Seasonal temperature trends on the Spanish mainland: A secular study (1916–2015)

RESEARCH ARTICLE Seasonal empe a u e ends on he Spanish mainland: A secula s udy (1916–2015) Dhais Peña-Angulo 1 | José Ca los Gonzalez-Hidalgo 2,3 | Lei e Sandonís 2,3 | San iago Begue ía 4 | Miquel Tomas-Bu gue a 4 | Joan Albe López-Bus ins 5 | Ma c Lemus-Cano as 5 | Ja ie Ma in-Vide 5 1 Ins i u o Pi enaico de Ecología, Consejo Supe io de In es igaciones Cien í icas (IPE-CSIC), Za agoza, Spain 2 Depa amen o de Geog a ía, Uni e sidad Za agoza, Za agoza, Spain 3 IUCA, Uni e sidad Za agoza, Za agoza, Spain 4 Es ación Expe imen al Aula Dei, Consejo Supe io de In es igaciones Cien í icas (EEAAD-CSIC), Za agoza, Spain 5 G upo de Clima ología, Uni e sidad de Ba celona, Ba celona, Spain Co espondence José Ca los González-Hidalgo, Depa amen o de Geog a ía, Uni e si y Za agoza, Campus San F ancisco, Za agoza 50009, Spain. E-mail: [email p o ec ed] Funding in o ma ion Spanish Minis y o Economy and Compe i i eness, MEC; Regional Go e nmen o A agon, Geo- en i onmen al and Global Clima e Change Resea ch G oup, G an /Awa d Numbe : E02-17R; Spanish Go e nmen , G an /Awa d Numbe s: CGL2017-83866-C3-3-R, CGL2017-83866-C3-2-R, CGL2017-83866-C3-1-R Abs ac T ends in seasonal mean alues o maximum and minimum empe a u e a e analysed in he Spanish mainland om he new MOTEDAS_cen u y da abase. This new da a se has been de eloped combining he digi alized a chi es om he Spanish Me eo ological Agency (AEMET) wi h in o ma ion e ie ed om Annual Books published by he o me Me eo ological Agency da ing back o 1916, and co e s he pe iod 1916–2015. In all ou seasons, mean seasonal em- pe a u e o maximum (Tmax) and minimum (Tmin) inc eased. The aising occu ed in wo main pulses sepa a ed by a i s pause a ound he middle o he 20 h cen u y, bu di e ed among seasons and also be ween maximum and mini- mum empe a u e. Analysis o he pe cen age o land a ec ed by signi ican ends in maximum empe a u e e eals wo inc easing phases in sp ing and summe o Tmax, and in sp ing, summe , and au umn o Tmin. Howe e , win e Tmax only ose du ing he ecen decades, and au umn Tmax in he i s decades. Nega i e signi ican ends we e ound in ex ended a eas in sp ing Tmax, and in sp ing, au umn, and summe Tmin, con i ming he i s pause a ound he 1940's–1960's. T ends o seasonal mean alues o Tmax and Tmin a e no signi ican o a leas he las 25–35 yea s o he s udy pe iod, depending on he season. The a eas unde signi ican posi i e end a e usually mo e ex ended o Tmin han Tmax a any season and pe iod. A eas wi h signi ican end expand and con ac in ime acco ding o wo spa ial g adien s: sou h-eas o no h-wes (eas -wes ) o Tmax, and wes o eas o Tmin. We hypo hesize a ela ionship be ween a mosphe ic p e alen ad ec ion and elie as igge ing ac o s o unde s and spa ial and empo al di e ences in seasonal empe a u es a egional scale du ing he 20 h cen u y in he Ibe ian Peninsula. KEYWORDS MOTEDAS_cen u y da abase, seasonal ends, Spain, empe a u e maximum, empe a u e minimum Recei ed: 15 Sep embe 2020 Re ised: 7 Janua y 2021 Accep ed: 8 Janua y 2021 DOI: 10.1002/joc.7006 This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion-NonComme cial-NoDe i s License, which pe mi s use and dis ibu ion in any medium, p o ided he o iginal wo k is p ope ly ci ed, he use is non-comme cial and no modi ica ions o adap a ions a e made. © 2021 The Au ho s. In e na ional Jou nal o Clima ology published by John Wiley & Sons L d on behal o he Royal Me eo ological Socie y. In J Clima ol. 2021;1–14. wileyonlinelib a y.com/jou nal/joc 1 1|INTRODUCTION Mos empe a u e end analyses ace he p oblem o gen- e alized poo da a a ailabili y, which becomes pa icu- la ly no iceable o he i s hal o he 20 h cen u y. To be e unde s and cu en wa ming, howe e , a secula con ex o longe mus be adop ed be e ame he cu - en ends in pe spec i e. Se e al s udies ha e shown ha empe a u e e olu ion du ing he 20 h cen u y was no mono onic bu cha ac e ized by pe iods o ising em- pe a u e and pe iods in which empe a u e emained s a- ble o e en dec eased, usually e e ed o as ‘pauses’o ‘hia us’(Eas e ling and Wehne , 2009; Fy e e al., 2013; Kosaka and Xie, 2013; Dong and McPhaden, 2017). Tem- pe a u e ise pe iods a e usually iden i ied be ween ci ca 1910–1940 and 1976–1997, while pauses ha e been epo ed be ween ci ca 1941–1975 and 1997–2013 (Folland e al., 2018). In he Wes e n Medi e anean basin, and pa icu- la ly in he Spanish mainland, he e a e only a ew secu- la analyses, mos ly ca ied ou wi h a e y low s a ion densi y. B une e al. (2007) analysed he seasonal em- pe a u e e olu ion in 22 s a ions o he pe iod 1850–2005. They ound ha he highes con ibu ion o he annual mean inc easing end was in win e and au umn, wi h he a es o maximum empe a u e (diu - nal ime, Tmax) being s onge han hose o minimum empe a u e (nigh - ime, Tmin). They analysed he i s ising pe iod o he 20 h cen u y (1901–1949), also he i s pause (1950–1972), and he second ising pe iod (1973–2005). Howe e , hey did no p o ide any in o - ma ion ega ding he second pause. They ha e shown, howe e , ha empe a u e ends we e no signi ican in se e al pe iods: in win e o Tmax and Tmin, and in summe and au umn o Tmin o e 1901–1949. Fu he - mo e, nega i e and signi ican seasonal ends we e ound om 1950 o 1972, and du ing he inal pe iod analysed (1951–2005), au umn Tmax and win e Tmin ends we e no signi ican . Di e en esul s we e ound by S aud e al. (2007) who de ec ed and co ec ed he u ban e ec in 43 s a ions, and sugges ed ha Tmin end a es we e highe han hose o Tmax. Un o u- na ely, hese au ho s nei he s udy sub-pe iods, no combined he s a ions in o a egional se ies. O he a i- cles analysing he secula ends a sub- egional scale (see e iew in Gonzalez-Hidalgo e al., 2015) ag ee wi h B une e al. (2007). This si ua ion con as s wi h he la ge body o esea ch co e ing pos -1950 yea s, ei he using wea he s a ions (del Río e al., 2011, 2012), g ids (Gonzalez- Hidalgo e al., 2015; He e a e al., 2015; Gonzalez- Hidalgo e al., 2018) o a e age egional and sub- egional se ies (Guija o, 2013; Gonzalez-Hidalgo e al., 2016). Gene ally speaking, hese pape s analyse he pe iod since he end o he i s pause, hus co e ing only he second pe iod o empe a u e ise, and in some cases de ec he second pause. S ill, hey do no p o ide he secula o he e olu ion o empe a u e in he egion. Gonzalez-Hidalgo e al. (2015) p esen ed a mo e ho ough e iew o hese s udies. This pape p esen s an analysis o seasonal mean maximum and minimum empe a u e ends o e he Spanish mainland, co e ing he pe iod 1916–2015, and based on high spa ial esolu ion in o ma ion p o ided by he new MOTEDAS_cen u y da abase. The main objec- i e is o explo e he spa ial a ia ions o seasonal ends o Tmax and Tmin along he s udy pe iod, by analysing hei beha iou a di e en empo al windows. 2|DATA AND METHODS The ecen ly p esen ed MOn hly TEmpe a u e DA ase o Spain (ac onym MOTEDAS_cen u y) combines in o - ma ion digi alized om he Annual Books (Lib os Resúmenes Anuales, LRA) edi ed by he o me Me eo o- logical Agency o Spain, and digi al da a om he Banco Nacional de Da os del Clima (BNDC) o he Na ional Me eo ological Agency o Spain (AEMET). The da a includes mon hly maximum (Tmax, diu nal- ime) and minimum (Tmin, nigh- ime) empe a u e, and co e s he 1916–2015 pe iod. A gene al desc ip ion o da a escue, s a ion ma ching, quali y con ol, and g id calcula ion a e desc ibed in Gonzalez-Hidalgo e al. (2020), al hough he mos c i ical in o ma ion will be p o ided he e. The esul ing g id has been he e o e ob ained using he highes spa ial in o ma ion cu en ly a ailable, in pa ic- ula o he pe iod 1916–1949. No wi hs anding, he numbe o s a ions a ies la gely du ing he pe iod o s udy, wi h a minimum o 228 in 1939 (end o he Span- ish Ci il Wa ), and a maximum o 2030 in 1994. MOTEDAS_cen u y o e s a spa ial esolu ion o 10 ×10 km. Da a escued om he LRA illed gaps and ex ended back he BNDC se ies, because abou 30% o da a escued om he LRA we e no included in he BNDC. Thus, LRA da a inc eased signi ican ly he in o ma ion a ail- able o he p e-1950 pe iod. Ne e heless, da a se ies om bo h, BNDC and LRA, p esen some d awbacks because he e a e a la ge numbe o s a ions wi h e y sho empo al eco ds. Fo ins ance, 3,969 (75.5%) se ies om a o al o 5,259 ha e less han 30 yea s (see ables 1 and 2 in Gonzalez-Hidalgo e al., 2020). This cha ac e - is ic is accen ua ed in he 1916–1950 pe iod, when 13.2% o all s a ions (156) eco ded da a o only 1 yea ; 22% o only 2 yea s (259 s a ions); and 63% o he s a ion (748) 2D. PEÑA-ANGULO ET AL. o less han 10 yea s. Pa icula ly, he LRA se ies end o be e y sho : 351 s a ions om a o al o 1107 (31.7%) eco ded less han 3 yea , and mo e han 70% (807) eco ded less han 10 yea s. The de elopmen o he g id was done conside ing ha : (i) he numbe o s a ions changed yea by yea ; and also (ii) ha he e we e many changes in he loca- ion o some s a ions. In o de o maximize he amoun o in o ma ion we decided o compu e mon hly ields independen ly using he o al amoun o da a a ailable each mon h, ins ead o econs uc ing a selec ed num- be o ime se ies. This app oach is simila o he one desc ibed and used in he eanalysis p ojec (Sli inski e al., 2020), and we applied i a e alida ing i s adequacy by compa ing al e na i e p ocedu es using illed ( econs uc ed) and non- illed se ies. The in e pola ion p ocedu e was a modi ica ion o he model sugges ed by B une i e al. (2006) ha combines dis ance weigh ing and azimu h, a e con e ing he o iginal da a in o anomalies o a oid ele a ion e ec s. To do so, he di e - ence be ween he obse ed alues and he mean clima- ology was compu ed, using he 1951–2010 clima ology compu ed by Peña-Angulo e al. (2016). P elimina y quali y con ol o aw da a was done o a oid suspicious da a. De ailed in o ma ion on he ad an ages and d aw- backs o he p ocedu e can be ound in Gonzalez- Hidalgo e al. (2020). Seasonal ime se ies we e ob ained o each g id cell bo h o Tmax and Tmin using he classic agg ega ion (Decembe –Janua y–Feb ua y o win e , Ma ch–Ap il– May o sp ing, June–July–Augus o summe , and inally Sep embe –Oc obe –No embe o au umn). The sign and signi icance o he empo al end we e calcu- la ed by he Mann–Kendall es (Mann, 1945) using a sig- ni icance le el o α=.05, a e disca ding any au oco ela ion (see Gonzalez-Hidalgo e al., 2015), and he a e o he end was es ima ed ollowing Sen (1968). Ra es a e shown as C pe decade (C/dec). I is gene ally accep ed ha signi icance and a e in empe a u e ends depend on he chosen empo al win- dow and leng h. Howe e , he e is no ag eemen on he minimum leng h equi ed o obus ly disc imina e a ending signal om na u al a iabili y o noise. In ou case, we se he minimum empo al span a 20 yea s, la gely in excess o he ecommenda ions o s udies such as hose o Knigh e al. (2009), who sugges ed a mini- mum o 15 yea s, o San e e al. (2011), who sugges ed a minimum o 17 yea s. Mo e ecen ly Huang (2013) no iced ha ‘ he mos de ec able secula end signals appea in he CO 2 band and he ime i akes o see hese adiance changes is much less han 12 yea s’, wi h simi- la opinions exp essed by Loehle (2009), Liebmann e al. (2010) and McKi ick (2014). Following he a o emen ioned esea ch, empo al a ia ions in ends we e analysed by using inc easing empo al windows spanning a minimum o 20 yea s (1916–1935), un il co e ing he en i e pe iod 1916–2015. Addi ionally, dec easing empo al windows we e also used, s a ing om he 1916–2015 window o he mos ecen 1996–2015 one (again, using a minimum leng h o 20 yea s). Inc easing and dec easing empo al win- dow end analyses enabled us o ind he e ec s o selec ed pe iods and empo al leng hs on he esul ing ends. In he i s case, he e ec o inc easing he size o he pe iod o analysis is simila o he p ocess o upda ing a da abase yea by yea . The esul s, he e o e, ake in o conside a ion he cumula i e e ec on he endonbo hmagni udeandsigni icance.When he empo al windows dec ease, he e ec is equi alen o dec easing he age o he eco ds, and i in o ms on he mos ecen pe iod, mos ly he second pause, and iden- i ies i s beginning, empo al leng h, and spa ial a ia- ion, among o he ea u es. Finally, 30-yea s mo ing windows we e also used o a oid he e ec o he leng h o he windows on he con idence in e als o he ends (size e ec ). 3|RESULTS 3.1 |Seasonal empo al e olu ion o Tmax and Tmin on he Spanish mainland 1916–2015 The ime se ies o mean seasonal Tmax and Tmin (as anomalies o e he pe iod 1916–2015) o he whole Spanish mainland a e shown in Figu e 1. The main esul is ha Tmin shows lowe empo al a iabili y han Tmax, excep du ing win e . The se ies shows a gene alized inc ease in empe a- u e du ing he s udy pe iod, bo h in Tmax and Tmin and in he ou seasons. I was easy o iden i y he di e - en pe iods de ec ed a a global scale. Also, we no ed he ex emely low empe a u e eco ded in he summe o 1977. The highes end a e o he 1916–2015 pe iod co e- sponds o Tmax in sp ing (0.16C/dec ±0.08) and sum- me (0.13C/dec ±0.06). Fo Tmin he highes a es occu ed in summe (0.13C/dec ±0.04) and au umn (0.11C/ dec ±0.05). Table 1 shows seasonal a es o di - e en pe iods o egional se ies. Global Tmax a es a e no signi ican o win e since he 1965–2015 window, and since 1945–2015 o Tmin. Fo sp ing he no signi ican pe iods a e 1985–2015 and 1992–2015, espec i ely. Fo summe hey a e 1981–2015 and 1987–2015. Fo au umn he e is no clea beha iou SPANISH SECULAR SEASONAL TEMPERATURE 3 in Tmin because a es ha e changed o signi ican o e he las ew decades, while no signi ican end we e de ec ed in Tmax. The highes seasonal end a es occu ed a ound he 1970s, and in ecen decades he a es o bo h a iables dec eased o all seasons. Tmax end a es we e highe han Tmin in sp ing and win e when hey we e signi i- can . In summe and au umn, Tmin a es we e usually highe han win e Tmax, and highe in summe . In gene al, he signi icance o a es emained o longe pe iods o Tmin han Tmax. 3.2 |Spa ial a ia ion in ends using 30-yea s mo ing windows The e olu ion o he su ace pe cen age acco ding o he end sign and signi icance a 30-yea s mo ing windows FIGURE 1 Regional seasonal mean se ies (Spanish mainland) o Tmax and Tmin. Values as anomalies o e 1916–2015 [Colou igu e can be iewed a wileyonlinelib a y.com] 4D. PEÑA-ANGULO ET AL. is p esen ed in Figu e 2. Two main ising pe iods a e iden i ied, and also in e media e and a inal pause. The pe iods di e be ween Tmax and Tmin o season and annual alues. The i s ise in Tmax a ec ed less han 50% o land in sp ing, summe , and au umn, and did no occu in win e . On he o he hand, he second ise was mo e ex ended in win e and s a ed ea lie han sp ing and summe , while being p ac ically absen in au umn. Also he i s pause is clea ly de ec ed in Tmax, when signi i- can nega i e ends a ec ed ex ended a eas (>50%) in sp ing, and o a lesse ex en in summe (c. 25%) and au umn (c. 10%). Meanwhile, he second pause is de ec ed in he las decades in he ou seasons. No clea e idences o win e ise a e de ec ed in Tmin du ing he s udy pe iod, while wo ising pe iods a e iden i ied in sp ing, summe , and au umn Tmin. Du ing he i s ising pe iod he a ea unde signi ican end was a ound 25% (highe alues han Tmax), and du ing he second posi i e phase i ex ended o almos he en i e a ea in sp ing and summe , bu also in au umn du ing some decades a ound he 1970's. Ex ended a eas wi h sig- ni ican nega i e end ha e been iden i ied (>75%) du - ing he i s pause in sp ing and au umn and o a lesse ex en in summe , and no signi ican nega i e end a eas ha e been de ec ed in win e . No signi ican ends ha e also been de ec ed du ing he las decades in Tmin, again showing he ecen pause o e ex ended a eas. These esul s sugges ha he end obse ed in he mean annual Tmax depend mos ly on sp ing, summe , and au umn du ing he i s ising pe iod; and win e , sp ing and summe du ing he second one. Fo Tmin, on he o he hand, sp ing, summe , and au umn de ined he wo ising pe iods. The spa ial e olu ion o ends acco ding o 30-yea mo ing windows is p esen ed in Figu e 3. In Tmax he i s posi i e pe iod un il he mid-1950's mos ly a ec s he sou h-eas e n a eas in sp ing, summe and au umn, while he second pe iod begun a ound he 1960's and las ed un il 2010 a mos , being mo e ex ended. The Figu e also shows ha , o mo e han 30 yea s, no signi i- can end is de ec ed o Tmax. Wi h espec o he i s pause, he nega i e signi ican end is only de ec ed in sp ing o e ex ended a eas. The spa ial e olu ion o Tmin end sign and signi i- cance also shows wo di e en posi i e phases, las ing he second in sp ing and summe un il he inal window (1986–2015) o e ex ended a eas and pa icula ly o he eas . No signi ican ends a e ound in any pa o he s udy a ea in win e and au umn in he pe iod TABLE 1 Selec ed seasonal Tmax and Tmin end a es in di e en empo al windows (10-yea in e als) Sig. Win e (±) Sig. Sp ing (±) Sig. Summe (±) Sig. Au umn (±) Maximum 1916_2015_100 *** 0.11 0.06 *** 0.16 0.08 *** 0.13 0.06 * 0.09 0.07 1926_2015_90 ** 0.12 0.07 * 0.13 0.10 *** 0.13 0.07 0.05 0.09 1936_2015_80 ** 0.13 0.09 * 0.15 0.12 *** 0.16 0.09 0.07 0.11 1946_2015_70 ** 0.15 0.11 * 0.19 0.15 *** 0.18 0.11 0.03 0.14 1956_2015_60 ** 0.17 0.13 ** 0.31 0.18 *** 0.27 0.14 0.13 0.17 1966_2015_50 0.13 0.16 *** 0.52 0.21 *** 0.33 0.17 0.07 0.25 1976_2015_40 0.04 0.22 *** 0.57 0.28 ** 0.41 0.27 0.03 0.31 1986_2015_30 −0.16 0.42 0.36 0.44 0.27 0.48 0.18 0.43 1996_2015_20 −0.51 0.57 0.36 0.89 0.53 0.78 0.69 0.81 Sig Win e (±) Sig Sp ing (±) Sig Summe (±) Sig Au umn (±) Minimum 1916_2015_100 * 0.08 0.07 ** 0.09 0.05 *** 0.13 0.04 *** 0.11 0.05 1926_2015_90 * 0.09 0.09 * 0.08 0.06 *** 0.13 0.05 ** 0.09 0.06 1936_2015_80 0.09 0.11 * 0.08 0.07 *** 0.15 0.07 * 0.09 0.08 1946_2015_70 0.09 0.12 ** 0.12 0.09 *** 0.18 0.08 * 0.10 0.10 1956_2015_60 0.09 0.16 *** 0.21 0.13 *** 0.27 0.09 ** 0.19 0.13 1966_2015_50 0.08 0.23 *** 0.45 0.13 *** 0.38 0.12 *** 0.31 0.17 1976_2015_40 −0.01 0.33 *** 0.48 0.16 *** 0.44 0.18 * 0.30 0.25 1986_2015_30 −0.06 0.46 ** 0.33 0.26 * 0.30 0.25 0.13 0.42 1996_2015_20 −0.59 0.66 0.06 0.53 0.30 0.52 0.65 0.77 No e: Signi icance exp essed by (***) p<.001, (**) p<.01, and (*) p<.05. Ra e alues in C/dec. SPANISH SECULAR SEASONAL TEMPERATURE 5 1980–2015. Signi ican ends ended o appea in he cen al-wes e n a eas du ing he i s posi i e pe iod, while du ing he second one he a ea a ec ed was mo e ex ended and shi ed o he eas . Ex ended a eas o nega- i e and signi ican ends a e ound a ound he pe iod 1945–1985, when a ound 100% o he s udy a ea was a ec ed by signi ican cooling. This pe iod was mo e ex ended in sp ing han in summe and au umn, and mos ly a ec ed cen al-wes e n a eas. The pe cen age o land a ec ed by signi ican ends du ing he wo posi i es pe iods and he i s pause is highe in Tmin han Tmax in he ou seasons. The abo e esul s seem o be an imp o ed desc ip o o wa ming han simple annual mean a es om a egional se ies. 3.3 |Spa ial a ia ions in ends using inc easing empo al windows Figu e 4 p esen s he a ia ion o he pe cen age o land a ec ed by posi i e and nega i e, signi ican and non- signi ican ends unde inc easing empo al windows. The i s empo al window (20 yea s, on he le side) co - esponds o 1916–1935, and he las one (on he igh side) includes he en i e 1916–2015 pe iod, hus he e exis s a cumula i e e ec along he cha ha mus be conside ed. The o al a ea a ec ed acco ding o he sign and signi icance o he end di e s by seasons h ough- ou he empo al windows. Globally, win e Tmax ends we e posi i e and signi ican and a ec ed mo e han 50% o he a ea a e he 1916–1982 window, inc easing p og essi ely o co e up o 94% o he e i o y in 1916–2015. Sp ing and summe Tmax also showed a gene alized posi i e end in he en i e pe iod, in acco dance wi h p e ious s udies (see B une e al., 2007). In sp ing, he e olu ion o he pe cen age o land displays wo pulses. The i s one s a ed in he empo al windows om 1916–1944 o 1916–1974, whe e posi i e and signi ican ends a ec ed mo e han 50% o he a ea. Subsequen ly, gi en he e ec o nega i e and signi ican ends de ec ed in he i s pause, no signi ican ends we e de ec ed un il 1916–1987, when he a ea a ec ed by posi- i e end in Tmax s a s again o inc ease o co e 93.4% o he a ea a he end o he s udy pe iod. The spa ial e olu ion o land a ec ed by posi i e and signi ican ends in summe was simila o sp ing excep ha , du ing he i s pulse, he a ec ed a ea was smalle . In he second aising pulse, s a ing in 1916–1994 up o he p esen , he a ea a ec ed by posi i e and signi ican ends co e ed he s udy a ea almos comple ely. The posi i e and signi ican ends in summe a ec ed mo e han 20% o he a ea in he in e media e pe iod, whe eas i was close o 0% in sp ing. E olu ion o he a ea a ec ed by posi i e and signi i- can ends in au umn di e ed: wo pulses we e de ec ed in which he a ea unde posi i e and signi ican ends inc eased (simila o sp ing and summe ), bu a no ime co e ed mo e han 50% o he a ea (only one pulse o inc ease). The esul s sugges ha he use o annual mean alues o Tmax may mask impo an di e ences among FIGURE 2 The su ace pe cen age unde posi i e/nega i e end and signi icance (p<.05); 30-yea s mo ing windows. Tmax (le ), Tmin ( igh ) [Colou igu e can be iewed a wileyonlinelib a y.com] 6D. PEÑA-ANGULO ET AL. seasons and along ime. Tempe a u e ise un il he 1970's, o ins ance, seems o be d i en mos ly by he e o- lu ion o sp ing and summe empe a u es and, o a lesse ex en , au umn. On he o he hand, empe a u e e olu ion du ing he las ew decades seems o be ela ed wi h all seasons, albei a lesse con ibu ion o au umn. The empo al e olu ion o Tmin was qui e simila in sp ing, summe and au umn. All o hese seasons show Tmin inc easing in wo pulses h oughou he 20 h cen- u y, wi h a high pe cen age o land a ec ed by posi i e and signi ican ends (>50% o o al land in he i s case and >75% in he second one). Also, he i s pause is clea ly ep esen ed wi h nega i e signi ican ends de ec ed in sp ing and o a lesse ex en in summe . Win- e , howe e , showed a di e en beha iou again, and he pe cen age o land a ec ed by posi i e and signi ican ends was less han 10% un il 1990's, and he inal maxi- mum alue eached 50%. These esul s ag ee wi h hose p e iously p esen ed and show ha he i s ise in Tmin was mos ly d i en by sp ing, summe , and au umn and he second one by win e , sp ing, summe , and au umn. Figu e 5 shows a sequence o maps wi h ime s eps o 5 yea s, and he pe cen age o land a ec ed by posi i e- signi ican ends. Gene ally, he pe cen age o land a ec ed by signi ican posi i e ends is highe o Tmin han Tmax in bo h pe iods o empe a u e ise, and pa - icula ly in summe and au umn. The igu es show he impo ance o conduc ing de ailed spa ial analyses, because egional mean a es as hose p esen ed in Table 1 hide he ich spa ial a iabili y o empe a u e ends in he egion. In Spanish mainland signi ican pos- i i e Tmin ends ha e been mo e ex ended and p o- longed han Tmax in bo h ising pe iods. Two g adien s o spa ial a ia ion domina e he de el- opmen o ends along he empo al windows. The a eas a ec ed by posi i e-signi ican ends in Tmax expand FIGURE 3 Spa ial a ia ion in seasonal mean ends in Tmax and Tmin acco ding o 30-yea s mo ing windows (5-yea in e als). The igu e includes he pe cen age unde posi i e/nega i e p edominan signi ican end (Mann–Kendall es p<.05) [Colou igu e can be iewed a wileyonlinelib a y.com] SPANISH SECULAR SEASONAL TEMPERATURE 7 FIGURE 4 The su ace pe cen age unde posi i e and nega i e ends and hei signi icance (p<.05); inc easing empo al windows. Tmax (le ), Tmin ( igh ) [Colou igu e can be iewed a wileyonlinelib a y.com] 1916 1940 1916 1945 1916 1955 1916 1960 1916 1950 1916 1965 1916 1970 1935 1916 1916 1995 1916 2000 1916 2005 1916 2010 1916 2015 1916 1975 1916 1985 1916 1980 1916 1990 8% 48% 70% 54%58% 74% 45%1% 28% 42% 55% 75% 93%8% <1%0% 10% 0% 15% 29% 26%46% 29% 32%2% 52% 56% 77%82%89%29% 30%23% 39% 1% <1% 48% 16%62% 20% 21%<1% 41% 31% 28%32% 57%14% 44%8% 74% 2% 0% 0% 0%0% 0% <1%<1%61% 98% 99% 93% 94%<1% 38%11% 50% 11% 29% 65% 40%91% 46% 51%3% 65%80%92%97%100%35% 40%30% 57% Annual Win e Sp ing Summe Au umn Tmax <1% 0% 1% 4%<1% 7% 8% <1% 7% 22%41% 53% 58%1% 6%8% 10% 4% 48% 55% 64%61%75%54%1% <1%7% 7% 34% 60% 81%3%5% 19%<1%16% 11%<1% 8% 32% 58% 48%63% 59% 54%31% 58% 80% 92% 98% 99%12%1% <1%5%<1%6% 15% 22% 24% 69% 62%71% 71% 70%4% 63% 67% 82% 82% 94%11% 17% 1%1% 62% 18% 30% 63% 63%64% 69% 66% <1% 53% 76% 86% 93% 100% 12%6% 7%6%5%10% 42%<1% Annual Win e Sp ing Summe Au umn Tmin –0.05 +0.05– +–0.01 +0.01 FIGURE 5 Spa ial a ia ion in seasonal mean ends in Tmax and Tmin acco ding o inc easing empo al windows (5-yea in e als). The igu e includes he su ace pe cen age unde posi i e/nega i e signi ican p edominan end (Mann–Kendall es p<.05) [Colou igu e can be iewed a wileyonlinelib a y.com] 8D. PEÑA-ANGULO ET AL. and con ac om he eas (Medi e anean coas land) o he wes (inland), while he e e se is ue o Tmin om wes A lan ic-coas land o he eas -inland. The wo opposing g adien s can be de ec ed du ing he ising pe iods bo h in Tmax and Tmin in sp ing, summe , and au umn, while only he second one p e ails in win e . The i s pause is appa en in a wide a ea in sp ing Tmax, when posi i e signi ican ends we e p ac ically absen , whe eas a small signi ican posi i e end a ea emained in he eas e n coas land in summe and au umn. Fo Tmin, sp ing and summe showed a signi i- can cooling pe iod mos ly loca ed o he no h-eas , longe -las ing in sp ing han in summe . 3.4 |Spa ial a ia ion in ends using dec easing empo al windows Figu e 6 shows he empo al e olu ion o he pe cen age o land wi h signi ican posi i e and nega i e ends unde dec easing empo al windows. F om le o igh , he plo s con i m a global ise in empe a u e du ing he 20 h cen u y in Tmax a he annual and seasonal scales. The p esence o he second pause is, howe e , e y ma - ked, since empe a u e ceased o ise signi ican ly in all ou seasons. Taking he 20% o o al land a ec ed by a signi ican end as a h eshold, he s a ing da e o he second pause can be iden i ied p ecisely o Tmax: in win e in 1965–2015; in sp ing and summe in 1985–2015 and 1983–2015. In au umn, he a ea a ec ed by signi i- can ends is less han 20% since 1933, bu he e is a b ie eco e y pe iod a ound 1983. The same is ue o Tmin. Howe e , he win e pause s a ed e y ea ly (1946–2015), while da es o he onse o he hia us we e 1992–2015 o sp ing, and 1988–2015 o summe ; he au umn pause s a ed in 1978–2015. 4|DISCUSSION 4.1 |Gene al commen s I has been assumed ha global empe a u e ise has been con olled by he e olu ion o Tmax, mos ly in win- e (Jones e al., 1999). Howe e , he nega i e end o he diu nal empe a u e ange since mid-cen u y does no con i m his hypo hesis (Rohde e al., 2013; Sun e al., 2018a,b). On he o he hand, disc epancies a ise in egional s udies a he seasonal scales as hose p esen ed he e. The e is de ailed in o ma ion o mainland Spain on empe a u e e olu ion om 1950 onwa d, cha ac e ized by a global ise, mos ly in summe and sp ing (B une e al., 2007; del Rio e al., 2011, 2012; Rios e al., 2012; Guija o, 2013; Gonzalez-Hidalgo e al., 2015). Howe e , he e is no ag eemen on whe he he inc ease was s on- ge o Tmax o Tmin. On he o he hand, less in o ma- ion exis s o he i s decades o he 20 h cen u y. Ou esul s show ha , globally, seasonal mean alues o Tmax and Tmin on he Spanish mainland ose be ween 1916 and 2015, in a simila sequence o pe iods as in he global da a se and in p e ious esea ch in he s udy a ea (see B une e al., 2007). Also, he esul s con- i m ha di e ences exis among Tmax and Tmin a he FIGURE 6 The su ace pe cen age unde posi i e/nega i e end and signi icance (p<.05); dec easing empo al windows. Tmax (le ), Tmin ( igh ) [Colou igu e can be iewed a wileyonlinelib a y.com] SPANISH SECULAR SEASONAL TEMPERATURE 9