Finisterra , LX ( 129 ), 202 5, e37506 ISSN: 0430 - 5027 doi: 10.18055/Finis37506 Artigo Published under the terms and conditions of an Attribution - NonCommercial - NoDerivatives 4.0 International license. THE VANISHING SNOW COVER IN SERRA DA ESTRELA: LEVERAGING SCARCE DATA FOR DIAGNOSTICS AND FUTURE SCENARIOS C ARLA M ORA 1 , 2 G ONÇALO V IEIRA 1,2 ABSTRACT This study investigates the evolution of snow cover in Serra da Estrela, the highest mountain in mainland Portugal, by integrating observational data from the Penhas Douradas meteorological station, remote sensing imagery, and climate model outputs. Results reveal a marked decline in the number of snow cover days at Penhas Douradas, droppi ng from an average of 53 days in the 1950s to 28 days in the 2010s, corresponding to an overall decrease of 5.4 days per decade. The rate of decline has slowed since the 1980s, as the station’s elevation still allows for snow during cold events but not for its sustained presence. To address the lack of observational data from higher elevations, we integrated multiple data sources including low - elevation records, satellite imagery, and climate model projections to reconstruct snow cover dynamics on the T orre Plateau since the late 19th century. Our reconstruction indicates a decrease from approximately 170 to 120 snow cover days in recent decades. Future projections under different climate scenarios suggest a continued and substantial reduction in snow co ver, potentially reaching zero days by the end of the century under high - emission scenarios. Under the most extreme scenario, the Torre Plateau could experience snow conditions similar to those currently observed at 900 metres of elevation. Keywords: Snow , climate change , remote sensing , Serra da Estrela. RESUMO O DECLÍNIO DA COBERTURA DE NEVE NA SERRA DA ESTRELA: APROVEITAMENTO DE DADOS ESCASSOS PARA DIAGNÓSTICO E CENÁRIOS FUTUROS. E ste estudo investiga a evolução da cobertura de neve na Serra da Estrela, a montanha mais elevada de Portugal continental, através da integração de dados observacionais da estação meteorológ ica das Penhas Douradas, imagens de deteção remota e resultados de modelos climáticos. Os resultados revelam uma redução acentuad a no número de dias com cobertura de neve nas Penhas Douradas, passando de uma média de 53 dias na década de 1950 para 28 dias na década de 2010, o que corresponde a um decréscimo global de 5,4 dias por década. A taxa de declínio abrandou desde a décad a de 1 980, uma vez que a altitude da estação ainda permite a ocorrência de neve em eventos frios, mas não a sua permanência. Para colmatar a escassez de dados observacionais nas cotas mais elevadas, integrámos diversas fontes incluindo registos de menor altitu de, imagens de satélite e projeções de modelos climáticos de modo a reconstruir a dinâmica da cobertura de neve no Planalto da Torre desde o final do século XIX. A reconstrução aponta para uma redução do número de dias com neve de aproximadamente 17 0 par a 120 dias nas últimas décadas. As projeções futuras, baseadas em diferentes cenários climáticos, indicam uma redução continu ada e significativa da cobertura de neve, podendo esta desaparecer por completo até ao final do século em cenários de elevadas em is sões. No cenário mais extremo, o Planalto da Torre poderá apresentar condições de cobertura de neve semelhantes às atualmente observadas a 900 metros de altitude. Palavras - chave: Neve , alterações climáticas , deteção remota , Serra da Estrela. HIGHLIGHTS Snow cover at Penhas Douradas declined by 5.4 days per decade since the 1950s. A warming trend of +0.17 °C per decade was recorded from 1883 to 2020. The snow season now starts one month later and ends one month earlier, with major losses in December, Ja nuary, and March. Torre Plateau snow cover dropped from approximately 170 to 120 days per year since the late 19 th century. All scenarios except SSP1 - 2.6 project less snow on Torre Plateau than current levels at Penhas Douradas. Recebido: 9 /0 9 /2024. Aceite: 15 / 05 /202 5 . Publicado: 21/07/2025. 1 Centro de Estudos Geográficos, Instituto de Geografia e Ordenamento do Território , Universidade d e Lisboa , Edíficio, IGOT, R. Branca Edmée Marques, 1600 - 276 , Lisboa , Portugal . E - mail: carlamora@edu.ulisboa.pt , vieira@edu.ulisboa.pt 2 Laboratório Associado TERRA
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 2 1. INTRODUCTION Snow is a key element of Earth's cryosphere with a major impact on climate at the global level and representing an important natural resource. Snowmelt provides drinking water to approximately 17% of the world’s population and is used for hydropower, agric ulture, and industry (Bormann et al ., 2018). Despite the limited observational data from mountains, a significant decline in Northern Hemisphere spring snow cover has been reported, with the largest decreases over higher latitudes and in mountain regions, a fact strongly associated to atmospheric warming (Bormann et al ., 2018). The amplified effect of the decrease of snow and ice at high latitudes is well - known, but there is growing evidence that the rate of warming is also amplified with elevation and, as such, mountain environments are and will suffer more drastic changes under a warming climate (Pepin et al. , 2015). Changes in the duration, extent, and timing of the onset and melting of snow cover have major impacts in mountains (Liston & Hiemstra , 2011). Modifications on the snow albedo and snow melt also influence energy fluxes and mountains’ local and microclimates (Brun & Pomeroy, 2001). The duration of water storage in the snowpack affects soil moisture and catchment hydrology (Diffenbaugh et al. , 201 3; Brown, 2019) and impacts water availability for consumption, agriculture, tourism, and hydropower production (Barnett et al ., 2005; Bormann et al ., 2018). The changes in mountain snow cover also impact the sensitive alpine and sub - alpine ecosystems. The impacts of climate change on the cryosphere in mountains in warm and dry mid - latitudes, such as the Mediterranean, is highly relevant. These have been shown in several mountain regions where snowmelt is the main component of the surface water resources du ring the summer, such as the Spanish Sierra Nevada (Perez - Palazón et al ., 2018), the eastern Pyrenees (López - Moreno et al ., 2009) and the Atlas Mountains (Marchane et al ., 2015). Mediterranean mountains show hotspots of high environmental value, such as bi osphere reserves with rare endemisms. They host an exceptionally high number of cold - adapted endemic plants and are exposed to a high risk of biodiversity loss due to climate change (Di Musciano et al ., 2018; Lamprecht et al ., 2021). Species are subject to great stress and need biotic adaptation responses to modifications on climate, such as the decrease of precipitation and the increase of the air temperatures that together give rise to a shortening of the snow cover extent and duration, and ultimately tra nslate to a reduction in summer water availability (Lamprecht et al ., 2021). Increasing variability of precipitation and air temperatures also affect mainland Portugal, as a result of its position in the transition between the Mediterranean basin and the N E Atlantic (Carvalho et al ., 2014; Ramos et al ., 2011; Soares et al . 2017). Most precipitation in Portugal falls as rain and snow below 1500m occurs only in rare events. Hence, snow cover is confined to the highest mountains of northern and central Portugal. The Serra da Estrela, rising to 1993m above sea level is the range where snow cover lasts for longer, but data allowing for its quantification is scarce and studies are lacking. The Estrela has be en traditionally the location in Portugal for snow leisure activities (Andrade et al . 1992; Carvalho 2007), with a small ski resort installed in the plateau between 1850 and 1990m above sea level, which struggles to maintain conditions f or ski practice. Un til the 1970 s, a small ski piste was maintained at approximately 1700m above sea level close to the Penhas da Saúde at Piornos but was then abandoned due to lacking snow. Despite the reduction in snow cover duration, the Serra da Estrela continues to be as sociated to snow in Portugal, with numerous tourists visiting the upper areas (Silva et al . , 2018), which are easily accessible by car on a national road. The elevation of the Serra da Estrela and geomorphology traits marked by the Pleistocene glaciations , together with a long history of grazing, gave origin to a rich biodiversity. The Estrela plateaus are integrated in a Natural Park, are protected by the Natura2000 network and include a Ramsar site under the Convention on Wetlands, as well as biogenetic reserves. They show the last remnants of alpine and sub - alpine ecosystems in Portugal and are home to several important species, such as the Lacerta monticola in the Red List of the vertebrates of Portugal (Oliveira et al ., 2005), several of them depending on the effects of snow cover. Most direct observations on snow cover in the Estrela plateaus ceased in the mid - 1980 s and are currently done only at the meteorological station of Penhas Douradas, which at 1 380m above sea level is now too low for snow resea rch. The snow and its fate, an issue that crosscuts different environment - related disciplines and socio - economic activities, are central issues for a proper management of this sensitive mountain area, which holds a status of both a Natural Park and a UNESC O Global Geopark (Vieira et al ., 2020). This study aims at filling the gap on understanding the evolution of the snow cover and its spatial
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 3 distribution in the Serra da Estrela since the late 19 th century, as well to provide an outlook on its evolution dur ing the 21 st century. Our key objectives are to: i ) provide a characterization of the evolution of the snow cover based on observational data from the mid - elevatio n plateau of Penhas Douradas (1 360m asl ) since the 1950’s, ii ) reconstruct the potential time - series of snow cover for the Penhas Douradas and the upper plateaus of the Serra da Estrela since 1880, and iii ) assess the possible futures of the Estrela snow cover and briefly discuss the potential impacts of the expected changes. To overcome the lack of observational data at high elevation, we analysed the series from the Penhas Douradas observatory, together with remote sensing imagery and climate modelling data. Scarce observational data, coarse - grained re - analysis that poorly represents the upper mountain conditions and a cold season with high cloudiness, limiting the application of multispectral remote sensing, pose significant limitations to the analysis we present here. However, the synergistic exploration of the available data allows for a firs t insight into the recent evolution and fate of the Estrela snow cover. Given this framework, the results need to be used with special care. 2. STUDY AREA The Serra da Estrela is part of the Iberian Central Cordillera and is the highest mountain in mainland Portugal (fig. 1). Its plateaus that rise between 1400 and 1993 m westwards of the Zêzere Alforfa val leys, and between 1400 and 1750 m eastwards, are a key feature of the mountain’s relief, affecting its climate and hydrology. In the Pleistocene c old periods, these plateaus were responsible for fast and extensive glacial inception due to their position just below the equilibrium line altitude (Vieira, 2008; Vieira & Woronko, 2021; Vieira et al ., 2021), and they are key to the present - day snow dynam ics. Fig. 1 Relief of the Serra da Estrela and location of sites mentioned in the text. Water bodies represented in blue. Colour figure available online. Fig. 1 Relevo da Serra da Estrela e localização dos sítios mencionados no texto. As massas de á gua estão representadas a azul. Figura a cores disponível online.
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 4 The mountain massif shows a NNE - SSW orientation and forms a barrier to the prevailing atmospheric circulation that brings moist air masses from the Atlantic into inner Iberia (Mora & Vieira , 2020). The climate is typically Mediterranean, with dry and warm summers and a wet cold semester (Mora, 2010). Daveau et al . (1977) estimated a mean annual pr ecipitation for 1931 - 60 of 2500 mm at the summit of the mountain. Data from the Penhas Dourada s s hows a precipitation of 1817 mm f or 1941 - 1970 decreasing to 1511 mm in 1981 - 2010. Mean annual air temperatures were estimated by Vieira and Mora (1998) to be close to 4°C at the summit in 1941 - 70. Andrade et al . (1992) analyzed snow cover data from 1957 to 1 985 from three meteorological stations: Penhas Douradas (1380m), Penhas da Saúde (1510m) and Lagoa Comprida (1604m). They showed that snowfall events were more frequent from December to March, but only in February the number of snowfall days surpassed the number of rainfall days, highlighting the warm characteristics of the precipitation, even during the winter. Snow cover was characterized by large interannual and intermonthly irregularity, and snow cover was more frequent, above 1800m, lasting for severa l weeks in winter. However, even at that elevation, warm rainy episodes generated extensive snow melting, interrupting the persistence of winter snow cover. The authors reported a median of 26 days with snow cover at 1600m and estimated about 70 days for t he Torre plateau. 3. DATA AND METHODS 3.1. Introduction Studies of snow cover in the Serra da Estrela, as well as in other Portuguese mountains have been hampered by the lack of good data - series and absent data from high elevation sites. To characterize the past changes in snow cover we have analyzed monthly precipitation, air temperature and snow cover duration from 1954 to 2019 using data from the Penhas Douradas Observatory, the highest meteorological station in Portugal still operational (fig. 1). To complement these observations and extend the analysis to higher elevations, we analyzed daily MODIS (Moderate Resolution Imaging Spectroradiometer) optical imagery from 2001 to 2020. This allowed us to generate a dataset with the presence of snow in different sectors of the plateaus. Besides the observational data, we analyzed modelling datasets and compared them to observations. This was done for the ERA - 5 Land reanalysis snow cover p roduct that extends back to 1950 at 9 km spatial resolution, as well as for the FSM - WRF Iberian Peninsula dataset at 10km resolution from 1981 to 2014, that models snow thickness for different elevations using the ERA - Interim (Alonso - Gonz ález et al . , 2019 ). The evolution of the snow cover duration for the upper areas of the Estrela range was estimated by linear regression modelling at decadal intervals using temperature data from the Penhas Douradas, with model results validated for the period with observati on data. This allowed us to reconstruct the snow cover duration in the Torre Plateau since the 1890 s. The future evolution of the snow cover climate scenarios was assessed using climate change scenarios obtained from WorldClim 2.1 until 2080 - 2100. 3.2. Data f rom the Penhas Douradas Meteorological station The highest meteorological station with a long time - series is located at the Penhas Douradas, an observatory of the Portuguese Institute of the Sea and Atmosphere (IPMA), at 1 380m above sea level (fig. 1). Despite its low elevation when compared to the plateaus that range from 1 400 to almost 2000m, it is the only high mountain station in Portugal. Thus, its observational data set is crucial for analyzing the snow cover and for validating the remot e sensing and modeling data that complement our analysis, enabling insight into the snow cover regimes at higher elevations. The Penhas Douradas station is located at the top of the Z êzere valley slope, in the upper Manteigas basin at in the northeastern l imit of the Penhas Douradas plateau . Surrounded by a Pinus sylvestris forest, the site shows different topographic and vegetation conditions from the rest of the Estrela plateaus at similar elevation, which are wide - open areas mostly dominated by heathland . Monthly mean air temperatures and precipitation data is available from 1883 to 2020 for the Penhas Douradas, but the data series has several gaps. The station changed location before being definitively installed at the current site in 1938. From 1882 un til 1898 it was located at the nearby
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 5 Poio Negro at 1450 m above sea level, while from 1899 to 1903, was moved d ownslope to Carvalheira at 1216m . Since 1903, the Penhas Douradas station has been at the current location, but until 1938 the instruments were a t the roof of the observatory building, before being moved to the ground at the current location. Considering the changes in elevation and the location in the vicinity of the current site, we have corrected the annual average temperatures from 1883 to 1898 and from 1898 to 1903 using a lapse rate of - 0.6 ºC/100m as estimated by Mora (2006) from a regional network of meteorological stations. At sub - daily time scales and mainly under stable anticyclonic conditions and in the early morning, temperature inversio ns have been recorded in the Serra da Estrela, especially in valleys and glacial cirques ( Mora, 2009 , Mora et al ., 2001). However, these are not long lasting and have a minimal impact in the annual averages. No corrections were made for the temperatures from 1903 to 1938, neither for the precipitation. The Penhas Douradas station provides data on the daily snow depth, a series that we have obtained from the IPMA for 1954 to 2019. From it, we calculated the number of days with snow c over by considering the days with more than one cm of snow in the ground. After 2004 , the series has many short gaps without observations. These concentrate in weekends or following high snowfall events, a pattern likely due to the fact, that the meteorolo gical observer, which until then lived at the observatory, since 2004 commutes daily to the observatory and in deep snow cannot reach it (oral information). To address the missing snow cover data in Penhas Douradas from 2004 to 2019, we focused on the per iod from October to May, when snow can potentially cover the ground for more than one day. This analysis resulted in 1,517 days without observations. We then examined these days to identify those with snow cover using visible MODIS imagery on cloud - free da ys, along with air temperature and precipitation records: 1. Periods between observed days without snow, which showed no snow in MODIS imagery or had positive temperatures during cloudy conditions, were classified as snow - free. 2. Periods of a few days between o bserved days with snow were classified as snow - covered. 3. Longer periods that were only preceded or only followed by snow cover were analyzed using MODIS imagery. If the scenes were cloud - covered, half of the period was classified as snow - covered and the oth er half as snow - free. These periods never exceeded two days and were very rare, accounting for a potential error below 1.6% of all missing data. This approach resulted in a good - quality dataset for the period with gaps, which was 2004 to 2019. All trends p resented were calculated using the Theil - Sen slope method and the statistical significance evaluated using the Mann - Kendall test. 3.3. Optical remote sensing data Optical satellite imagery allow for the accurate mapping of snow cover, both using visual analysis of single scenes or by applying multispectral indexes (Rees, 2006 ). The MODIS Terra and Aqua satellites provide visible daily imagery at a resolution of 250m since 2001. Optical sensors are affected by cloudiness, a very important limiting factor in the Serra da Estrela, where cloudy conditions are frequent during the cold semester. Hence, optical satellite imagery is useful only in clear sky conditions. Synthetic Aperture Radar (SAR) satellite data may be an alternative (Mora et al ., 2017), but lo ng data series of SAR at high resolution, such as Copernicus Sentinel 1 are still lacking, since these have only been active since 2015. Other SAR satellites show either low spatial resolution (passive microwave) or lack continuous acquisitions at the glob al scale. Even though MODIS imagery presents the above - mentioned limitations, it is still the best satellite data to complement the observations from the Penhas Douradas meteorological station from 2001 to 2020, allowing to provide spatial and temporal i nformation about the snow cover regime in the Serra da Estrela plateaus. Hence, in addition to using MODIS to fill data gaps at the Penhas Douradas Observatory series as explained above, we used the daily visible imagery dataset to identify the spatial dis tribution of snow cover in the Estrela plateaus from 2001 to 2020. The days were classified based on the presence of snow as follows (fig. 2): 1. Snow present only in the Torre Plateau. 2. Snow present in the Lagoa Comprida Plateau and above. 3. Snow present in both the Penhas da Saúde and Lagoa Comprida plateaus and above. However, the data - series show long periods with cloud cover that limit the ground observations.
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 6 Fig. 2 Examples of MO DIS RGB composite scenes at 250 m resolution for the Serra da Estrela us ed for snow extent classification: A) Snow cover in the Penhas da Saúde plateau and above; B) Snow cover in the Lagoa Comprida and above; C) Snow cover in the Torre Plateau. Colour figure available online. Fig. 2 Exemplos de imagens compósitas RGB de sat é lite MODIS com resolução de 250 m para a Serra da Estrela, usadas na classificação da cobertura de neve: A) Presença de neve nas Penhas da Saúde e no planalto superior; B) Presença de neve na Lagoa Comprida e áreas mais altas; C) presença de neve no planal to da Torre. Figura a cores disponível online. Source: https://wvs.earthdata.nasa.gov/ These affect 48% of the days and in some cloudy winters, MODIS imagery provides essentially sample glimpes of the snow cover. In order to be able to make best use of the exiting MODIS imagery, we applied the following approach to fill the gaps without data: 1. If snow cover was observed before and after the cloudy period, all cloud - obscured days were counted as snow - covered. Hence, in these situations, our classification scheme is based on the assumption that in these conditions, the period was fully snow - covered. 2. If the cloudy period wa s preceded by a snow - free day and followed by a snow - covered day (or vice versa), the snow - covered days were counted as half the duration of the cloudy period. These periods were very rare and account only for about 3% of cloudy days. 3.4. ERA - 5 Land reanalysi s data Reanalysis products generate physically modeled data from multi - source observations from across the world into a global gridded dataset. ERA5 - Land covers the period since 1950 at 9km resolution, with its core being the tiled ECMWF Scheme for Surfac e Exchanges over Land incorporating land surface hydrology (H - TESSEL), using the version CY45R1 of the IFS ( Muñoz - Sabater et al ., 2021 ). Despite the coarse spatial resolution and the difficulty to accurately resolve the relief of mountains, such as the Serra da Estrela, the fact that this is a plateau mountain, reduces the terrain complexity. Here, we have evaluated the ERA5 - Land data feasibility for the reconstruction of the snow cover and to complement observationa l datasets. We have used the snow cover product at 00 UTC, referring to the percentage covered area within the cell within which the coordinate 40.3205 ºN, 7.613ºW is located. The number of monthly and annual days with snow cover was classified by counting the days with: i) more than 0%; ii) more than 5%; iii) more than 10% and iv) more than 25% area covered by snow, within the cell. From these, we selected the one better describing the snow cover duration in the Estrela plateaus. This was done by correlati on analysis with observational data from the Penhas Douradas observatory (1954 - 2019) and from the MODIS analysis (2000 - 2019). 3.5. FSM - WRF Iberian Peninsula snow depth at 10km dataset Alonso - Gonz ález et al. (20 20 ) developed the FSM - WRF dataset of daily snow d epth and snow water equivalent at 10km resolution for the Iberian Peninsula for 1980 - 2014. The dataset was produced using the MODIS data for 2000 - 2014 and a physically based snow model the Factorial Snow Mod el (FSM) driven by the WRF Weather Research a nd Forecast model. The results were validated
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 7 using observations from meteorological stations in Spain. Using a k - means clustering, four main types of snow cover regime were shown to characterize the Iberian mountains. The two plateaus of the Serra da Estr ela show up as the only areas in Portugal classified as cluster 3, which encompasses mountains with significant snow cover in most cold seasons, but with the possibility of occurrence of years with shallow snowpacks. The accurate identification of the Est rela plateaus in this regional model led us to evaluate the FSM - WRF dataset to better assess the snow cover characteristics. The dataset provides data for different elevation levels at 100m intervals. We have selected a cell based on a data point at the Al to da Torre and analysed the daily snow - depth data for 1500, 1600, 1700, 1800, 1900 and 2000m above sea level. The series was transformed into daily occurrence of snow using thickness thresholds of >0, >2 and >5 cm, which were compared with data from PD to assess the possibility to use FSM - WRF data for improving the characterization of the Estrela snow cover. 3.6. WorldClim 2.1 WorldClim 2.1 is a global - grided dataset of downscaled climate data for past and future climate. For the past data, the dataset uses interpolation from weather station data using thin - plate splines and covariates such as elevation, distance to the coast, and MODIS - deri ved maximum and minimum land surface temperatures, as well as cloud cover (Fick & Hijmans, 2017). Mean air temperature, which is the variable we used here, have shown low RMSE values and are accurate, especially in regions with a high density of meteorolog ical stations (Cerasoli et al ., 2022; Fick & Hijmans, 2017). Future scenarios in WorldClim 2.1 have been downscaled from CMIP6 Earth System Models. Here, we have used the EC - Earth3 - Veg model (D öscher et al ., 2022) mean annual air temperature data for the periods 2040 - 60 , 2060 - 80 and 2080 - 2100 at a 30sec grid (c. 1 km at the equator). We use the standard socioeconomic pathways (SSP) 1 - 2.6, 2 - 4.5, 3 - 7.0 and 5 - 8.5. SSP5 - 8.5 is the so - called fossil - fueled development scenario at the upper end of the warming pa thways, with high emissions (Meinshausen et al ., 2020; O’Neill et al . , 2016). SSP3 - 7.0 is based on nationalism driving policy and a focus on regional and national problems, rather than on global development. It represents the medium to high end of the range of pathways. SSP2 - 4.5 is an intermediate scenario, which corresponds approximately to the pre vious RCP4.5, and includes moderate conditions for land use change and aerosols. SSP1 - 2.6 represents the low end of the pathways and is a sustainability - based pathway, with policies addressing human well - being, clean energy, and nature conservation. It rep resents values below 2ºC warming at 2100 as compared to pre - industrial levels (O’Neill et al ., 2016). The data was collected from the bioclimatic variables data set of WorldClim 2.1 . However, ssp370 for 2041 - 2060 and ssp245 for 2081 - 2100 had missing data. Hence, for those two scenarios, we used the climate variables to average the mean temperatures from the monthly maxima and minima. 4. RESULTS 4.1. Climate and snow cover evolution in the Penhas Douradas Observatory 4.1.1. Air temperature and precipitation in Penhas Douradas from 1883 to 2020 A clear warming trend is present in the mean annual air temperature (MAAT) in the Penhas Douradas meteorological station for the last circa 140 years (1883 to 2020) (fig. 3). MAATs in the early 1900 ’s were around +8.6ºC and in 2019 were close to +10.6ºC. Despite the trend, three periods can be distinguished. From 1883 until around 1949 the temperatures increased, with the warmest period being between 1945 and 1949. The following three decades were do minated by a cooling trend that lasted until around 1972. MAATs in the 1970’s were close to those of the 1920’s, but since then temperatures started to increase at fast pace, until they peaked at 11.6ºC in 2017. The warming trend from 1883 to 2020 in the P enhas Douradas was +0.164ºC/decade with a p - value <0.0001.
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 8 Fig. 3 (A) Mean annual temperature , (B) total annual precipitation , and (C ) Mean annual air te mperatures (MAAT) per decade in the Penhas Douradas meteorological station from 1884 t o 2020 ( MAAT corrected for site changes of the observatory before 1938). Colour figure available online. Fig. 3 (A) Temperatura média anual , (B) precipitação anual e (C) temperatura média anual por década na estação meteorológica das Penhas Douradas entre 1884 e 2020 ( a temperatura média foi corrigida em função da altitude devido a modificações no local de observação antes de 1938). Figura a cores disponível online. The decadal data evidences the period of 1941 - 50 as anomalously warm, interrupting a series that otherwise would have been stable to slightly warming from the beginning of the 20 th century until 1971 - 80 (fig. 3). Since 1981 - 90 , all the decades are warmer than the previous and the warming rate is steeper . The precipitation regime needs to be analyzed w ith care , especially before 1938 due to the changes in location in the Penhas Douradas observatory before that year. The data shows high interannual variability, with a decrease in annual totals. Pre cipitation values averaged c. 2300mm in 1920 - 40, c. 2000 m m in 1941 - 50 and cont inued decreasing until around 1 600 to 1400 mm since the 1980 s. However, the high values before 1938 are possibly biased due to changes in station location, with the trend since then being of - 51.16mm/decade with a p - value of 0.014. 4.1.2. Sno w cover in Penhas Douradas from 1954 to 2020 Penhas Douradas is the only meteorological observatory at high elevation in Portugal with a long data - series for snow cover. The data from 1954 to 2020 shows a very strong reduction in snow cover, with a trend of - 5 days/decade (at p<0.00003), from an aver age of 53 days in 1951 - 1960 to 28 Days in 2011 - 2020 (table I). The data shows a period with longer duration of snow cover from 1969 to 1976, which is reflected in a peak with an average of 68.6 days in 1971 - 1980. A sharp change occurred in the following de cade, with a strong reduction to an average of 34 days in 1981 - 1990. Since then, the reduction rate was slower at - 2 days/decade, with some years in the early 2000 s and in the mid of the decade of 2011 - 2020 showing longer snow cover (fig. 4).
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 9 Table I Mean annual air temperature and mean annual snow cover days per decade in the Penhas Douradas, from 1951 to 2020. Quadro I Temperatura media anual e média anual do número de dias com cobertura de neve por década nas Penhas Douradas entre 1951 e 2020. Mean annual air temperature (ºC) Number of days with snow cover 1951 - 60 8.8 53.3 1961 - 70 9.1 47.0 1971 - 80 8.6 68.6 1981 - 90 9.5 33.6 1991 - 00 9.6 32.5 2001 - 10 9.9 31.0 2011 - 20 10.4 28.0 Source: Instituto Portugu ês do Mar e da Atmosfera, with data estimated for snow cover after 2004 Fig. 4 Number of days with snow cover at Penhas Douradas meteorological station from 1954 to 202 0. Upper graph: annual values. L ower graph: decadal box - plots with D1, Q1, Median, Q3 and D9. Colour figure available online. Fig. 4 Número de dias com cobertura de neve na estação meteorológica das Penhas Douradas de1954 a 2020. Gráfico superior: valores anuais. G ráfico inferior: box - plots decenais com indicação do D1, Q1, Mediana, Q3 e D9. Figura a cores disponível o nline. Source: Instituto Portugu ês do Mar e da Atmosfera, modified after 2004 4.1.3. Seasonal changes in snow cover Besides the reduction trend in the number of days with snow cover at Penhas Douradas, the monthly distribution has also changed since 1954 (fig. 5). September and October have always been months of very scarce snow, but while until the late 1960 ’s snow cov er occurred for a few days per year in September, it disappeared after that.
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 10 Fig. 5 Number of days with snow cover in the Penhas Douradas Observatory from 1954 to 2020. Colour figure available online. Fig. 5 Número de dias com cobertura de neve na estação meteorológica das Penhas Douradas de 1954 a 2020. Figura a cores disponível online. Source: Instituto Português do Mar e Atmosfera, corrected in the period 2004 - 2020 October frequently recorded a few days of snow cover in the beginning of the ser ies, but after the 1980 s became a month with very rare occurrences, starting to mirror the characteristics of September at the beginning of the 1950s (fig. 5). These are two months with still warm ground and sporadic snowfall events. November had always a small number of days with snow cover (decadal averages from 1.5 to 4.4 days) and high interannual irregularity. No trend for this month is visible, meaning that sporadic snow events continued to occur during the whole period, but that snow did not last for more than a few days, independently of the year (fig. 6).
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 11 Fig. 6 Decadal variability of days with snow cover for the cold season months in the Penhas Douradas Observatory. Decadal box - plots with D1, Q1, Median, Q3 and D9. Colour figure available online. Fig. 6 Variabilidade decenal do número de dias com cobertura de neve para os meses da estação fria no observatório das Penhas Douradas. Box - plots decenais com D1, Q1, Mediana, Q3 e D9. Figura a cores disponível online. Source: Instituto Portugu ês do Mar e da Atmosfera, modified after 2004 December showed marked changes in the 1950’s, with two distinct periods (fig. 5). Until the mid - 1970s, the ground was frequently snow - covered for about two weeks, but after that, the snow cover showed very short duration lasting only for a few days each year. This became clearer after 1980 and resulted in a remarkable reduction of the snow during that month (fig. 6). January has been a very irregular month and is characterized by its interannual irregularity (fig . 6), having recorded full snow cover in 1971/72 to 1973/74 (fig. 5). In the last two decades, it became more irregular and even lacked snow cover in some years after 2000. In the second decade of the 21 st century January recorded a regime comparab le to December in the late 1970 s. February, similar to January was marked by high irregularity, but since 2002 it clearly became the month with more snow cover (figs. 5 and 6). March was a very irregular month, influenced by sporadic snowfall events. Its genera l behavior during the period is comparable to December, reflecting the strong reduction of the snow cover in the shoulder seasons (fig. 5). However, March reflects a much stronger reduction in snow cover after the 1980's, than any other month (fig. 6). Apr il has been a month of sporadic snow and despite the general reduction, its behavior is very irregular (fig. 5), depending on the advection regime of spring cold and wet air masses. It seems to be losing its original characteristics to March (fig. 6). May has been typically a month with little snow, peaking in the cold early 1970s, but with negligible snow cover days before and after (fig. 5 ). April in the 2010s becam e comparable to May in the 1970 s, which may mean a fast shift towards a Mayification of Ap ril. The warm season, lasting from June to September is a regular period without snow cover in the Penhas Douradas. In synthesis, the Penhas Douradas snow cover regime has been marked since the mid - 20 th century by an increasing interannual irregularity and by a shortening of the snow period. Snow cover at the observatory only approached what could be called a seasonal snow cover from 1971 to 1974, but otherwise, snow rarely covered the ground for over a month in a row. The cold period of the early 1970 s is remarkable on its effects on the duration of snow cover, but overall, a trend for a strongly decreasing snow cover is clear during the whole time - series. The season with snow on the ground starts about one month later and ends one month earlier. November a nd April, show small reduction
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 12 trends, with December, March and January being the months showing larger reduction in snow cover duration. 4.2. Attempting to evaluate the recent evolution of snow cover from satellite imagery (2001 - 2020) The analysis of the MOD IS optical imagery at 1 - day intervals and 250m spatial resolution allowed insight on the snow cover duration at the plateaus of the Serra da Estrela from 2001 to 2020. This is currently the best option for reconstructing the spatial extent of snow in the E strela plateaus, but it is an approach that is based on continuity assumptions during cloudy periods, which implies that it needs to be used with extreme care, since cloudy days are very frequent in the cold season. Figure 7 represents the reconstructed da ys with snow cover at least down to the Penhas da Sa úde Plateau (circa 1500 m), down to the Lagoa Comprida upper Penhas Douradas Plateau (circa 1600 m) and the days with snow only in the Torre Plateau (circa 1800 m). Similarly to what has been obser ved in the Penhas Douradas at 1 360m, the observations show the high interannual irregularity of the snow cover. In some years, the Torre plateau showed continuous snow cover for several months, such as 2008/09 with snow from late November to early May, or 200 5/06 with snow from late November to mid - April. However, in several years, especially towards the end of the series, as in the winters of 2011/12 and 2019/20, snow lasted for only a few days even at upper areas of the Estrela. Fig. 7 Reconstructed snow cover in the Serra da Estrela plateaus (from MODIS imagery analysis, 2001 - 2020). Colour figure available online. Fig. 7 Reconstrução da cobertura de neve nos planaltos da Serra da Estrela a partir da análise de imagens MODIS ( 2001 - 2020). Figura a cores disponível online. The observations show a trend for shorter snow seasons, a significant delay in the onset of snow cover and for less days of snow at lower elevations, as well as a regime where years with seasonal snow alternat e with years with sporadid snow, even at the Torre Plateau (fig. 7). While the Torre Plateau may have snow cover at the beginning of the cold season, snow lasting for several days down to Penhas da Saúde before mid - January almost ceased to occur in the las t decade. Typically, longer lasting snow cover starts from higher to lower elevation, but with the delaying of the snow onset to January - February, several years showed an abrupt initiation of multi - day snow cov er at all elevations above 1500 m. This may be also related with the warmer ground until later in the cold season, limiting snow accumulation or by later and scarcer snow events. The end of the snow period was generally more gradual and started from lower to higher elevation. Snow did not last long at the Penhas da Saúde plateau after mid - March, especially in the last decade. In April, snow cover was frequent only in the Torre Plateau, but in some years, it extended down to Lagoa Comprida.
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 13 In rare years, April has shown snow down to Penhas da Saúde for several days, but these have been mainly years of late snow occurrence and probably marked by cold snow events, with snow cover benefiting from the cool er ground from a preceding snow - free period. May showed very rare snow cover. The data clearly shows that despite the delay in the onset of the snow cover, the start of the snow - free period did not change and concentrates in April. The period 2011 - 2020 has shown more snow in April than the preceding decade. 4.3. Evaluation of climate models for reconstructin g the snow cover 4.3.1. Snow cover reconstruction from ERA5 - Land ERA5 - Land provides meteorological data since 1950 for a global grid with 9km resolution. Here we evaluate its application to the Serra da Estrela by comparing ERA5 - Land data with observations from the Penhas Douradas and with the MODIS snow cover series retrieved for the plateaus. From the ERA5 - Land, we calculated the annual days with snow cover for different limits of snow - covered area within the grid point: >0, >1, >5, >10 and >25% area. The comp arison of the resulting series with the number of days with snow cover at Penhas Douradas resulted in r from 0.57 to 0.63, significant at p<0.05 (table II and fig. 8A). Correlations with the MODIS series for the Torre Plateau are worst (0.32 to 0.41), but still statistically significant. Table II Correlations (r) between the number of days with snow cover at Penhas Douradas and at the Torre Plateau, with the ERA5 - Land days with snow cover above different area thresholds (>0%, >1%, >5%, 10% and >25%), using annual days and mean annual days per decade. Quadro II Correlações (r) entre o número de dias com cobertura de neve nas Penhas Douradas e no planalto da Torre, com o número de dias acima de vários limiares de cobertura de neve (>0%, >1%, >5%, 10% e >25%) da ERA5 - Land. São apresentados o número anual e a média por década. ERA5 - Land Snow Cover days (area) Penhas Douradas days with snow cover (Observations) Torre Plateau days with snow cover (MODIS) Annual Decadal Annual >0% 0.57 0.97 0.41 >1% 0.62 0.96 0.37 >5% 0.63 0.93 0.33 >10% 0.61 0.92 0.33 >25% 0.59 0.91 0.32 Fig. 8 Correlations between ERA5 - Land of snow cover days and observations from the Penhas Douradas Observatory. (A) ERA5 - Land days with more than 5% area of snow cover (annual totals). (B) ERA5 - Land days with more than 0% area of snow cover (decadal averages). Fig. 8 Correlações entre o número de dias com cobertura de neve da ERA5 - Land e as observações d as Penhas Dou radas. (A) Dias com mais de 5% da área coberta por neve na ERA5 - Land (totais anuais). (B) Dias com mais de 0% da área coberta por neve (médias decenais).
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 14 The analysis of the decadal means is useful for evaluating long - term scenarios. The comparison of ERA5 - Land snow cover days for the Serra da Estrela with the snow cover days at the Penhas Douradas observatory resulted in very strong positive correlations from 0.91 to 0.97 (t able II ), while the data for the Torre Plateau did not provide statistically signif icant results. The MODIS results cannot be used for the decades since the series started only in 2000. Figure 8B shows the example of the correlations between ERA5 Land days with snow presence at the grid and the observations from Penhas Douradas. It is, however, important to notice that the high correlations may be a result of the overall control of mean annual air temperature on the snow cover duration, rather than of the real performance of ERA5 - Land in modeling the snow cover dynamics. Actually, as we show below, at the decadal level, colder periods have resulted clearly in longer - lasting snow cover. 4.3.2. FSM - WRF Modeled snow cover duration (1980 - 2014) The FSM - WRF modeled snow cover thickness is a dataset gridded at 10 km resolution for the Iberian Peninsula from 1981 to 2014, for 100m elevation intervals, from which we calculated the number of days with over 0, 2 and 5cm of snow depth. We assessed the s eries performance for the Serra da Estrela from 1500 to 2000m elevation by comparing it with the Penhas Douradas observations and with the MODIS - derived dataset. Statistically significant correlations at p<0.05 were found for elevations above 1900 m, but no t for all variables (table III). The best correlations were obtained for the Torre Plateau snow cover from MODIS, with a correlation of 0.79 for days over five cm of snow (fig. 9). The decadal means are not analyzed due to the short time - series of the FSM - WRF. Table III Correlation (r) between the observed number of days with snow cover in Penhas Douradas (PD_Snow), Torre Plateau (TSnowRS), above the Lagoa Comprida (PTSnowRS) and above the Penhas da Saúde Plateau (PSSnow_RS), with the number of days with snow depth (SD) above 0, 2 and 5 cm at 2000 and 1900m, from the FSM - WRF model (1981 - 2019). Correlations at p<0.05 in bold. Quadro III Correlações (r) entre o número observado de dias com cobertura de neve nas Penhas Douradas (PD_Snow), Planalto da Torre (TSnowRS), acima da Lagoa Comprida (PTSnowRS) e acima do planalto das Penhas da Saúde (PSSnow_RS), com o número de dias com espessura de neve (SD) maior que 0, 2 e 5cm a 2000 e 1900 m de altitude. Resultados baseados no modelo FSM - WRF (1981 - 2019). Correlaçõ es p<0.05 a negrito. SD FSM - WRF PD_Snow TSnowRS PTSnow_RS PSSnow_RS > 0 cm (2000 m) 0.32 0.50 0.65 0.47 > 2 cm (2000 m) 0.40 0.65 0.77 0.59 > 5 cm (2000 m) 0.41 0.68 0.79 0.61 > 0 cm (1900 m) 0.22 0.40 0.53 0.42 > 2 cm (1900 m) 0.16 0.47 0.56 0.43 > 5 cm (1900 m) 0.16 0.43 0.51 0.38 4.3.3. Modeling snow cover duration at Penhas Douradas a) Mean annual air temperature and snow cover The correlation of the mean annual air temperatures (MAAT) with the number of days with snow cover in the Penhas Douradas shows a r of - 0.66 at p<0.05, reveals both the control of air temperature on snowfall, but also on melting (fig. 9). Based on this correlation, we approach reconstructing the Estrela snow cover series along the 20 th century. This implies the assumption of a similar climatic regime during the period of analysis, especially on the seasonality of precipitation. Since observational and model data show that precipitation has been decreasing and temperature increasing duri ng last century (fig. 3), our approach is based on the broad assumption that present conditions of higher temperature and lower precipitation at high elevation, may be comparable with past conditions at lower elevations. To improve the statistical signifi cance, we reconstructed long - term snow cover trends using the correlation of the decadal averages of the MAAT and snow cover. For Penhas Douradas, the time series from 1955 to 2019 was used, but the values are constrained by the low elevation of the site, not showing large numbers of days with snow nor very low MAAT. To overcome this problem, we have included in the regression observations from days with snow cover from the MODIS reconstruction
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 15 for the decades of 2001 - 2010 and 2011 - 2020, paired with extrapo lated MAAT for the Torre Plateau (1800 m), Lagoa Comprida (1600m) and Penhas da Sa úde (1500m). This allowed for better constraining of the best - fit line (fig. 10), resulting in a r value of - 0.97 at p<0.05, allowing for high confid ence in the application of the equation: Snow cover days (decadal mean) = - 27.114 MAAT + 298.89 (Eq. 1) . The plateau - dominated relief and smooth summit surfaces of the Serra da Estrela favor the applicability of this elevation - based model, but it is important to note that small t opographical effects that depend on slope and aspect are not resolved with this approach. Fig. 9 Mean annual air temperature and number of days of snow cover at the Penhas Observatory from 1954 to 2020 . Colour figure available online. Fig. 9 Temperatura média anual do ar e número de dias com solo coberto de neve no observatório das Penhas entre 1954 to 2020. Figura a cores disponível online. Source: Instituto Portugu ês do Mar e da Atmosfera, modificado a partir de 2004 Fig. 10 Correlation between the decadal mean annual air temperatures and days with snow cover based on observational data from the Penhas Douradas for 1954 - 2019, and on MODIS reconstructed snow cover and extrapolated decadal mean annual tempe ratures for 1500, 1600 and 1800 m for 2001 - 2020. Colour figure available online. Fig. 10 Correlação entre as médias decenais da temperatura média do ar e o número de dias com solo coberto de neve, baseada nos dados do observatório das Penhas Douradas (1954 - 2019) e na análise de imagens MODIS e extrapolação da temperatura média anual decenal p ara 1500, 1600 e 1800 m de altitude (2001 - 2020). Figura a cores disponível online. b) Reconstruction of the snow cover duration at the Alto da Torre We used the mean annual air temperatures fo r the Alto da Torre at 1993 m asl estimated at the decadal level t o reconstruct the number of days with snow since 1883 (fig. 11). Although the results need to be examined with care, they allow for insight into the likely evolution of the snow cover duration in the upper reaches of the Serra da Estrela. Following atmosph eric warming, snow cover duration has been decreasing from around 170 days in the late 19 th century, to about 160 days from
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 16 1951 to 1980, and sharply declined to an average of about 120 days in the last decade. The last decade also showed increased irregularity with some winters showing barely any snow cover, which according to the estimations, had never occurred since 1884. The decade of 1941 - 50 showed anomalously low snow, while the cold 1970s showed values like those of the first half of the 20 th century. The model shows a reduction of almost two months in snow cover duration in the Alto da Torre since the late 19 th century (fig. 11). Figure 11 Estimated duration of the snow cover in the Alto da Torre from 1884 to 2020. Figure 11 Duração estimada da cobertura de neve no Alto da Torre entre 1884 to 2020. 4.3.4. Snow cover at the end of the 21 st Century The projection of the mean annual air temperatu res for the Torre Plateau (1900 m asl) until 2100 on the EC - Earth3 - Veg under different socioeconomic pathways and WorldClim 2.1, shows very high warming ( f ig. 12). Fig. 12 Modeled decadal averaged mean annual air temperatures (A) and mean days with snow cover (B) at the Torre Plateau (1900m) from 1880 to 2020 and scenarios following different Standard Socioeconomical Pathways until 2100. The scenarios are 20 - year means from the EC - Earth3 - Veg (WorldClim2.1). Colour figure available online. Fig. 12 Modelos das médias da temperatura média anual por década (A) e do número médio de dias com cobertura de neve (B) no planalto da Torre (1900 m) de 1880 a 2020, segundo os cenários Standard Socioeconomical Pathways até 2100. Os cenários usados são médias de 20 anos do EC - Earth3 - Veg (WorldClim2.1). Figura a cores disponív el online.
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 17 Temperatures at the end of the century may reach around 13 ºC under ssp5 - 8.5, 11.6ºC under ssp3 - 7.0, 10.2ºC under ssp2 - 4.5 and 8.6ºC under ssp1 - 2.6, being only in the latter that a slight cooling is expected to start in 2080 - 2100. Accounting for the assumptions presented above and for the limitations in the modeling, the past and future evolution of the snow cover days for the Torre Plateau can be estimated using Equation 1. The results show that the number of snow cover days at the Torre Plateau will be 0 around 2070 under ssp5 - 8.5 and 10 years later under ssp3 - 7.0 (fig. 12). Ssp2 - 4.5 shows about 20 days of snow by the end of the century, a value below today ’s Penhas Douradas, while ssp1 - 2.6 shows a stabilization at around 60 days o f snow cover, which is a reduction to almost half of present - day’s values, even in the best climate scenario. 5. DISCUSSION 5.1. The evolution of snow cover in the Penhas Douradas since the 1954 Snow cover since the mid - 20 th century in the Penhas Douradas station at 1380m shows a clear declining trend of c. 5.4 days/decade, with a peak in the late 1960 s to early 1970s. The records show change in average number of days with snow cover from 53 days in 1951 - 1960 to 28 d ays in 2011 - 2020. Similar reductions in snow are widespread in other mountain regions in Europe and the World ( Intergovernmental Panel on Climate Change , 2019) and follow the atmospheric warming that has been occurring since the industrial revolution (Carvalho et al ., 2014, Ramos et al ., 2011, Soares et al ., 2017) . Since the 1980 s , the decline in snow cover duration has occurred at a slower pace in the Penhas Douradas, despite the continued warming. This seemingly relates to the elevation of the observatory being just high enough to record snowfall in cold meteorological events, but not high enough to maintain the snow in the ground for longer periods. The 1980s seem to have been the transition decade to these new conditions, where snow lasting for several weeks c eased to occur. This hypothesis needs to be further tested by analyzing the events with snowfall and the changes in synoptic conditions causing them. On the other hand, the reduction in snow cover days in December and January, may also be controlled by the warmer ground temperatures in late Fall and early Winter, not allowing for snow to be maintained before, with snow onset moving towards February. 5.2. The evolution of snow cover in Serra da Estrela since the late 19 th century Most of the area of the Serra d a Estrela plateaus is above the Penhas Douradas Observatory and lacks observational data, while those are also the areas where snow is more relevant for the ecosystems and as a natural resource. The lack of observations in the plateaus is a major issue tha t has been seriously affecting the assessment of the impacts of the changing snow cover in the Estrela. Here, we tried to exploit data available, by using data from the Penhas Douradas observatory, from remote sensing imagery and from reanalysis. However, the joint use of these sources for the reconstruction of the snow cover duration time - series was constrained by the low elevation of the observatory, by the short time - series of remote sensing imagery and by problems with clouds affecting the optical image ry. Furthermore, the reanalysis - based models showed limited ability for accurately resolving the topography of the Estrela. The comparison of observations from the Penhas Douradas with remote sensing imagery and reanalysis data at the monthly and annual le vels, showed limitations, which were only solved using decadal averages to reconstruct the snow, cover duration time - series. The interannual variability was lost, but we were able to capture the overall signal associated with the changing climate conditions. As such, the reconstructions that we present here are supported by the decadal correlation be tween the snow cover duration and mean annual air temperatures and allow a first insight into the recent evolution of the snow cover in the Estrela. The reconstructed time - series for the days with snow cover in the Torre Plateau show 170 days in the late 19 th century, which changed to about 160 days from 1951 to 1980, and sharply declined to an average of about 120 days in the last decade. Interannual variability has increased significantly, with more years with scarce snow towards the end of the period. L autensach (1929) refering to oral information from the meteorological observer of the Penhas Douradas Observatory, indicated that snow in the Torre Plateau lasted for about 8.5 months, from mid - October to late June, while in Penhas
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 18 Douradas it lasted for a bout five months. It is not clear if these values referred to a continuous snowpack, or rather to isolated snow patches. Our reconstruction is more conservative, which suggests that Lautensach’s information may relate to presence of snow patches. The gene ral warming in the Estrela has led to a shortening and a delay of the snow cover season even at the Torre plateau, where MODIS - derived observations for the last two decades show a cold semester marked by irregularity in snow conditions. While some years sh ow snow from December to April, most years show sporadic snow with snow - free periods even in mid - winter at the Torre Plateau. Extreme years like 2018/19 have almost entirely lacked snow cover. If we compare the snow cover in the Alto da Torre in 2011 - 2020 with the conditions in the 1970s, the current conditions are close to those of the upper Lagoa Comprida area (at c.1700m) then, which gives a good approximation of the magnitude of the changes that have occurred. Similar to the Serra da Estrela, a decrease of snow cover has been recorded in other Iberian mountains during the second half of the 20 th and early 21 st centuries ( Bonsoms et al . , 2021; L ópez - Moreno, 2005; L ópez - Moreno et al ., 2020). Annual and winter precipitation in the Iberian Peninsula between 1961 and 2011 decreased at a rate of 18.7mm/decade (Vicente - Serrano et al ., 2017), with snowfall days in the the northern Iberian Peninsula having recorded a 50% decrease since the mid - 1970s ( Lopez - Moreno et al . , 2020 ; Pons et al ., 2010;). The North Atlantic Oscillation (NAO) has been identified as the main driver of precipitation in Iberian Peninsula in winter, as well as of snowfall in the western and central Pyrenees (López - Moreno, 2005; Revuelto et al ., 2012). The predominance of NAO positive phases in winter over the 20 th century has lead to a reduction of weather types associated with precipitation ( Bonsoms et al ., 2021 ; López - Bustins et al ., 2008). Further south, in the Atlas mountain range in North Africa the large interan nual variability of winter precipitation (400%) is affecting the mass balance of the snowpack. The disappearance of many perennial snowpatches in the last decades reflects the warming trend of summer air temperatures since the 1970s (Hughes et al . , 2020). 5.3. The fate of the Serra da Estrela snow cover Our estimates of the evolution of the duration of snow cover for the 21 st century show drastic changes occurring at fast pace. All scenarios, except ssp126 place the Torre Plateau before the end of the century with less snow than today ’s Penhas Douradas (fig. 12). This means that the Serra da Estrela will clearly enter a sporadic snow cover regime, which will result in dramatic changes in the high elevation ecosystems. However, even at the short - term, a lmost al l scenarios show the snow cover duration in 2030 - 2040 to become about half that recorded in the early 1970 ’s, which calls for urgent actions in the management of the plateaus. Snow reduction scenarios are also foreseen in other mountain ranges in Iberia. For example, Pérez - Palazón et al. (2018) modeled snowfall for the end of the century for the Sierra Nevada and found a decreasing tren d in snowfall from 0.21 to 0.55 mm ·year 1 , under representative concentration pathways (RCP) 4.5 and RCP 8.5, respectively. Snowfall days are expected to decrease from −0.068 days·year 1 under RCP 4.5 to −0.111 days·year 1 under RCP 8.5, accompanied by an increase in the torrentiality of snowfall. As an example of the socioeconomic impacts of the changes in snow, Spandre et al . (2019) showed that under RCP8.5 projections, there would no longer be snow ski resort in the French Alps and the Pyrenees (France and Spain) by the end of the century . 5.4. The future environmental setting Serra da Estrela The impacts of the modeled 21 st century warming may be analyzed by plotting future air temperatures as its corresponding changes in elevation. Since precipitation is also forecasted to diminish significantly, a future pseudo - lowering of the Estrela accompanied by a precipitation reductio n, will result in climate conditions closer to current ones in lower mountains in the region, as precipitation is much controlled by elevation in Central Portugal (Daveau, 1977; Mora & Vieira, 2020). For this, we used the altitudinal lapse rate for the Ser ra da Estrela by Mora (2006). Figure 13 shows simulated comparative altitudinal modifications of the Torre Plateau, considering the SSPs and changes in mean annual temperatures compared to present - day. The results show that at 2080 - 2100 under ssp585, the Alto da Torre will hav e MAAT close to those of c. 900 m asl today (Caldas de Manteigas). SSP370 will lead to warming comparable to an elevation of c. 1150 m today (Po ço do Inferno). SSP245 will lead to a Va le do Rossim - like climate (1400 m), and the best
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 19 scen ario, SSP125 to a lowering of about 200 m bringing the summit to the current Lagoa Comprida conditions (1650 m). In 2040, the Torre Plateau should have MAAT between those today at c. 1600m (ssp585) and c. 1750m (ssp126). It reaches present - day ’s Vale do Ro ssim’s (c. 1450m) conditions around 2055 under ssp585, 10 years later for ssp370 and by 2080 for ssp245. Only under ssp126 those conditions will not be met. Fig. 13 Estimated pseudo - lowering of the Torre Plateau during the 21 st century using different air temperature scenarios from the EC - Earth3 - Veg model downscaled in WorldClim2.1. Colour figure available online. Fig. 13 Pseudo - subsidência do Planalto da Torre ao longo do século XXI com base em vários cenários para a temperat ura do ar do modelo EC - Earth3 - Veg obtidos a partir do WorldClim2.1. Figura a cores disponível online. 5.5. Impacts of a warmer and snow - depleted Estrela The Serra da Estrela will suffer drastic warming and snow cover reduction during the 21 st century, with ve ry fast modifications already under way. These changes will have profound impacts on its sensitive ecosystems and land management systems. The plateaus will be drier and much warmer with conditions similar to those of much lower mountains today, such as Se rra da Gardunha or even lower. This lowering of the Estrela will affect the sub - alpine ecosystems of the plateaus, including key conservation habitats in the biogenetic reserve and Ramsar site, such as the peatlands, with high impacts for endemic species. The last 140 years have seen strong changes in the E strela, with a warming of +0.17 ºC/decade and a strong decrease of snow cover. The Torre Plateau is accompanying these changes by moving from a clearly seasonal snow - covered environment to one with sporadic snow and with some years showing almost no snow at all. With the reduction of she pherding in the plateau and with the warming, an expansion of the shrub formations is expected to occur. But the drying and warming will also lead to more extreme soil drought during the summer, increasingly exposing soils to water erosion and degradation during Fall and reducing even more its ecological potential. The impacts of the vanishing snow on the hydrology can also impact water availability. An increase of water retention in the plateaus, which can be properly done having healthier and deeper soil s and a denser protecting vegetation cover, is urgently needed. Conditions must be found to adapt to climate change and properly manage the vegetation in the plateaus, protecting key locations and controlling grazing, while protecting the area from wildfir e. Despite the scarce research on changes in snow cover, the reduction of snow is clear. Several tourism and regional development studies during the last decade have emphasised on the needs for adaptation of the regional tourism economy in order to divers ify and become more resilient to climate change (Mota et al. , 2023). Public policies and private investments have also been following this line, with tourist numbers increasing in all season. Despite more research needed, it is likely that the changes in s now cover will not have a strong impact on the local tourism industry, which is adapting to the new climate scenarios. However, it is important to note that the forecasted disappearance of seasonal snow and climate warming clearly shows that the Torre ski infrastructure has no future and that the investments made in recent years were poorly advised. The ski resort expansion in the last two decades has furthermore resulted in high impacts on the landscape, with urgent action needed for
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 20 renaturalizing the slo pes and for the timely removal of the infrastructure. Similar approaches are being conducted in other mountains, such as the Alps in Austria, France and Italy (Moreno - Gené et al ., 2018; Polderman et al ., 2020). 6. CONCLUSIONS The analysis of the snow cover trends in the Serra da Estrela, specifically at the Penhas Douradas station, reveals a significant decline in snow cover days from the mid - 20 th century to the present, with a reduction rate of approximately 5.4 days per decade. This decline, from an average of 53 snow cover days in the 1950s dro p ping to just 28 days in the 2010s, mirrors similar trends observed in other European and global mounta in regions, largely attributed to ongoing atmospheric warming since the industrial revolution. Despite warming, the rate of decline in snow cover at Penhas Douradas has slowed since the 1980s. This deceleration is likely due to the station's elevation, whi ch is sufficient to record snowfall during cold episodes but not high enough to sustain prolonged snow cover. The 1980s mark a shift to these new conditions, where snow lasting for several weeks has become rare. Further complicating the assessment of snow cover trends in Serra da Estrela is the lack of observational data from the higher plateaus, which are crucial for understanding the full impact of changing snow cover on the ecosystem and natural resources. Attempts to reconstruct snow cover duration usin g data from Penhas Douradas, remote sensing, and reanalysis models faced significant challenges due to the observatory's low elevation, limitations in data resolution and optical imagery completeness. Hence, our observations need to be used with care, espe cially the spatial reconstruction attempt using MODIS for the period of 2001 to 2020. The reconstructed time - series indicate a substantial decline in snow cover days in the Torre plateau, from about 170 days in the late 19 th century to approximately 120 da ys in the last decade, with increased interannual variability and more years experiencing scarce snowfall. Projections for the 21 st century under various climate scenarios predict drastic reductions in snow cover for the Torre Plateau, with some scenarios suggesting conditions by the end of the century will resemble those currently observed at much lower elevations. This anticipated decline will have profound impacts on the high - elevation ecosystems, hydrology, and land management systems of the Serra da Es trela. The sensitive sub - alpine ecosystems, including peatlands and endemic species habitats, will face increased threats from warming and drying conditions, leading to soil degradation and reduced ecological potential. In response to these changes, adapti ve management strategies are urgently needed to protect and enhance the resilience of the plateau's ecosystems. This includes improving water retention through soil and vegetation management, controlling grazing, and mitigating wildfire risks. The foreseen disappearance of seasonal snow covering the ground also calls for a reevaluation of local infrastructure investments, particularly the Torre ski resort, which is unlikely to remain viable. Instead, efforts should focus on renaturalizing impacted areas, fo llowing examples from other mountain regions like the Alps. Overall, the findings underscore the critical need for comprehensive and forward - looking strategies to manage the impacts of climate change on snow cover and associated environmental and socioecon omic systems in the Serra da Estrela. ACKNOWLEDGEMENTS This research is a contribution to the Project ReMonStar (Ref PD23 - 00005) funded by Fundação La Caixa and F undação para a Ciência e a Tecnologia through the PROMOVE program. Thanks are due to the tw o anonymous referee s for the comments which contributed to an improvement of the manuscript. The Instituto Português do Mar e da Atmosfera is thanked for providing d ata from the Pe n has Douradas meteorological station .
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 21 AUTHORS CONTRIBUTIONS Carla Mora : Conceptualization; Methodology; Software; Validation; Formal analysis; Investigation; Resources; Data curation; Writing original draft preparation; Writing review & editing; Visualization. Gonçalo Vieira : Conceptualization; Methodology; Software; Val idation; Formal analysis; Investigation; Resources; Data curation; Writing original draft preparation; Writing review & editing; Visualization. ORCID Carla Mora https://orcid.org/0000 - 0002 - 0843 - 3658 Gon ç alo Vieira https://orcid.org/0000 - 0001 - 7611 - 3464 REFERÊNCIAS BIBLIOGRÁFICAS Alonso - González, E., López - Moreno, J. I., Navarro - Serrano, F. M., & Revuelt o, J. (2020). Impact of North Atlantic Oscillation on the Snowpack in Iberian Peninsula Mountains. Water , 12 (1), 105. https://doi.org/10.3390/w12010105 Andrade, E., Mora, C., Neves, M., & Vieira, G. T. (1992). Desportos de Inverno na Serra da Estrela [ Winter Sports in the Serra da Estrela ]. Finisterra Revista Portuguesa de Geografia, XXVII (53/54), 187 - 193. https://doi.org/10.18055/Finis1887 Barnett, T. P., Adam, J. C., & Lettenmaier, D. P. (2005). Potential impacts of a warming climate on water availability in snow - dominated regions. Nature , 438, 303 309. https://doi.org/10.1038/nature04141 Bormann, K . J., Brown, R. D., Derksen, C., & Thomas H. Painter (2018). Estimating snow - cover trends from space. Nature Climate Change , 8, 924 - 928. https://doi.org/10.1038/s41558 - 018 - 0318 - 3 Bonsoms, J., Salvad or - Franch, F., & Oliva, M. (2021). Snowfall and snow cover evolution in the Eastern Pre - Pyrenees (NE Iberian Peninsula). Cuadernos de Investigación Geográfica, 47 (2), 291 - 307. http://doi.org/10.18172/cig.487 9 Brown, I. (2019). Snow cover duration and extent for Great Britain in a changing climate: Altitudinal variations and synoptic - scale influences. International Journal of Climatology, 39 (2), 4611 - 4626. https://doi.org/10.1002/joc.6090 Carvalho, A., Schmidt, L., Santos, F. D., & Delicado, A. (2014). Climate change research and policy in Portugal. WIREs. Climate Change, 5 (2),199 217. https://doi.org/10.1002/w cc.258 Carvalho, M. J. P. J. (2007). Os desportos de inverno e o reposicionamento da oferta na Região de Turismo da Serra da Estrela [ Winter Sports and the Repositioning of Tourism Offerings in the Serra da Estrela Tourism Region ]. [Master thesis in Stra tegical management and tourism development, IST - UTL]. https://core.ac.uk/reader/61796572 Cerasoli, F., D’Alessandro, P., & Biondi, M. (2022). Worldclim 2.1 versus Worldclim 1.4: Climatic niche and grid r esolution affect between - version mismatches in Habitat Suitability Models predictions across Europe. Ecology and Evolution, 12 (2), e8430. https://doi.org/10.1002/ece3.8430 Daveau, S., & Coelho, C. (1977) Répartition et rythme des précipitations au Portugal [ Distribution and Pattern of Precipitation in Portugal ]. CEG. Di Musciano, M., Carranza, M. L., Frate, L., Di Cecco, V., Di Martino, L., Frattaroli, A. R., & Stanisci, A. (2018). Distribution of Plant Sp ecies and Dispersal Traits along Environmental Gradients in Central Mediterranean Summits. Diversity, 10 (58). https://doi.org/10.3390/d10030058 Diffenbaugh, N., Scherer, M., & Ashfaq, M. (2013). Response of snow - dependent hydrologic extremes to continued g lobal warming. Nature Climate Change , 3, 379 384. https://doi.org/10.1038/nclimate1732 Döscher, R., Acosta, M., Alessandri, A., Anthoni, P., Arsouze, T., Bergman, T., … & Zhang, Q. (2022). The EC - Earth3 Earth system model for the Coupled Model Intercomparison Project 6. Geoscientific Model Development, 15 (7), 2973 3020. https://doi.org/10.5194/gmd - 15 - 2973 - 2022
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 22 Fick, S. E., & Hijmans, R. J. (2017). W orldClim 2: New 1 - km spatial resolution climate surfaces for global land areas. International Journal of Climatology, 37 (12), 4302 4315. https://doi.org/10.1002/joc.5086 Hughes, P. D., Fletcher, W. J., Bell, B. A., Braithwaite, R. J., Cornelissen, H. L., Fink, D. F., & Rhoujjati, A. (2020). Late Pleistocene glaciers to presentday snowpatches: a review and research recommendations for the Marrakech High Atlas. Mediterranean Geoscience Reviews , 2, 163 - 184. https://doi.org/10.1007/s42990 - 020 - 00027 - 4 Intergovernmental Panel on Climate Change. (2019) Summary for Policymakers. In H. - O. Pörtner, D.C. Roberts, V. Masson - Del motte, P. Zhai, M. Tignor, E. Poloczanska, … N.M. Weyer (Eds.), Special Report on the Ocean and Cryosphere in a Changing Climate. IPCC. https://www.ipcc.ch/srocc/chapter/summary - fo r - policymakers/ Lamprecht, A., Pauli, H., Fernández Calzado, M. R., & Winkler, M. (2021). Changes in plant diversity in a water - limited and isolated high - mountain range (Sierra Nevada, Spain). Alp Botany , 131, 27 - 39. https://doi.org/10.1007/s00035 - 021 - 00246 - x Lautensach, H. (1929). Eiszeitstudien in der Serra da Estrela (Portugal) [ Glacial Studies in the Serra da Estrela (Portugal) ]. Zeitschrift für Gletcherkunde, 17, 324 - 369. Liston, G. E., & Hiems tra, C. A. (2011). The Changing Cryosphere: Pan - Arctic Snow Trends (1979 2009). Journal of Climate, 24 (21), 5691 - 5712. https://doi.org/10.1175/JCLI - D - 11 - 00081.1 López - Bustins, J. - A., Martin - Vide, J., & Sanchez - Lorenzo A. (2008). Iberia winter rainfall trends based upon changes in teleconnection and circulation patterns. Global and Planetary Change, 63 (2 3), 171 - 176. https://doi.org/10. 1016/j.gloplacha.2007.09.002 López - Moreno, J. I., Soubeyroux, J. M., Gascoin, S., Alonso - González, E., Durán - Gómez, N., Lafaysse, M., & Morin, S. (2020). Long - term trends (1958 2017) in snow cover duration and depth in the Pyrenees. International Journal o f Climatology, 40 (14),1 - 15. https://doi.org/10.1002/joc.6571 López - Moreno, J. I. (2005). Recent variations of snowpack depth in the central Spanish Pyrenees. Arctic Antarctic Alpine Research 37 (2), 253 - 260. https://doi.org/10.1657/1523 - 0430(2005)037[0253:RVOSDI]2.0.CO;2 López - Moreno, J. L., Goyette, S., & Beniston, M. (2009). Impact of climate change on snowpack in the Pyrene es: Horizontal spatial variability and vertical gradients. Journal of Hydrology, 374 (3 - 4), 384 - 396. https://doi.org/10.1016/j.jhydrol.2009.06.049 Marchane, L., Jarlan, L., Hanich, A., Boudhar, S., Gascoin, A., Tavernier, N. … Berjamy, B. (2015). Assessment of daily MODIS snow cover products to monitor snow cover dynamics over the Moroccan Atlas mountain range. Remote Sensing of Environment , 160, 72 - 86. https://doi.org/10.1016/j.rse.2015.01.002 Meinshausen, M., Nicholls, Z. R. J., Lewis, J., Gidden, M. J., Vogel, E., Freund, M. … Wang, R. H. J. (2020). The shared socio - economic pathway (SSP) greenhouse gas concentrations and their extensions to 2500. Geoscientific Model Development, 13 (8), 3571 3605. https://doi.org/10.5194/gmd - 13 - 3571 - 2020 Mora, C. (2010). A synthetic map of the climat opes of the Serra da Estrela (Portugal). Journal of Maps, 6 (1), 591 - 608. https://doi.org/10.4113/jom.2010.1112 Mora, C. (2009). Padrões térmicos no alto vale do Zêzere. Condicionantes topográficas e me teorológicas [Thermal Patterns in the Upper Zêzere Valley: Topographic and Meteorological Constraints]. Finisterra Revista Portuguesa de Geografia , XLIV (87). https://doi.org/10.18055/Finis1377 Mora, C. ( 2006). Climas da Serra da Estrela: características regionais e particularidades locais dos planaltos e do Alto Vale do Zêzere [Tese de Doutoramento, Universidade de Lisboa]. Repositório da Universidade de Lisboa. https://repositorio.ulisboa.pt/handle/10400.5/98023 Mora, C., Vieira, G. T., & Alcoforado, M. J. (2001). Daily minimum air temperatures in the Serra da Estrela, Portugal. Finisterra Revista Portuguesa de Geografia, XXXVI (71). https://doi.org/10.18055/Finis1647 Mora, C., Jímenez. J. J., Pina, P., Catalão, J., & Vieira. G. (2017). Evaluation of single - band snow patch mapping using high resolution microwave remote sensing: an applic ation to the Maritime Antarctic. The Cryosphere , 11, 139 - 155. https://doi.org/10.5194/tc - 11 - 139 - 2017
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 23 Mora, C., & Vieira, G. (2020). The climate of Portugal. In G. Vieira, J. L. Zêzere & C. Mora (Eds.), Landforms and Landscapes of Portugal (pp. 33 - 46). Springer. Moreno - Gené, J., Sánchez - Pulido, L., Cristobal - Fransi, E., & Daries, N. (2018). The Economic Sustainability of Snow Tourism: the case of ski resorts in Austria, France, and Italy. Sustainability, 10 (9), 3012. https://doi.org/10.3390/su10093012 Mota, M., Nossa, P., & Oliveira Moreira, C. (2023). The Impact of Health and Wellness Tourism in the Regional Economy of Estrela UNESCO Global Geopark, P ortugal. Sustainability, 15 (20), 15151. https://doi.org/10.3390/su152015151 Muñoz - Sabater, J., Dutra, E., Agustí - Panareda, A., Albergel, C., Arduini, G., Balsamo, G. Thépaut, J. - N. (2021). ERA5 - Land: a s tate - of - the - art global reanalysis dataset for land applications. Earth System Science Data, 13 (9), 4349 - 4383. https://doi.org/10.5194/essd - 13 - 4349 - 2021 Oliveira, M. E., Brito, J. C., Dellinger, T., Almeida, N. F., Loureiro, A., Martins, H. R. … Teixeira, J. (2005). Lacerta monticola Lagartixa - da - montanha [ Lacerta monticola Mountain lizard ]. In In M. J. Cabral, J. Almeida, P. R. Almeida, T. Dellinger, N. Ferrand de Almeida, M. E. Oliveira, J. M. Pa lmeirim, A. L. Queiroz, L. Rogado, & L. Santos - Reis (Eds.), Livro Vermelho dos Vertebrados de Portugal [ Red Book of the Vertebrates of Portugal] (pp. 1 - 6). Instituto da Conservação da Natureza, Lisboa. O’Neill, B. C., Tebaldi, C., Vuuren, D. P., Eyring, V., Friedlingstein, P., Hurtt, G., & Sanderson, B. M. (2016). The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6. Geoscientific Model Development, 9 (9), 3461 - 3482. https://doi.org/10.5194/gmd - 9 - 3461 - 2016 Pepin, N., Bradley, R. S., Diaz, H. F., Baraer, M., Caceres, E. B., Forsythe, N. … Yang, D. Q. (2015). Elevation - dependent warming in mountain regions of the world. Nature Climate Ch ange , 5, 424 - 430. https://doi.org/10.1038/nclimate2563 Pérez - Palazón, M. J., Pimentel, R., & Polo, M. J. (2018). Climate Trends Impact on the Snowfall Regime in Mediterranean Mountain Areas: Future Scen ario Assessment in Sierra Nevada (Spain). Water , 10, 720. https://doi.org/10.3390/w10060720 Polderman, A., Haller, A., Viesi, D., Tabin, X., Sala, S., Giorgi, A. ... Daragon, Q. (2020). How Can Ski Resorts Get Smart? Transdisciplinary Approaches to Sustainable Winter Tourism in the European Alps. Sustainability , 12, 5593. https://doi.org/10.3390/su12145593 Pomeroy, J. W., & Eric Brun. (2001). Physical prop erties of snow. In H. G. Jones, J. W. Pomeroy, D. A. Walker & R. W. Hoham (Eds.), Snow ecology: An interdisciplinary examination of snow - covered ecosystems (pp. 45 118). Cambridge University Press. Pons, M. R., San - Martín, D., Herrera, S., & Gutiérrez, J. M. (2010). Snow trends in Northern Spain: analysis and simulation with statistical downscaling methods. International Journal of Climatology, 30 (12), 1795 - 1806. https://doi.org/10.1002/joc.2016 Ramos, A. M., Trigo, R. M., & Santo, F. E. (2011). Evolution of extreme temperatures over Portugal: recent changes and future scenarios. Climate Research , 48, 177 - 192. http://dx.doi.org/10.3354/cr00934 Rees, G. (2006) R emote sensing of snow and ice . Taylor & Francis. Revuelto, J., López - Moreno, J. I., Morán, E., Fassnacht, S., &Vicente, S. M. (2012). Variabilidad interanual del manto de nieve en el Pirineo: tendencias observadas y su relación con índices de teleconexión durante el periodo 1985 - 2011. In C. Rodríguez, A. Ceballos, N. González, E. Morán, S. Pacheco & A. Hernández (Eds.), Asociación Española de Climatología (pp. 613 - 621). Madrid. http://hdl.handle.net/20.500.11765/8325 Silva, C., Abrantes, J. L., & Kastenholz, E. (2018). A imagem da Serra da Estrela, na perspetiva dos turistas [ The image of Serra da Estrela from the tourists’ perspective. In A. Correia & A. P. Barbas Homem ]. In A. Correia & A. P. Barbas Homem (Eds.), Turismo no Centro de Portugal: Potencialidades e Tendências [ Tourism in Central Portugal: Potential and Trends ] (pp. 279 - 298). Almedina. Soares, P. M. M., Cardoso, R. M., Lima, D. C. A, & Miranda, P. (2017) Future prec ipitation in Portugal: high - resolution projections using WRF model and EURO - CORDEX multi - model ensembles. Climate Dynamics , 49, 2503. https://doi.org/10.1007/s00382 - 016 - 3455 - 2 Spandre, P., François , H., Verfaillie, D., Pons, M., Vernay, M., Lafaysse, M., & Morin, S. (2019). Winter tourism under climate change in the Pyrenees and the French Alps: relevance of snowmaking as a technical adaptation. The Cryosphere, 13 (4), 1325 - 1347. https://doi.org/10.5194/tc - 13 - 1325 - 2019
Mora, C., Vieira, G ., Finisterra, LX (129), 2025, 37506 24 Vicente - Serrano, S. M., Rodríguez - Camino, E., Domínguez - Castro, F., Kenawy, A., & Azorín - Molina, C. (2017). An updated review on recent trends in observational surface atmospheric variables and their extremes over Spain. Cuadernos de Investigación Geográfica, 43 (1), 209 - 232. https://doi.org/10.18172/cig.3134 Vieira, G., & Woronko, B. (2021). The glaciers of Serra da Estrela. In M. Oliva, D. Palacios & J. M. Fernández - Fernández (Eds.), Iberia, Land of Glaciers . Elsevier. https://doi.org/10.1016/B978 - 0 - 12 - 821941 - 6.00020 - 7 Vieira, G., Palacios, D., Andrés. N., Mora, C., Vázquez - Selem, L., Woronko, B., & Goyanes, G. (2021). Penultimate Glacial Cycle glacier extent in the Iberian Peninsula: New evidence from the Serra da Estrela (Central System, Portugal). Geomorphology , 388, 10778. https://doi.org/10.1016/j.geomorph.2021.107781 Vieira, G., Castro, E., Gomes, H., Loureiro, F., Fernandes, M., Patrocínio, F. … Forte, J. P. (2020). The Estrela Geopark: from planation surfaces to glacial erosion. In G. Vieira, J. L . Zêzere & C. Mora (Eds.), Landforms and Landscapes of Portugal (pp. 341 - 357). Springer. Vieira, G. (2008). Combined numerical and geomorphological reconstruction of the Serra da Estrela plateau icefield, Portugal. Geomorphology, 97 (1 - 2), 190 - 207. http://doi.org/10.1016/j.geomorph.2007.02.042 Vieira, G. T., & Mora, C. (1998). General characteristics of the climate of the Serra da Estrela, in Vieira, G.T. (ed), Glacial and Periglacial Geomor phology of the Serra da Estrela. Guidebook for the field - trip, IGU Commission on Climate Change and Periglacial Environme nts, 26 - 28 August 1998, CEG, University of Lisbon. 26 - 36.