Surface Temperature Variation Prediction Model Using Real-Time Weather Forecasts

Combination of climate change and urbanization are heating up cities and putting the lives of millions of people in danger. More than half of the world's total population resides in cities and urban centers. Cities are experiencing urban Heat Island (UHI) effect. Hotter days are associated with serious health impacts, heart attaches and respiratory and cardiovascular diseases. Densely populated cities like Manhattan, New York can be affected by UHI impact much more than less populated cities. Even though many studies have been focused on the impact of UHI and temperature changes between urban and rural air temperature, not many look at the temperature variations within a city. These studies mostly use remote sensing data or typical measurements collected by local meteorological station networks. Local meteorological measurements only have local coverage and cannot be used to study the impact of UHI in a city and remote sensing data such as MODIS, LANDSAT and ASTER have with very low resolution which cannot be used for the purpose of this study. Therefore, predicting surface temperature in urban cities using weather data can be useful.Three months of Field campaign in Manhattan were used to measure spatial and temporal temperature variations within an urban setting by placing 10 fixed sensors deployed to measure temperature, relative humidity and sunlight. Fixed instrument shelters containing relative humidity, temperature and illumination sensors were mounted on lampposts in ten different locations in Manhattan (Vant-Hull et al, 2014). The shelters were fixed 3-4 meters above the ground for the period of three months from June 23 to September 20th of 2013 making measurements with the interval of 3 minutes. These high resolution temperature measurements and three months of weather data were used to predict temperature variability from weather forecasts. This study shows that the amplitude of spatial and temporal variation in temperature for each day can be predicted by regression of weather variables. In addition amplitude of spatial variations were most dependent on temperature, north winds, and high level lapse rate and the temporal variations were most dependent on temperature and lapse rates.