Application of remote sensing technology in smart city construction and planning

The construction of a smart city is complex and requires many considerations and improvements. Builders usually use remote sensing technology and related integration methods as tools to assist smart city construction. Compared with traditional methods, remote sensing technology usually has the advantages of high efficiency and low cost. This paper mainly expounds on the background of smart cities, focusing on remote sensing-related technologies and methods that can be applied to different aspects of smart city construction. As for urban traffic pollution monitoring, remote sensing-related technologies have a positive correlation accuracy in traffic pollution monitoring. But it exposes the limitations of their practical application because they cannot detect the pollutants emitted by diesel engines. The advanced optical and radar sensors integrated with the satellite system could generate high-resolution 3D images. MODIS and SAR sensors usually have higher cost characteristics than PALSAR and Lidar, and cloud-free PALSAR, as a functional sensor in the SAR branch, is suitable for use in tropical and subtropical regions. The depth learning-based remote sensing systems in traffic management have the characteristics of drawing complexity and image fluency. The remote sensing-based open-source software of QGIS and the visualization of relevant plug-ins could map the escape areas in the event of an earthquake. The impact of natural disasters on the safety of citizens can be effectively reduced in a low-cost and efficient way, which is conducive to the construction of smart cities.


Introduction
With the development of urbanisation to enhance the living standard of human communities, smart city planning and construction has become the development direction chosen by regional officials or government departments in many countries after the city has already had a specific economic scale and reached pertinent indicators that required to be a smart city such like data foundation.The network structure is the first thing to consider, then is the connectivity of IoT (internet of things), more precisely Smart cities are generally based on the development of conventional cities in terms of economic structure, science, and technology foundation, and support from government and citizen primarily Based on people's sufficient understanding of smart cities.
Aerial photography was the primary source of information for mapping urban green spaces (UGSs), a critical and popular method used for city planning between the 1970s and 1990s Due to technical and methodological limitations.Based on the state-of-the-art remote sensing (RS) technology, pertinent indices such as Normalized Difference Vegetation Index (NDVI) provided by Landsat are used to measure the quantity and greenness of the vegetation in contemplation of getting a better understanding of vegetation health.Others like the broad application of unmanned aerial vehicle-based RS technology (UAV-RS) in traffic monitoring, management, and data collection.
Compared with the novel tech used for city planning, traditional measurement methods do not meet the volume standards of modern society.Consequently, the problem of time efficiency and cost has gradually eliminated the conventional techniques and technologies of urban planning and development.
There are many advantages of RS technology in the perspective of the functions compared with traditional methods in city planning and construction, like polyphyly, high resolution, and real-time [1].So far, there has been considerable progress in research and application in this area, from the perspective of city greening, the use of the PALSAR-based classification method to get high temporal and spatial resolution images in forest cover change during the process of afforestation, and some of the affiliated improved model based on the phased array L-band synthetic aperture radar-based stochastic gradient boosting classification to get a more accurate value of normalised difference vegetation index (SGB-NDVI).
Although many pieces of research on RS technology in smart cities are not systematic, therefore, this study selects typical and state-of-the-art technology and some research models and methods, explains the advantages and disadvantages and prospects of applying specific RS technology in smart cities, and scrutinises its application in smart city construction and planning.

The background of the smart city
Here is a comparative systematic explanation of the smart city concept by integrating a series of ideas.However, there has been no accurate answer to the definition of a smart city over the years of evolution.Still, it is constantly changing according to the development direction of the city.Bowerman et al. proposed that a "Smart city" is customarily defined as the effective use of state-of-the-art relevant technologies in all aspects of urban construction to bring a safe and comfortable urban environment and an efficient and convenient lifestyle to people to achieve a higher index of human development (HDI) and economic growth [2].Toppeta proposed six main aspects of the construction of the smart city: an intelligent economy, intelligent mobility, an intelligent environment, intelligent people, intelligent living, and intelligent governance.However, many measures in the process of smart city construction may not have been included before, so with the different needs of modern society and urban modernisation in other countries, the measures in the process of smart city construction have also increased.Figure 1 shows the model framework for representing the components involved in smart city construction and planning.This model offers the necessary factor in the planning and construction of a smart city comprehensibly and perspicuously.It includes the environment, economy, inhabitant life quality, education, energy usage, and government [3].It can be seen that the concept and measurement standard of the smart city continues to improve with the city's development as time goes by.

Figure 1. Different aspects of smart cities [3].
There is also a systematic theory for the conceptual development and evolution of smart cities.A recently developed concept is proposed based on the continuation of the previous content, which is the "smart city 4.0" [4].Before the idea of 4.0 was proposed, the first three were "SC 1.0", "SC 2.0", and "SC 3.0", separately.The concept of "SC 1.0" refers to the smart city in the early stage of its creation.The standard driver is usually the proposal put forward by the ICT company.Generally, the companies consider the industry chain's development and the company's competitiveness in the market.Irrespective of whether these proposals are necessary for the development of the city."SC 2.0" is a crucial stage in smart city construction.At present, most countries are in "SC 2.0".The driving force in this stage mostly comes from official departments to improve people's quality of life.The "SC 3.0" is a city based mainly on the creative involvement of its inhabitants.Based on the first three general development goals of smart cities at different stages put forward by N. Kaminionos, Zbigniew et al. put forward the fourth main development direction of smart city construction according to today's society development direction and the official government's emphasis on sustainable development.For instance, blockchain application for encrypting and distributing data is the technological foundation of cryptocurrency systems.This paper introduces and explains the application of RS technology in several aspects, such as urban pollution, urban traffic monitoring, city greening, and afforestation.

Application of remote sensing technology in monitoring urban air pollution
For many cities in developing countries, the air quality problem caused by cars with high emissions of harmful gases is a huge factor that hinders the construction of smart cities because it not only affects people's living standards but also imperceptibly causes immeasurable economic losses.According to statistics, nearly 400,000 minor deaths were related to the impact of harmful gases produced by highemission vehicles on the body.There was also a loss of almost one billion US dollars Invested in medical assistance in this area [5].Although from the point of view of car manufacturers, they must install threeway catalytic (TWC) and exhaust gas recirculation (EGR) systems on the car to ensure that the harmful gas, such as NO, CO, emitted by the vehicle is degraded into harmless inorganic compounds (water molecules, carbon dioxide etc.) [6].But there are many uncontrollable factors, such as the driving time of the car, the function of the harmful gas filter system declines, the driver's maintenance of the vehicle, and the driver's driving habits.These factors may cause the increase of carbon emissions from cars imperceptibly.
Nevertheless, harmful gases are inevitably generated and dispersed in the air.For better application, remote sensing technology in pollution detection.It combines other methods and equipment with RS to achieve a more accurate measurement output.Hong Kong Environmental Protection Department (HKEPD) integrated the RS technologies, chassis dynamometer testing, and air quality monitoring and selected some LPG and petrol vehicles as research samples to complete information analysis [5].The research was carried out for re-maintenance of the broken engine to test the percentage reduction of harmful gas emissions and compared engines with varying degrees of damage, then evaluate the accuracy of this method.as one of the research projects has many related pieces of research in this area.The data help the completion of the Hong Kong Transient Emission Test (HKTET) for the improvement and implementation of the enforcement program.
The measurement results of RS and related experimental methods are more efficient and accurate than traditional ones and have lower costs and a wider measurement range.At the same time, this research program also has specific challenges and limitations.First, the horizontal optical configuration limits its application on single-lane roads.The second HKEPD project only measured Emission data for LPG and petrol but not diesel because many uncertain factors make measuring diesel engine emissions very difficult.The possible production of the diesel engine in combustion mode may only account for a small part of the overall production, but diesel is a significant contributor to NOx and particulate matter (PM), so it is still an area worthy of study [6].

Application of remote sensing technology in afforestation of city
Urban greening can bring many benefits to the city.First, it can reduce air pollution, reduce the probability of citizens suffering from respiratory diseases, and improve the hidden value of the real estate.People are more inclined to choose places with gardens or near botanical gardens to settle down.On the other hand, the absorption capacity of urban drainage systems is limited.Plants can help the drainage system with limited capacity to achieve a part of rainwater absorption to reduce the paralysis of urban transportation systems caused by improper rainwater management [7].
Two kinds of remotely sensed data can be used for forest change observation.One is the Time series data, and the other is triple-temporal, also known as time-temporal satellite images.Triple-temporal cannot effectively capture the changing image of the forest spectrally.However, it can provide images with a high spatial and temporal resolution to ensure high efficiency in data analysis.The time series data can ensure the real-time transmission of forest change data and long-term monitoring of significant area ROI of change.However, it is complex for time series to get a high-resolution image.Therefore, one of the two types of data cannot meet the requirements of data detection.Then the integration of the synthetic aperture radar (SAR), Light detection and ranging radar (Lidar), moderate resolution imaging spectroradiometer (MODIS), and the Phased Array L-Band Synthetic Aperture Radar (PALSAR) on satellite (e.g.Advanced Land Observing Satellite (ALOS)) can generate 3D images with high resolution accurately.Data availability is greatly improved, but the cost of PALSAR and Lidar sensors is relatively high compared with the SAR and MODIS [8].
Cloud-free PALSAR is good for monitoring and observing rainy and cloudy areas in tropical areas.And it can also use the advantages of optical sensors to compensate for the disadvantages of radar data by integrating usage.For example, Landsat can help PALSAR collect data, sometimes confusing cities, vegetation, or other areas [9].
Four main PALSAR-based classifiers were used for topography and vegetation analysis classification.Contingent on the NDVI value calculated using the Japan aerospace exploration agency phased array L-band synthetic aperture radar (JAXA-PALSAR), there is a tremendous improvement in PALSAR-based stochastic gradient boosting (SGB-NDVI) from the different calculation periods (from 2005, 2010, 2016).Nevertheless, the accuracy of the PALSAR-based SGB classification is not as good as the vegetation change tracker (VCT), and global land cover (GLC) type classification in perspective of the accuracy of measurement in vegetation detection separately [8,9].

Application of remote sensing technology in traffic management
Traffic monitoring can extract road information by integrating RS and deep learning algorithms.Because this method is more efficient than conventional approaches, the results can be utilised for road condition monitoring.In addition, road feature extraction based on a deep learning model can also be applied to traffic management, emergency rescue operations, road construction, mapping, and updating.
Road feature extraction through UAV and orthographic projection can produce RS images with relatively high resolution.Meanwhile, multispectral and hyperspectral imaging has images with high spectral resolution but moderate spatial resolution.Both methods share the characteristics of being simultaneously expensive, inefficient, and poorly automated.These problems can be successfully reduced by combining RS systems (UAV; RS-UAS) and deep learning (DL) techniques and working jointly.Unmanned aerial vehicles (UAVs) can effectively reduce time expenditure, improve measurement efficiency and prevent cloud cover from covering the camera and sensors during operation.Up to date, a range of tests based on various RS databases and particular forms of coefficient analysis (F1 scores), convolutional networks (CNNs) and generative adversarial networks (GANs), as well as CNN-based fully convolutional networks (FCNs), have been used.They have shown good measurement accuracy (95%-96%).By combining RS and DL methods, road feature extraction can be more effective, but there are still many things that could be accomplished better.Table 1 illustrates the features and drawbacks of the four methods, as well as the characteristics of the content output.Although these methods are imperfect, the accuracy of the DL method measurements can be improved, and the potential for error is reduced by continually enhancing the RS database in related areas [10].Some practical problems still haven't been solved by applying deep learning techniques with UAVbased RS, because the UAV may have accidents with human-crewed aircraft during flight.After all, the UAS is incompatible with the airspace system, which causes a lot of public property losses and endangers the life safety of citizens.National Aeronautics and Space Administration (NASA) has proposed the UAS traffic management (UTM) program to remedy and improve the technical deficiencies in this area.In a series of tests and development processes, significant and influential progress has been made in flight and UAV health data detection reports, UAV crash automatic repair, and, navigation communication in the environment affected by global navigation satellite system (GNSS), but there are still many unsolved problems in the communication between UTM and UAS-UAVs still waiting to be improved [11].

Application of remote sensing technology in natural disaster
Prevention and control of natural disasters and relevant follow-up measures are also worthy problems for constructing smart cities.With the acceleration of urban industrialisation, the abnormal increase in temperature caused by the failure to pay attention to the urban green island effect and the regional carbon dioxide concentration in high carbon emission areas during the construction process may lead to fires in specific seasons.In this process, urban greening and carbon emission treatment play a vital role in fire prevention and control.However, using RS technology to generate analysis images in areas prone to fire Possible to create a segmentation map with no rough edges is also effective.Some other disasters can also be monitored and relevant measures taken with the same method (e.g.floods, landslides, earthquakes, tornados, etc.).This study introduces the most state-of-the-art methods and integrated models based on the application of RS technology in predicting and preventing landslides, earthquakes, tsunamis, and tornadoes.
6.1.Application of remote sensing technology in the landslide 6.1.1.Prediction for landslide.To effectively reduce the economic losses and casualties caused by landslides to the city, a landslide susceptibility map with Greece as the area of interest (AOI) effectively displays the degree of landslides in this area in different colours, which is a high visualisation.In this process, the researchers evaluated the possibility of landslide occurrence through suitable methods based on the geological information system (GIS) environment.
Figure 3 shows the possibility of landslides in different areas of a modern city in Greece, visualized in different colours (e.g. Green mainly represents a low probability of landslides in this area, orange represents a moderate probability of landslides, and red represents a high probability of a high-risk landslide).

The process of getting this result
The possibility of landslides in different areas is based on the relevant landslide, altitude, and geological parameters obtained by RS (Figure 4).Then the Landslide susceptibility (LS) is calculated by the formula, and finally, the classification is carried out based on the GIS environment to obtain the final result [12].First, it is possible to indirectly infer the prediction of natural disasters by observing animal behaviour based on changes in animal behaviour found in many studies.Second, monitoring changes in groundwater chemical properties because many chemicals are dissolved in groundwater.Additionally, changes in groundwater levels can be observed to infer the likelihood of occurrence.Third, use sensors and related technologies to perceive and receive the direction and speed of seismic waves propagating in faults as indicators to predict possible future disasters (e.g.primary waves, secondary waves, and rayleigh waves); usually on the eve of an earthquake, there is a large amount of energy generated from the interior of the earth and causes changes in the earth's magnetic field.Fifth, The generation of radium emanation can also be sensed through related technologies; this is mainly generated from the radioactive properties of groundwater and rocks.Monitoring this gas is also an effective way of seismic monitoring [13].

Earthquake prevention and evacuation instructions.
It can be obtained by analyzing relevant parameters through QGIS, which can give people an escape area in the event of an earthquake.Through the images obtained by satellite, the distance between different buildings can be calculated to estimate the threat of building debris in a particular area, as well as digital elevation model (DEM) imagery calculations of elevations in different parts of the city and Assess the possibility of safe evacuation.Figure 6 shows the analysis of escapeable areas using related software and open-source plug-ins.

Application of remote sensing technology in tsunamis
A tsunami is a less common natural disaster for inland areas.Still, it is relatively common in coastal areas located in volcanic seismic belts and threatens the coastal industrial chain and the safety of citizens.Using RS and related integration technologies, tsunami protection can be achieved.Therefore, it can effectively reduce economic losses and casualties.This chapter will introduce several cutting-edge RS techniques used in this area.IOP Publishing doi:10.1088/1742-6596/2608/1/01205210 6.3.2.Tsunami and using SAR (synthetic aperture radar).Figure 6 is a set of high spatial resolution SAR images obtained by satellite (AOI is the area north of Tokyo, Japan, that was hit by the tsunami).Figure 6(a) shows the image before the tsunami hit the area.

Conclusions
A smart city itself is a dynamic concept.The description of the smart city is continuously adjusted with time, the degree of urban development, and national policies and development directions.Most human cities do not complete all parts of smart city construction.Adapting to environmental changes is the main direction of smart city development.
In the planning and constructing of smart cities, RS and RS-related methods are more efficient and low-cost than traditional methods.As for pollutant monitoring, the HKEPD carried out the HKTET project to assess the accuracy of RS measurement and monitoring and the impact of engines with different degrees of damage on the environment by comparing engines with varying degrees of damage and monitoring their carbon emissions through RS technology.
As for urban greening, several common optical and radar sensors are introduced, and their advantages and disadvantages are compared, such as synthetic aperture radar (SAR), light detection and ranging radar (Lidar), model resolution imaging spectrometer (MODIS), and the phased array L-band synthetic aperture radar (PALSAR).
As for traffic management, four mainstream deep learning models based on RS technology are introduced by listing their advantages and disadvantages, the most complex tasks that can be completed, the characteristics of their output results and the smoothness of images.To evaluate the scenarios that are appropriate for their use.
As for natural disaster prevention, LSM produced by using relevant formulas and engineering software effectively shows the possibility of landslides in the city and surrounding mountainous areas and visualises it.It is possible to use QGIS and relevant plug-ins to complete the evacuation site in case of an earthquake.

Figure 3 .
Figure 3.The landslides in different areas of a modern city in Greece, visualised in different colours [12].

Figure 4 .
Figure 4.The safe escape area in the city was obtained after visual classification processing through DEM data [12].6.2.Application of remote sensing technology in earthquake 6.2.1.Earthquake Prediction.So far, it is difficult for humans to directly predict an earthquake's exact time and area through sensors or RS technology.Still, it is possible to indirectly observe and analyse changes in other related natural environments to achieve the purpose of predicting earthquakes.First, it is possible to indirectly infer the prediction of natural disasters by observing animal behaviour based on changes in animal behaviour found in many studies.Second, monitoring changes in groundwater chemical properties because many chemicals are dissolved in groundwater.Additionally, changes in groundwater levels can be observed to infer the likelihood of occurrence.Third, use sensors and related technologies to perceive and receive the direction and speed of seismic waves propagating in faults as indicators to predict possible future disasters (e.g.primary waves, secondary waves, and rayleigh waves); usually on the eve of an earthquake, there is a large amount of energy generated from the interior of the earth and causes changes in the earth's magnetic field.Fifth, The generation of radium emanation can also be sensed through related technologies; this is mainly generated from the radioactive properties of groundwater and rocks.Monitoring this gas is also an effective way of seismic monitoring[13].

Figure 6 (
b) shows the area's image after being hit by the tsunami.By comparing Figure6(a) and Figure6(b), it can be observed that the part of the submerged area has turned white.By modifying the RGB value of the SAR image, Figure6 (c)shows a FALSE-color image.The submerged area can be observed more clearly, to facilitate the implementation of post-disaster recovery measures.

Figure 6 (
d) shows that the area affected by the tsunami is based on Figure 6 (b).

Table 1 .
Different deep-learning techniques for road extraction are advantages and drawbacks.