9:30 AM - 9:45 AM
[HTT19-03] Proposal of nocturnal light data as a proxy variable for building height
Keywords:Night Time Light, Digital Surface Model, Suomi NPP
In this study, we examine whether night light can be a proxy for building height.
The height of the building is a very important value for the damage situation of the city at the time of disaster and the urban design. However, building height data can only be obtained for high-rise buildings such as landmarks on Open Street, and it is necessary to purchase data to obtain height data including low-rise buildings nationwide. Another major disadvantage is that it is difficult to acquire data in chronological order. Building height data can be confirmed in chronological order, and user verification can be performed for various studies only after the aging of cities and towns.
In this study, we verified whether it is possible to estimate the height of a building with a high cost by using night light data whose time scale has been released in detail for free in recent years.
So far, active research has been conducted to estimate GDP using Night-Time Light Data. In particular, developing countries with low reliability of statistical data and rural poor areas that have not been surveyed in the first place have achieved very great results in the study area.
On the other hand, economic forecasts using satellite images in urban areas are not very active due to the coarseness of the spatial resolution of satellite images and the complex urban structure. In addition, since it is necessary to verify the accuracy of night light GDP estimation using data based on the presentations of each country, the difficulty of accuracy verification is described in various papers in developing countries due to its low reliability. It has been mentioned as an issue. When using night light to estimate GDP in this way, there is a strong tendency to choose a region.
In this study, we will verify whether it can be a proxy for the height of a building as a new usage of night light.
This takes advantage of the fact that the more floors a building have, the more light it has per unit area. Since this law is universal, it is possible to replace the height of buildings around the world with night light if the correlation between night light and building height can be found.
In this study, we used the 1m-resolution 3D data set released by Hyogo Prefecture for the first time in Japan on January 10, 2020, and the nighttime light data of the Suomi NPP satellite jointly managed by NOAA and NASA, which has been in operation since 2011. Correlation analysis was performed. Among the aerial laser survey results, it is obtained by the difference between the digital elevation model, which is a digital elevation model of the ground surface with buildings, vegetation, etc. removed from the digital surface model of the entire Hyogo prefecture, which is the numerical surface layer including the ground surface and buildings, vegetation, etc. Data on the height of buildings and vegetation above the ground surface. The observation period is from 2012 to 2013. In addition, there were no missing values in each of the Digital Surface Model and Digital Elevation Model. Since the observation period of Hyogo prefecture was from 2012 to 2013, the annual data of 2014 was used for the night light. The spatial resolution was about 450 meters, excluding outliers of the night light and setting the non-light areas to zero.
In order to classify the building as vegetation, the outer circumference data of the building was acquired from the basic map information of the Geographical Survey Institute, and the data of only the height of the building was extracted. The night light was layered, and the pixels with 70% or more of the buildings in the 450m square were selected, and the average height of the buildings in those pixels was correlated with the luminous intensity of the night light. The x-axis is the value of night light weighted and averaged by the number of observations in each month for the pixel, and the y-axis is the average height of the buildings in the pixel.
As a result of the correlation analysis, the number of data was 1028 and the coefficient of determination was 0.392. The null hypothesis was rejected because the P-value was well below the significance level of 0.05. In this way, it was found that the night light has a certain correlation with the height of the building. This can be evaluated as having a correlation even if the building uses data with a roughness of 70% or more in Pixel. We calculated more than 50%, more than 60%, more than 80%, and more than 90% of buildings in Pixel, but 70% or more was the most accurate result. In the future, we would like to conduct verification in other regions and countries to investigate whether there is any regional difference in the correlation between night light and building height.
The height of the building is a very important value for the damage situation of the city at the time of disaster and the urban design. However, building height data can only be obtained for high-rise buildings such as landmarks on Open Street, and it is necessary to purchase data to obtain height data including low-rise buildings nationwide. Another major disadvantage is that it is difficult to acquire data in chronological order. Building height data can be confirmed in chronological order, and user verification can be performed for various studies only after the aging of cities and towns.
In this study, we verified whether it is possible to estimate the height of a building with a high cost by using night light data whose time scale has been released in detail for free in recent years.
So far, active research has been conducted to estimate GDP using Night-Time Light Data. In particular, developing countries with low reliability of statistical data and rural poor areas that have not been surveyed in the first place have achieved very great results in the study area.
On the other hand, economic forecasts using satellite images in urban areas are not very active due to the coarseness of the spatial resolution of satellite images and the complex urban structure. In addition, since it is necessary to verify the accuracy of night light GDP estimation using data based on the presentations of each country, the difficulty of accuracy verification is described in various papers in developing countries due to its low reliability. It has been mentioned as an issue. When using night light to estimate GDP in this way, there is a strong tendency to choose a region.
In this study, we will verify whether it can be a proxy for the height of a building as a new usage of night light.
This takes advantage of the fact that the more floors a building have, the more light it has per unit area. Since this law is universal, it is possible to replace the height of buildings around the world with night light if the correlation between night light and building height can be found.
In this study, we used the 1m-resolution 3D data set released by Hyogo Prefecture for the first time in Japan on January 10, 2020, and the nighttime light data of the Suomi NPP satellite jointly managed by NOAA and NASA, which has been in operation since 2011. Correlation analysis was performed. Among the aerial laser survey results, it is obtained by the difference between the digital elevation model, which is a digital elevation model of the ground surface with buildings, vegetation, etc. removed from the digital surface model of the entire Hyogo prefecture, which is the numerical surface layer including the ground surface and buildings, vegetation, etc. Data on the height of buildings and vegetation above the ground surface. The observation period is from 2012 to 2013. In addition, there were no missing values in each of the Digital Surface Model and Digital Elevation Model. Since the observation period of Hyogo prefecture was from 2012 to 2013, the annual data of 2014 was used for the night light. The spatial resolution was about 450 meters, excluding outliers of the night light and setting the non-light areas to zero.
In order to classify the building as vegetation, the outer circumference data of the building was acquired from the basic map information of the Geographical Survey Institute, and the data of only the height of the building was extracted. The night light was layered, and the pixels with 70% or more of the buildings in the 450m square were selected, and the average height of the buildings in those pixels was correlated with the luminous intensity of the night light. The x-axis is the value of night light weighted and averaged by the number of observations in each month for the pixel, and the y-axis is the average height of the buildings in the pixel.
As a result of the correlation analysis, the number of data was 1028 and the coefficient of determination was 0.392. The null hypothesis was rejected because the P-value was well below the significance level of 0.05. In this way, it was found that the night light has a certain correlation with the height of the building. This can be evaluated as having a correlation even if the building uses data with a roughness of 70% or more in Pixel. We calculated more than 50%, more than 60%, more than 80%, and more than 90% of buildings in Pixel, but 70% or more was the most accurate result. In the future, we would like to conduct verification in other regions and countries to investigate whether there is any regional difference in the correlation between night light and building height.