natural environments, which affects the natural plant system and reduces its density, and thus affects the overall socioeconomic activity of human beings. Human beings and their activities, including abuse of resources, inadequate planning, and even natural and climatic changes, as low precipitation, elevated temperatures, and drought, are the causes of degradation. Factors such as migrations and the decrease in agriculture work, after that a decrease in cultivated areas, and a decline in economic activities, mainly agriculture, are also causing deterioration. Based on the digital information collected from the analysis of satellite images of vegetation cover in the region of the study and the survey of the field, Table 7 and

CONCLUSION
In this study, a statistical method was presented to determine the impact of the change in the land cover/land use and the urbanization on local climate to Baghdad city for July. The results showed the following: the changes in the land cover/land use from 2001 to 2020 affected the local climate due to human activities. * The study detected a spatial variation of the Buildings and vegetation cover from 2001 to 2020 as follows; the Buildings increased from 863.1126 km2 to 1469.4588 km2, Moderate vegetation decreased from 88.4763 km2 to 59.1354km2, dense vegetation from 0.0252 km2 to 0.0135 km2. * the increasing temperatures during July, from (26-47 degrees Celsius) in 2001 to (32-56 degrees Celsius) in 2020, Most of the increase was in residential areas and areas devoid of vegetation in the center, east, southeast, and south of the city. * The NDVI index has an inverse relationship with the surface temperatures of the land, where the area with plant cover (such as the western part of the city) can minimize thermal variation. * The NDBI index has a positive relationship with the surface temperatures of the land, where the built areas (such as the center of the city), it can increase the temperature. The study found that the use of the spectral indexes method to determine the changes in the land cover/land use was better than the classifications (supervised, unsupervised).