Thermal echoes : refining land surface temperature monitoring over the Maltese island

Abstract

This research investigates Land Surface Temperature (LST) variations across the Maltese Islands using satellite remote sensing validated with in-situ observations. The main objective is to assess the accuracy and spatial variability of LST derived from Landsat 8, MODIS (Terra and Aqua), and Sentinel-3 satellites by comparison with ground-based measurements. Emphasis is placed on examining how rural and urban land use types influence the formation of urban heat island (UHI) effects. Monthly ground observations were conducted over a six-month period, cap turing both morning and nighttime conditions to assess diurnal temperature variation. Meteorological parameters such as humidity, wind speed, and wind direction,were also recorded to analyse their relationship with satellite in-situ LST discrepancies. Satellite derived and in-situ LST measurements showed strong correlations across all platforms. In the morning, Landsat 8 exhibited a slight positive mean bias (+1.04 °C), indicating a minor overestimation, whereas MODIS and Sentinel3 underestimated temperatures (-3.82 °C and -1.89 °C respectively). Despite these biases, high correlation coefficients (0.82 to 0.85) confirmed satellite reliability. Nighttime observations initially revealed significant negative biases for MODIS (-6.91 °C) and Sentinel-3 (-6.89 °C). After applying corrections, residuals reduced substantially, approaching zero, and correlation (r value) strengthened (0.87 for MODIS, 0.75 for Sentinel-3), significantly improving nighttime LST accuracy. Urban and rural areas exhibited distinct thermal patterns, with a pronounced nighttime UHI effect as urban regions retained more heat. Interestingly, daytime patterns reversed seasonally, with rural zones warmer than urban areas during hotter months due to sparse vegetation and soil exposure, highlighting the importance of land cover and solar geometry