Transfer learning reveals large discrepancies between air and land surface temperatures in cities

Understanding of urban weather and climate is severely limited by data poverty resulting from a dearth of true urban weather stations 1, 2 . As a result, land surface temperature (Ts), obtained from remote sensing platforms, has been widely used as a stand-in for near-surface air temperature (Ta) [3] [4] [5] [6] [7] despite their fundamental differences 8 , especially in urban areas. This has led to erroneous characterization of urban heat stress and urban climate impacts 9 . Here we develop a novel urban transfer learning framework (U-TL) to address this critical gap and to provide urban high-resolution air temperature (U-HAT) data at large scales across the contiguous United States (CONUS). U-TL demonstrates high accuracy and strong robustness in predicting urban Ta, even with limited training data. The resulting U-HAT is the first high-resolution urban Ta dataset capable of accurately reproducing observed and well-established urban climatology. U-HAT reveals substantial Ts-Ta discrepancies and therefore cautions the use of Ts to characterize urban heat. We show that satellite-measured Ts substantially overestimates both urban heat stress magnitude and intra-city disparities, which have very consequential implications for urban heat exposure, vulnerability, and adaptation policy making. Cities, despite being widely recognized as global centers of climate hazards and exposure [10] [11] [12] , remain severely underrepresented in terms of in-situ meteorological observations. This gap stems primarily from the scarcity of research-grade weather stations in urban areas 1,2 (Fig. 1 ). Standard weather stations are disproportionately placed in non-urban landscapes (Fig. 1a, b ). Even those stations that are classified as "urban" (see Fig. 1c-g ) are most often located at airports or adjacent open fields, thereby failing to capture the true meteorological and climatological characteristics of the built environments [13] [14] [15] . A key reason for this misrepresentation lies in the World Meteorological Organization (WMO) siting guidelines, which stipulate weather stations to be installed on open, grass-covered ground and far away from built obstructions 16 . While the WMO guidelines aim to ensure consistent, uncontaminated measurements for capturing broader weather and climate signals, adherence to these standards effectively precludes the placement of conventional weather stations in true built environments, leaving only peripheral locations such as airports or city outskirts as apparent viable options 17 . Consequently, very few stations are situated in core urban areas worldwide. Even in the rare occasions where stations do exist in true built environments, their sparse presence (e.g., just one or two stations per city) fails to capture the inherent large intra-city heterogeneity of urban landscapes 18 . This pervasive urban data poverty has been a major obstacle to urban climate research for decades, hindering observationbased studies, compromising validation efforts for climate and weather models, and limiting data assimilation in historical reanalysis products and weather forecasting systems, particularly for urban areas 19 .

Proportion of meteorological stations categorized by impervious surface fraction within a 200-meter buffer. Impervious surface fraction is based on the "built-up" category in the ESA WorldCover dataset 22 . c-g, Satellite images from Google Earth ( http://earth.google.com ) showing the surrounding environments of selected stations.

Land surface temperature (Ts), a bulk radiative skin temperature of objects on the ground as estimated from a satellite view, has therefore been commonly used as a stand-in for near-surface air temperature at the screen height (Ta) to study the urban thermal environment in recent years 3- "urban heat", which can be often confused with those from Ta. Ts offers several advantages over sparse weather stations, including global coverage, consistency through measurements from the same satellite sensor, and spatial continuity 23, 24 , making it particularly suitable for large-scale studies and detailed intra-urban analyses. However, Ts-based applications, while providing important insights into urban thermal conditions 25, 26 , do not directly reflect physiologicallyrelevant heat hazard and are thus subject to critical limitations when drawing impact-or riskrelated implications such as those related to public health or urban adaptation policies 9 . Indeed, Ts itself does not inherently represent human-perceived thermal conditions or other impact-relevant variables (e.g., heat-related health risks or energy use). Therefore, studies relying on Ts to infer urban climate impacts and adaptation strategies have to assumeexplicitly or implicitlythat either (i) Ts is strongly correlated with Ta, or (ii) urban-rural difference in Ts (surface urban heat island or UHIs) is strongly correlated with urban-rural difference in Ta (canopy urban heat island or UHIa). However, multiple recent observational studies in diverse cities or regions have demonstrated that neither of the two assumptions is justified 9, 13, 27 . Ts and Ta, while related, are conceptually and physically distinct 8 . Their relationship is influenced by the complex interplay of many factors, including land surface properties (e.g., vegetation cover, imperviousness, and aerodynamic roughness), sky conditions (e.g., water vapor and cloud cover), and boundary layer processes 8, 28 , especially in highly heterogeneous urban landscapes. This raises significant concerns about the validity of using Ts as a proxy for Ta.

Recent observational-based attempts to address this critical urban data gap have primarily focused on two approaches: crowdsourced citizen weather station (CWS) measurements 29, 30 and empirical data-driven methods to infer urban Ta 31, 32 . The former has emerged as a promising avenue for local and regional urban observational studies due to the recent proliferation of CWS networks 9, [33] [34] [35] . Their measurements, however, are subject to two critical limitations. First, CWS measurements are prone to substantial errors, biases, and uncertainties due to improper station placement (e.g., on rooftops or near HVAC [Heating, ventilation, and air conditioning] exhausts), lack of calibration and maintenance, instrument or internet malfunctions, and inconsistent metadata 30, 36, 37 . Second, CWSs are subject to uneven distribution both within cities and across regions 38, 39 . They are, by design, primarily installed in more conveniently accessible locations (e.g., backyards of residential buildings), and often in more affluent neighborhoods or regions, leading to sampling biases that undermine the representativeness of their measurements. These factors significantly limit the reliability of CWS data and complicate their use for global or large-scale urban climate studies.

On the other hand, data-driven approaches aim to empirically infer urban Ta using standard observations and/or auxiliary parameters. Broadly, these methods fall into two categories: spatial interpolation from standard weather stations [40] [41] [42] and statistical/machine learning techniques to predict urban Ta [43] [44] [45] [46] [47] [48] [49] [50] . Both, however, are fundamentally constrained by the overall dearth of stations in true urban environments. The former often produces interpolated Ta datasets substantially biased toward non-urban landscapes due to their much higher station density, resulting in substantial inaccuracy in urban areas 31 . The latter, particularly data-intensive deep learning techniques, struggle to establish robust relationships between predictors and urban Ta due to insufficient representative training samples from urban cores. Many of such methods have to rely heavily on Ts as a predominant input feature. Consequently, their Ta products often fail to reproduce observed urban climate signals, for example, the well-established reversal of diurnal patterns for UHIa and UHIs, preventing their applications in urban climate research 51 .

To address this critical data poverty issue, we present a novel framework based on the idea of transfer learning, termed U-TL (Extended Data Fig. 1a ), which provides urban high-resolution air temperature (U-HAT) data at large scales. U-HAT is developed at 1-km resolution, covering 120 To address this critical data poverty issue, we present a novel framework based on the idea of 121 transfer learning, termed U-TL (Extended Data Fig. 1a), which provides urban high-resolution 122 air temperature (U-HAT) data at large scales. U-HAT is developed at 1-km resolution, covering 123 384 largest cities in the contiguous United States (CONUS) from 04-01-2013 to 12-31-2023 124 twice a day, at 1:30 AM and 1:30 PM local time. Comprehensive validation of the U-HAT data 125 demonstrates strong performance in capturing urban-specific climate signals and gradients with 126 high fidelity. Unlike existing methods, our framework effectively reproduces observed urban 127 climatology both within and across cities, addressing a longstanding barrier in urban climate 128 research. Future releases of U-HAT will extend coverage to all urban areas globally.