PROMJENE KOLIČINE PADALINA U HRVATSKOJ OD SREDINE 20. STOLJEĆA DO DANAS CHANGES IN PRECIPITATION LEVELS IN CROATIA FROM THE MID 20th CENTURY TO THE PRESENT

Rezultati proučavanja klimatskih promjena na globalnoj razini u nekim se slučajevima nastoje prikazati toliko drastičnim da su doveli do klimatske uzbune. Osobito se to odnosi na naglašenu opasnost od globalnog zatopljenja, o čijoj bi se pak opravdanosti mogli naći pro et contra argumenti. Ostalim klimatskim elementima ne pridaje se tolika pažnja, iako bi značajnije promjene tih elemenata mogle bitno utjecati na život na Zemlji. Jedan od tih elemenata su svakako padaline.


Introduction
In some cases, the presentation of results of investigating climatic changes at the global level has been so drastic that it has led to climatic alarm. This particularly pertains to the pronounced threats of global warming, and there are arguments for and against as to whether these threats are justified. Other climatic elements receive less attention, though significant changes of those elements could significantly affect life on Earth. One such element is certainly precipitation.
Third and Fourth National Report of the Republic of Croatia to the UNFCCC (2006) also address the changes in precipitation levels in Croatia, and these reports are based on analyses of four weather stations in Croatia.
Trends in the quantity of precipitation in this part of Europe primarily indicate a decrease in precipitation levels (Romero et al., 1998;Ventura et al., 2002;Dore, 2005;Cannarozzo, 2006), which is consequently leading to aridification of the Mediterranean (Gao, Giorgi, 2008).
The objective of this paper was to investigate the changes in precipitation levels in Croatia from the mid 20 th century to the present day, and to determine whether regional differences can be ascertained.

Data and methods
Given Croatia's dynamic relief, the network of meteorological stations, particularly primary weather stations, is insufficient to provide a detailed regional analysis. This should certainly be taken into account while interpreting the results of climatic elements' spatial distribution. For this reason, certain authors often analyse the data from only one station, treating it as representative for a wider area, which need not always be the case.
In this paper, data on annual precipitation levels from 22 weather stations ( Figure 1) were analysed. The source of the data is the Hydrological and Meterological Service of the Republic of Croatia. A 60-year period, from 1950-2010 was analysed. Stations were selected as representative for continental Croatia (Osijek, Slavonski Brod, Donji Miholjac, Bjelovar, Križevci, Koprivnica, Varaždin, Zagreb-Maksimir, Karlovac, Ogulin, Gospić), and for the coastal region (Rovinj, Pazin, Rijeka, Senj, Mali Lošinj, Zadar, Šibenik, Split-Marjan, Hvar, Lastovo, Dubrovnik). The share of interpolated values was minimal, and amounted to only 0.3% for all stations and all annual data. Values were interpolated for the following stations: Osijek (from October 1991 to February 1992), Rovinj (from October 1986 to May 1988) and Dubrovnik (April 1978). The annual precipitation levels for all stations is shown graphically, and based on these data, 5-year and 10-year running means and linear trends were calculated using the least squares method. The use of a six-degree polynomial trend points to the existence of alternating precipitation cycles in more detail.

Results
The large annual fluctuations seen in precipitation levels were expected. To smooth the curve, 5-and 10-year running means were used ( Figure 2). With these, certain pronounced periods with higher levels and periods with lower annual precipitation levels became evident. In the early 1990s, the majority of stations recorded substantially lower precipitation levels, which is the consequence of the gradual decrease that started ten years earlier and continued to the end of the century. At the stations Koprivnica, Varaždin, Rovinj and Dubrovnik, this decline began even earlier. In the same period, stagnation instead of a decrease was recorded at the stations Križevci, Zagreb-Maksimir and Gospić. In line with this, two to three successive short-term trends could be isolated at all stations. Usporedbom linearnih trendova (Slika 2.) može se zaključiti da, osim Dubrovnika, niti na jednoj postaji nije došlo do izrazite promjene količine padalina, iako je na većini postaja evidentirano njihovo smanjenje (Tablica 1.). U istočnoj Hrvatskoj taj trend stagnira ili bilježi tek neznatan porast. U ostalim dijelovima Hrvatske trend pokazuje stagnaciju ili pad. Smanjenje količine padalina u promatranom je razdoblju najizrazitije u Dubrovniku (45 mm svakih 10 godina) čime se ta postaja bitno izdvaja od svih ostalih u Hrvatskoj, što bi se moglo tumačiti i time što je geografski najbliža dijelu Mediterana u kojemu je smanjenje količine padalina dosta izraženo (Norrant, Douguédroit, 2006.). To se smanjenje prvenstveno odnosi na Grčku. Iako su objavljeni brojni radovi o padalinama na Mediteranu koji dokazuju trend smanjenja njihove količine (npr. Palutikof, 2003.) uzroci trenda se zapravo još uvijek ne mogu sa sigurnošću utvrditi iako je primijećena inverzna korelacija između smanjenja količine padalina i porast tlaka zraka (Norrant, Douguédroit, 2003.). Neki autori smatraju da porast koncentracije stakleničkih plinova remeti hidrološki ciklus, pa to utječe na režim padalina (Piervitali, Colacino, 2003. From the analysis of linear trends ( Figure  2), it can be concluded that none of the stations, with the exception of Dubrovnik, experienced a pronounced change in the quantity of precipitation, though a decline was recorded at the majority of stations (Table 1). In eastern Croatia, that trend has either stagnated or recorded a slight increase. In the remainder of Croatia, the trends indicate stagnation or decrease. A reduction in precipitation levels during the observed period was mostly pronounced in Dubrovnik (45 mm every 10 years), making this station stand out. This could be explained by its position close to the part of the Mediterranean where precipitation declines have been most pronounced (Norrant, Douguédroit, 2006). This particularly refers to Greece. Although many papers dealing with precipitations decrease in the Mediterranean have been published (e. g. Palutikof, 2003), the reasons for such a decline have not been fully explained, though an inverse relationship has been observed between decreasing precipitation levels and increasing air pressure (Norrant, Douguédroit, 2003). There are opinions that the changes in hydrological cycle are influenced by a higher concentration of greenhouse gases (Piervitali, Colacino, 2003). The correlation between precipitation trends and ENSO and NAO has also been investigated (Price et al., 1998;Rodò et al., 1997). Investigation of correlation between ENSO and precipitations in the Mediterranean resulted in different conclusions: some scientists think that the correlation can be established, and the other have the opposite opinion (Colocino et al., 2001;Mariotti et al., 2002 Na većini ostalih analiziranih postaja smanjenje nije veće od 20 mm u 10 godina. Iznimku čini Rijeka sa zabilježenim uzlaznim trendom. S obzirom na geografski položaj i cirkulacijske uvjete, Rijeka je specifična. Do nje dopiru vlažne zračne mase iz južnog kvadranta koje relativno niski jadranski otoci ne uspijevaju zaustaviti. Tek je planinsko zaleđe Rijeke dovoljna visinska prepreka za kondenzaciju vlažnog zraka. Osim toga, zbog koncentracije kopna, maritimni utjecaj u Rijeci je slabiji nego duž ostatka obale. Sličnih izuzetaka nalazimo i u ostalom dijelu Mediterana (Ben-Gai i dr., 1994.).
Kako bi se naglasila periodičnost količine padalina u Hrvatskoj, konstruirani su polinomni trendovi šestog stupnja. To je učinjeno posebno za postaje u unutrašnjosti Hrvatske (Slika 3.), At the majority of the remaining stations in Croatia, the decrease in precipitation is not more than 20 mm in 10 years, whereas the Rijeka station recorded an increase. According to its geographical position and circulation circumstances, the Rijeka area is specific in terms of precipitation quantity. The flow of air from the southern quadrant passes relatively easily over the Adriatic islands, as their altitude does not represent a significant barrier to the movement of wet air masses. However, the tall mountainous inland of Rijeka causes the humid air to rise, which results in condensation. Besides that, due to the land concentration, the maritime influence in Rijeka is less than in other coastal areas. Similar exceptional cases exist elsewhere in the Mediterranean (Ben-Gai, T. et al., 1994).
In order to emphasise the periodicity of precipitation levels in Croatia, a six-degree polynomial trend line was constructed. This was performed separately for the stations in continental Croatia (Figure 3), and separately for the coastal and island stations (Figure 4). In the majority of analysed stations, a cycle in which the maximum precipitation occurred in the mid to late 1960s was observed. At certain stations, it is evident that the cycle began just after 1950, while at other stations (Lastovo, Hvar, Split-Marjan, Slavonski Brod), its beginning within the analysed period cannot be determined with certainty, i.e. it began prior to 1950. At Lastovo and Slavonski Brod the maximum occurred earlier. This cycle ended in the 1980s, and somewhat later at certain Adriatic stations (Lastovo, Hvar, Šibenik, Split-Marjan, Pazin, Dubrovnik), i.e. around 1990. This marked the beginning of a second cycle, the end of which occurred between 2000 and 2010 at most of the stations. There are stations (e.g. Lastovo) that stand apart from such a cyclic scheme. The differences between the minimum and the maximum precipitation level within the cycle are generally greater at the coastal and island stations than at inland stations. That could be related to the strong precipitation seasonality during the year. Slika 4. Polinomni padalinski trendovi šestog stupnja za obalne i otočne postaje u Hrvatskoj Figure 4 Six-degree polynomial precipitation trends for coastal and island stations in Croatia