• Energy Research

Abstract Attribution and disentanglement of the effects of global greenhouse gas and land-use changes on temperature extremes in urban areas is a complex and critical issue in the context of regional-to-local climate change mitigation and adaptation. Here, an innovative modelling framework based on a large ensemble of urban climate simulations, using SURFEX (a land-surface model) coupled to TEB (an urban canopy model), forced by E20C (a GCM-based reanalysis), is proposed, and applied to the capital of Portugal—Lisbon. This approach allowed to disentangle the main drivers of change of extreme temperatures in Lisbon, while also improving the simulated summer temperature variability compared to E20C, using station observations as reference. The improvements were physically linked to the strong sensitivity of summer mean and extreme temperatures to local land-use properties. The sensitivity was systematically investigated and robustly demonstrated here, with built-fraction (buildings + roads), albedo and emissivity emerging as key surface parameters. The results revealed a very strong summer temperature increase between 1951–1980 and 1981–2010 periods: 0.90 °C for daily maximum temperature (T max), and 0.76 °C for daily minimum temperature (T max). These changes were sensitive to considering different (but constant throughout the simulation) land-uses, varying by about 10% for T max, and around 17% for T min. Regarding the temperature extremes (quantified by extreme hot days, EHD, and extreme hot nights, EHN, respectively defined as exceeding the 95th-percentile of T max and T min) the changes and their dependencies with the land-use are much more drastic. The isolated effect of changing land-use (keeping the climate forcing unchanged) from rural/natural (low built-fraction) towards dense urbanization (high built-fraction) caused a significant increase in EHN (up to ∼+130 d per 30 years, larger than the effect due to climate forcing alone), and in EHD (∼+60 d per 30 years, which is similar to the effect due to climate forcing alone).

We present a detailed evaluation of wind energy density (WED) over Portugal, based on the EURO-CORDEX database of high-resolution regional climate model (RCM) simulations. Most RCMs showed reasonable accuracy in reproducing the observed near-surface wind speed. The climatological patterns of WED displayed large sub-regional heterogeneity, with higher values over coastal regions and steep orography. Subsequently, we investigated the future changes of WED throughout the twenty-first century, considering mid- and end-century periods, and two emission scenarios (RCP4.5 and RCP8.5). On the yearly average, the multi-model ensemble WED changes were below 10% (15%) under RCP4.5 (RCP8.5). However, the projected WED anomalies displayed strong seasonality, dominated by low positive values in summer (< 10% for both scenarios), negative values in winter and spring (up to − 10% (− 20%) under RCP4.5 (RCP8.5)), and stronger negative anomalies in autumn (up to − 25% (− 35%) under RCP4.5 (RCP8.5)). These projected WED anomalies displayed large sub-regional variability. The largest reductions (and lowest increases) are linked to the northern and central-eastern elevated terrain, and the southwestern coast. In contrast, the largest increases (and lowest reductions) are linked to the central-western orographic features of moderate elevation. The projections also showed changes in inter-annual variability of WED, with small increases for annual averages, but with distinct behavior when considering year-to-year variability over a specific season: small increases in winter, larger increases in summer, slight decrease in autumn, and no relevant change in spring. The changes in inter-annual variability also displayed strong dependence on the underlying terrain. Finally, we found significant model spread in the magnitude of projected WED anomalies and inter-annual variability, affecting even the signal of the changes.