Abstract
Beirut, a densely populated city along the Mediterranean coastline, has been witnessing temperature records well above thermal comfort levels especially during the past decade. The significant effect of existing urbanised materials on these rising urban temperatures has been verified through the previous modeling of the urban heat island phenomenon (UHI) of the city using the Town Energy Balance model (TEB). The predominantly dense mix of midrise buildings spread across the city with few trees has also been confirmed by previous local climatic zoning classification.
The first aim of this paper was to compare different input data used for the classifications. Two Landsat 8 scenes (optical and thermal, acquired in 2013 and 2014) as well as one Sentinel-1, and one Sentinel-2 scene (both acquired in 2015) were tested as input data. The best results were achieved with both optical sensors, while thermal and SAR data alone did not perform as well. Secondly, the transferability of training area between Beirut and Damascus city was tested. It was found that the diverse biophysical and climatic backgrounds make this difficult resulting in poor accuracies. Finally, long-term surface temperature characteristics from thermal remote sensing images were compared to the LCZ classifications. Although the surface heat island effects could be seen in annual cycle parameters, the patterns were dominated by land cover and topo-climatic effects due to the high vertical extent of the region of interest.
