Does quality control matter? Surface urban heat island intensity variations estimated by satellite-derived land surface temperature products

The temporally regular and spatially comprehensive monitoring of surface urban heat islands (SUHIs) have been extremely difficult, until the advent of satellite-based land surface temperature (LST) products. However, these LST products have relatively higher errors compared to in situ measurements. This has resulted in comparatively inaccurate estimations of SUHI indicators and, consequently, may have distorted interpretations of SUHIs. Although reports have shown that LST qualities are important for SUHI interpretations, systematic investigations of the response of SUHI indicators to LST qualities across cities with dissimilar bioclimates are rare. To address this issue, we chose eighty-six major cities across mainland China and analyzed SUHI intensity (SUHII) derived from Moderate Resolution Imaging Spectroradiometer (MODIS) LST data. The LST-based SUHII differences due to inclusion or exclusion of MODIS quality control (QC) flags (i.e., ΔSUHII) were evaluated. Our major findings included, but are not limited to, the following four aspects: (1) SUHIIs can be significantly impacted by MODIS QC flags, and the associated QC-induced ΔSUHIIs generally accounted for 24.3% (29.9%) of the total SUHII value during the day (night); (2) the ΔSUHIIs differed between seasons, with considerable differences between transitional (spring and autumn) and extreme (summer and winter) seasons; (3) significant discrepancies also appeared among cities located in northern and southern regions, with northern cities often possessing higher annual mean ΔSUHIIs. The internal variations of ΔSUHIIs within individual cities also showed high heterogeneity, with ΔSUHII variations that generally exceeded 5.0 K (3.0 K) in northern (southern) cities; (4) ΔSUHIIs were negatively related to SUHIIs and cloud cover percentages (mostly in transitional seasons). No significant relationship was found in the extreme seasons. Our findings highlight the need to be extremely cautious when using LST product-based SUHIIs to interpret SUHIs.