GS
Geunchan Song
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
1 records found
1
From satellite images to urban-scale air temperatures
A Multi-Layer Perceptron approach
A detailed understanding of the Urban Heat Island is crucial for energy efficiency, renewable integration, building system automation and demand response in district energy systems. Satellite images provide high resolution information about land surface temperatures (LST) at urban scale, but do not provide information about air temperature (AT), which affect energy performance and thermal comfort. This paper presents a methodology to estimate AT at urban scale for the case of Seoul, South Korea. A MultiLayer Perceptron (MLP) neural network was developed to convert satellite-derived LST to AT using 12 years of satellite imagery and air temperature measurements from local automatic weather stations (AWS). The method is then independently tested by comparing the predicted temperatures to the measurements from 1000+ distributed “Seoul Data of Things” (S-DoT) sensors in the city. The model was found to achieve an R2 value of 0.968 during validation using the AWS data from the period 2013-2024. The mean difference during the independent testing was 0.98°C, with an R2 of 0.807, confirming the model successfully predicts actual air temperatures rather than sensor-specific values.
...
A detailed understanding of the Urban Heat Island is crucial for energy efficiency, renewable integration, building system automation and demand response in district energy systems. Satellite images provide high resolution information about land surface temperatures (LST) at urban scale, but do not provide information about air temperature (AT), which affect energy performance and thermal comfort. This paper presents a methodology to estimate AT at urban scale for the case of Seoul, South Korea. A MultiLayer Perceptron (MLP) neural network was developed to convert satellite-derived LST to AT using 12 years of satellite imagery and air temperature measurements from local automatic weather stations (AWS). The method is then independently tested by comparing the predicted temperatures to the measurements from 1000+ distributed “Seoul Data of Things” (S-DoT) sensors in the city. The model was found to achieve an R2 value of 0.968 during validation using the AWS data from the period 2013-2024. The mean difference during the independent testing was 0.98°C, with an R2 of 0.807, confirming the model successfully predicts actual air temperatures rather than sensor-specific values.