Current Regional Climate Change Studies in Hungary : a Review

an important issue for many reasons. The diversity of the local consequences in even a small area could be very high. Before policies can be brought forward mak¬ ing optimal use ofthe expected changes and limiting the adverse affects, it is necessary to explore the regional consequences of global change. For the Simulation of climatical processes. coupled atmosphere-ocean general circulation modeis (GCMs)

Current Regional Climate Change Studies in Hungary: a Review Istvän Matyasovszky, Tamäs Weidinger, Judit Bartholy and Zoltän Barcza, Budapest 1 Introduction The possible effect increasing concentrations of atmospheric greenhouse gases could have on global climate is an important issue for many reasons.The diversity of the local consequences in even a small area could be very high.Before policies can be brought forward mak¬ ing optimal use ofthe expected changes and limiting the adverse affects, it is necessary to explore the regional consequences of global change.For the Simulation of climatical processes.coupled at- mosphere-ocean general circulation modeis (GCMs) are common.The Standard procedure for assessing cli¬ mate change is to run a GCM using the atmospheric CO2 content of the period before the industrial revolution (1 xCO: case) and then to run it again under twofold CO2 content (2xCO:> case).Due to the relatively low horizontal resolution (a few hundred kilometres) and simplistic parameterisations (atmosphere-surface feed- backs, radiation processes, cloud and preeipitation forming, etc.) of these modeis, the validity ofthe results for small areas like the Carpathian Basin, is very limit¬ ed.For practical reasons, the need to convert the largescale output of GCMs into something more applicable for small-scale research is vital.The downscaling of GCM data follows three different approaches (Giorgi & Mearns 1991).Empirically speaking, independent of the factor causing global change, similar large-scale climate changes have much the same local consequences.This hypothesis makes it possible to use temporal or spatial analogies (Mika 1992, RÄcz 1999).The second approach to downscal¬ ing operates with meso-scale numerical modelling.
Meso-scale modeis use GCM Outputs as initial and boundary conditions (Giorgi & Marinucci 1992,  Marinucci et al. 1995, McGregor 1997).This tech¬ nique requires a substantial amount of modelling and Computer programming and there is at present no satis- factory long-term Simulation available to assess ex¬ treme conditions.Alternatively, use could be made of stochastic downscaling procedures, particularly as they incorporate elements of both approaches mentioned above.Stochastic downscaling makes use of two key elements: Large-scale atmospheric circulation and an element which links local surface variables and largescale circulation.For the latter use is made of observed data.Then, this model may be ulilised with GCM Out¬ puts characterising atmospheric circulation (Bogärdi etal. 1993          Mika (1988,   1991, 1992) used an empirical downscaling method to describe climate change scenarios in Hungary.Later, Bartholy et al. (1995, 1998) focused on smaller and more vulnerable regions in Hungary, namely the watershed of Balaton-Siö and the Great Hungarian Piain (Fig. 1) to test the stochastic downscaling technique.Causes and magnitude of climate change, as well as the conse¬ quences for hydrology, agriculture, forestry and energy were evaluated (Faragö et al. 1990, 1991).More re- cently, changes monitored in the Great Hungarian Piain have been associated with the effects of climate change (Mika et al. 1995, Kertesz et al. 1999).The article is divided into four parts: A summary of long-term trend analysis of annual mean temperature and preeipitation is followed by a presentation of im¬ portant empirical downscaling results.In contrast, the results of a stochastic downscaling are discussed.The article closes with a brief summary.
2 Temperature and preeipitation trends in Hungary Hungary is situated at the centre of the Carpathian Ba¬ sin (Fig. 1).84% of the country (93 030 km2) has an altitude less than 200m a.s.l..At the other end of the scale, only 2% of the country ranges above the 400m mark.The climate is both oceanic, and Continental, with slight Mediterranean influences.The mean annual tem¬ perature is 10CC The distribution of temperature re- flects latitudinal influence.Maximum values (11 -11.5°C) are recorded in the southeast, the northern re¬ gions are colder with mean annual temperatures around 8-9"C January, witha mean of-1 --4°C is the coldest month and a mean of 18 -22°C is recorded in July.Mean annual preeipitation lies between 600 -650mm, although the spatial Variation is high.Maximum values (800 -900mm) are recorded in the western and southwestern parts of the country: the driest regions are in Hortobägy (below 500mm) (see Fig. 1) and in the south¬ east.Preeipitation underlies a characteristic annual cycle.At the beginning ofthe summer, rainfall is greatest, with a drier period following in the second part of win¬ ter.Due 10 the Mediterranean influence, a small second preeipitation peak oecurs in the south-west around October-November.As is the case all over the world, pre¬ eipitation in Hungary is a highly variable meteorological element.A threefold difference between the wettest and the driest years is possible.and any month may suf¬ fer from a lack of preeipitation (Okolowicz 1977;   Peczely 1981).Deux regions sensible ele la Hongrie utilisees pour la modele de «downscaling» stochastique Molnär and Mika (1997) performed a linear trend analysis of temperature and preeipitation for 16 locations in Hungary using homogenised time series for the period between 1881 and 1990.For more information about the method, refer to Szentimrey (1995, 1996).
Interestingly, recordings at each ofthe stations indicates a general warming (with over 95% Statistical significance) with the average temperature rising by 1°C per Century.The rise in temperature varies insignificantly and corresponds to other results in Central Europe (Böhm 1992).Preeipitation on the other hand appears to be decreas- ing.At 10 of the stations.the values have on average a negative trend of -90mm per 100 years (95% significance).The ränge varies from -230mm to -27mm per 100 years.Earlier research (Koflanovits-Adämy & Szentimrey 1986) carried out in the Carpathian Basin for 84 stations offers similar results for the period be¬ tween 1901 and 1984.Bartholy and Pongräcz   (1998) emphasise the drift towards aridity in the region.Slatistically significant is likewise the increasing number of dry, warm years.For their research, Tar (1992) and Molnär (1996) analysed the frequency of years warmer or colder than average, as well as those years drier or wetter than the annual mean average.
Climate change analyses generally need only concentrate on temperature and preeipitation, as the Perform¬ ance of other meteorological factors is closely linked.
Research on the local influences of changing global cli¬ mate for Hungary and for the whole Carpathian Basin was carried out by Mika (1988, 1991, 1992) and Molnär and Mika (1997) using: (i) a Statistical rela¬ tionship between local meteorological elements and Northern Hemisphere temperature and ocean-continent temperature contrast; (ii) paleoclimatological ana- logues; (iii) a simple regional energy balance model.
Results of the change of regional climate as a funetion of the change of average hemisphere temperature are summarised in Table I.As was to be expected, global and local hemisphere temperature change fluetuated to the same degree.However, preeipitation shows the opposite reaction.A minor global warming of 0.5 to 1°C generates a considerably drier climate, but in the North¬ ern Hemisphere, a 4°C temperature increase leads to a noticeably more humid local climate.In the course of a moderate global warming.sunshine hours increase by 20%, and an annual dry phase is extended from 1.4 months per year to 2 months per year.Obviously, long-term changes in temperature and pre¬ eipitation may cause a number of ecological, agricultural and economical problems.Ecological scenarios include amongst others, drought, a drying out the upper soil horizons and a lowering of the ground water table.Thus, the demand for irrigation increases and more res- ervoirs have to be built.The probability of the above is underlined by the current Situation in the vast lowlands ofthe Danube and Tisza (Danube-Tisza Interfluve): The ground water table has been sinking here since the mid 1970s (Kertesz et al. 1999).
Crop response can clearly be considered a significant element of climatic and soil milieu change.Simulations of maize and wheat yields indicate a 10 -20% yield loss.However, it is possible that other factors linked to climatic and soil milieu change will compensate the negative effects on yield: e.g.increasing atmospheric CO2 concentration, milder winters, and modest denitri- Stochastic downscaling methods make use ofthe corre¬ lation between large-scale atmospheric circulation and hydrometeorological variables.Based on the correla¬ tion between observed data and GCM Output, future scenarios for hydrometeorological parameter can be es- timated using, for example, a 2xCO?climate (Bogärdi et al. 1993).The authors developed and applied such a model to two vulnerable regions in the Carpathian Basin (Fig. 1), namely the watershed of Balaton-Siö (Bartholy et al. 1995, WEiDiNGERet al. 1995) and the Great Hungarian Piain (Bartholy & Matyasovszky   1998a).Although not discussed in detail here, the meth¬ od was also applied to other climatic zones: The dry, Continental climate of Nebraska (Matyasovszky et al.  1994, Mearns et al. 1999), the dry subtropical climte of Arizona (Bartholy & Duckstein 1994), the Mediter¬ ranean climate of Greece (Matyasovszky et al. 1995)  and the Alpine region of Austria (Nachtnebel et al.   1996).Computations were carried out using two differ¬ ent modeis: An ECHAM (a coupled ocean-atmosphere GCM developed by the Max Planck Institute, Germany (Cubash etal. 1991), and a GCM of the Canadian Cli- Ungarn, nachgewiesen mit einer stochastischen «downscaling» Technik bei verdoppelter atmosphäri¬ scher COz-Konzentration Changements de temperature et de preeipitation prevus en Hongrie.obtenue par la technique de «downscaling» stochastique.pour un doublement des concenlrations eitmospheriques de CO2 mate Centre (CCC).Global warming according to ECHAM is expected to increase by 1.5°C, whereas CCC predicts a 3.5°C global temperature increase (Boer et al. 1984).For the results computed using the ECHAM model, see Table 2. smaller than the predicted global mean temperature in¬ crease of 1.5"C (ECHAM) and corresponding values in  & Mika 1997).
Because of the spatial-temporal intermittent character of preeipitation.its prediction proves to be more com¬ plex than that of temperature (Bogärdi et al. 1993).Accordingly.both oecurrence and magnitude probability need to be calculated.Furthermore, wet/dry diurnal duration should be considered.Important results of our investigation on the Great Hungarian Piain are: Preeipi¬ tation frequency decreases and preeipitation magnitude during wet periods remains the same or indeed.could increase.Thus, preeipitation patterns are expected to On the other hand, preeipitation magnitude stays very much the same.Spatial variability is expected to be minimal (Bartholy & Matyasovszky 1998).Chang¬ es in preeipitation magnitude fluetuate between -10 -+5% in the summer months (Fig. 3), and between -15 --5% in the winter months.On the whole, annual preeip¬ itation change is in the ränge of 10%.
A more complex analysis was carried out for the water¬ shed Balaton-Siö on the basis of observed data from 28 preeipitation stations (Bartholy et al. 1995).Both the frequency and the amount of preeipitation on wet days is expected to decrease substantially in summer.The forecast for spatial distribution of preeipitation in the winter months is slightly more complicated.Preeipita¬ tion frequency will definitely decrease, but the amount of wet days will decrease in northern part of the water¬ shed and increase over the southern part. 25-35% less preeipitation is expected in the summer months; the winter months.with 0 -10% less preeipitation will only be slightly drier than is presently the case.Because the forecasted preeipitation fluctuations for the Great Hungarian Piain were smaller than those calculat¬ ed for the Balaton-Siö watershed, the development to¬ wards a drier climate will probably affect the Great Piain to a lesser degree than the watershed.A modified version of our stochastic downscaling meth- odology was also applied for predicting evaporation rates on Lake Balaton under a 2xCÜ2 climate.By adapting pan evaporation readings of surrounding locations, lake evaporation could be modelled (Weidinger et al.  1995).The calculated increase of 3 -4% in the summer months correlates well with the expected increase in temperature.
The spatial variability of temperatures calculated under a 2xC02 climate for the Great Hungarian Piain (Fig. 2) is minimal.The seasonal trend with 0.1 -0.5°C is posi¬ tive.Interestingly, the average temperatures in autumn are expected to exceed l.5°C (Bartholy & Matya¬ sovszky 1998).The expected mean annual temperature change is thus about +0.7°C, a value considerably 5

Summary
The main conclusions of the linear trend analysis for time series of homogenised annual mean temperature and annual preeipitation amounts in Hungary are as follows: Mean annual temperature is increasing at a rate of 1CC per 100 years (95% significance).Mean annual preeipitation is decreasing by 90mm every 100 years (95% significance).As becomes obvious from the points above, a trend towards aridification has set in.This tendency will probably be strengthened by the increasing concentration of atmospheric greenhouse gases.
Our stochastic downscaling modelling of two vulnera¬ ble climatic regions in Hungary, based on a twofold CO2 concentration, led to the following conclusions: A regional increase of the mean annual temperature by 0.7°C in a prevailing

1996).
The frequency of wet days is expected to decrease in future, whereas preeipitation magnitude will be increasingly variable.
The Balaton-Siö watershed is expected to have a noticeably greater preeipitation deficit than the Great Hungarian Piain.The scale of our climate change scenarios for preeipitation is similar to that used in previous estimations based on empirical downscaling modelling.The results however, highlight the neces- sity for greater spatial differentiation as the disparity between even small areas can be high.
Summary: Current Regional Climate Change Studies in Hungary: a Review After focusing on the changes in Hungarian tempera¬ ture and preeipitation during this Century, possible hydrological, agricultural and ecological consequences of a future climate change are described.These results have been obtained using a modified version of empiri¬ cal downscaling techniques, developed to analyse the local effects of global climate change in a twofold con¬ centration of atmospheric greenhouse gases scenario.In addition, regional changes in temperature and preeipita¬ tion were examined with the help of the more specific stochastic downscaling method.The climate of Hunga¬ ry has become warmer and drier over the last Century.It is to be expected that an increasing concentration of at¬ mospheric greenhouse gases will enhance the tendency towards aridification.
Resume : Une resume des etudes de la Situation actuelle des climatique changements des regions en Hongrie La premiere partie de cet article se concentre sur les changements de temperature et de preeipitation pendant le 20e siecle en Hongrie.Ensuite, les consequences pos¬ sibles d'un changement de climat futur sur les domaines hydrologique, agricoles et ecologique possibles sont recapitulees.Ces resultats ont ete obtenu apres Papplication des techniques de «downscaling» empriques developpees pour analyser les effets locaux du changement global de climat en fonction d'un doublement de la concentration des gaz ä effet locaux du chan¬ gement global de climat en fonction d'un doublement de la concentration des gaz ä effet de serre.Finalement, Devolution de la temperature ainsi que les changements de pre'cipitation obtenus avec une me'thode downscaling stochastique plus elaboree, sont presentes.Le climat de la Hongrie a dejä connu un rechauffement et un des- sechement dans ce siecle.Dans un climat influence par une hausse de la teneur atmospherique des gaz ä effet de serre, cette tendance ira en s'aecentuant.

Fig. 3 :
Fig.2: Temperature changes for selected stations on the Great Hungarian Piain as forecasted by a stochastic down¬ scaling model with twofold atmospheric CO2 concentration Erwartete Temperaturänderungen für ausgewählte Stationen im Grossen Ungarischen Tiefland, nachgewiesen von einer stochastischen «downscaling» Technik bei verdoppelter atmosphärischer CÖ2-Konzentration Changements de temperature prevus pour une selection de stations dans la Grande Plane Hongroise obtenue par la technique de «downscaling» stochastique, pour un doublement des concentrations atmospheriques de CO2 fication due to a drier climate (Bacsi & Hunkär 1994, Harnos 1998, KovÄcs & Dunkel 1998).Of further interest is research on the relationship between climate change and possible plant migration Mätyäs (1997) and Koväcs-Läng et al. (1998).4Local effects of global climate change in Hungary based on stochastic downscaling modelling Temperature and preeipitation change in Hunga¬ ry as forecasted by a stochastic downscaling model with twofold atmospheric CO2 concentration Erwartete Niederschlags-und Temperaturänderungen in fluetuate more in future (Matyasovszky et al. 1994, Matyasovszky et al. 1995, Nachtnebel et al. 1996), The local effects of global climate change have been studied in Hungary since the mid-80s.

Table 1 (
after Molnär 1.5CC global warming (ac¬ cording to ECHAM calculations) is considerably smaller than previous estimates.Anticipated temperature changes are considerably smaller than changes calculated for other regions at the same latitude (Matyasovszky et al. 1994; Matyasovszky et al. 1995, Nachtnebel et al.