Between Clouds, Carpathian Mountains-EUROPHOTOMETEO2012
Aeolian Harp, Dobrogea-EUROPHOTOMETEO2012
Sky Paintings, Cluj Napoca-EUROPHOTOMETEO2012
Peace, Balea Lake-EUROPHOTOMETEO2012
Water Magic, Alba County-EUROPHOTOMETEO2012
Mistic River, Alba County-EUROPHOTOMETEO2012
Summer Storm, Targoviste-EUROPHOTOMETEO2012
The Morning Fog-Ceata de dimineata-Letea, Tulcea-EUROPHOTOMETEO2012
Waiting for the Rain, Brasov-EUROPHOTOMETEO2012

Report 2014

Scientific report

on the implementation of the project "Changes in climate extremes and associated impact in hydrological events in Romania" (CLIMHYDEX) over the period 2012-2014


During the reported period (May, 11.2012 - Nov., 30 2014), the research activity has been developed according to the approved project schedule. The synthesis of the results is presented below, ordered by work packages (WP) and activities. The results obtained during this period have been presented at several national and international thematic conferences (43), being also subject to 21 articles: 15 published (11-ISI articles, 4-BDI articles) and 6 under evaluation. Additionally, 3 book chapters and 4 articles in other journals have been also published. The list of presentations and publications are posted on the project website ( under the sections “Meetings" and "Publications", respectively. The list of publications is also attached to this report (Annex 1).


WP2 – Data


2.1. Seasonal indexes associated to climatic and hydrological extremes.

As planned, NMA and NIHWM have computed all seasonal indexes associated to climatological / hydrological extreme events, according to the deadline of this activity (D.2.1, completed).


2.1.1 Climatic extremes.

Starting from daily measurements of minimum and maximum air temperature, precipitation and snow depth, from all available stations with full data records  for the reference period 1961-2010, the NMA has computed all seasonal indexes associated to climatic extremes, according to the schedule, as follows: the frequency of very warm days/nights (maximum/minimum temperature >= 90th percentile): Frtmax90 (1), Frtmin90 (2); longest period of very warm days/nights: Dtmax90 (3); Dtmin90 (4); longest dry period: Dmaxpp0 (5); frequency of very wet days (daily precipitation amount above 90th percentile): Frpp90(6); longest very wet period: Dmaxpp90 (7); frequency of snow/rain showers: Frap (8), Fraz (9); meteorological drought index, represented by the standardized precipitation index SPI (10); thermal stress indexes (warm/cold): ITU (11) / IR (12); frequency of days with ITU/IR above/below their critical thresholds: ITU >= 80 (FrITU (13)), IR <= 32 (FrIR (14)); maximum daily precipitation amount: Ppmaxz(15); maximum daily temperature:  Tmax (16); minimum daily temperature: Tmin (17); maximum snow depth: Grmaxz(18); pedological drought index- IP (19), represented by the soil moisture content for agricultural areas in Romania, at different soil depths (0-0.2 m, 0-0.5 m, 0-1 m); maximum annual precipitation amount recorded within 5 to 1440 minutes resulted from pluviographs (1965-2007) at 45 meteorological stations; a special new index expressing the maximum intensity (mm/min.) of continuous rainfalls (Imaxp) has been considered: nine stations representing various physical-geographical conditions in Romania have been considered for this analysis (1951-2007), among which the Bucuresti-Filaret station covers the period 1898-2008.


2.1.2. Hydrological extremes.

The NIHWM filled in the mean monthly discharges data series over the longest available observation periods for the Orsova station (on the River Danube) over the interval 1921-2010 and for other stations monitoring the first degree river confluences and the ones from certain representative basins controlling certain altitude steps (1931 -2010). For this, level observations and multiannual keys for the period 1931-1960, correlations for the levels between correspondent stations, as well as rainfall-runoff analysis methods, have been used. Finally, continuous and homogenous data series for the period 1931-2010 at 45 hydrometric stations were generated. The values of the 80% and 20% quantiles were calculated, used for sampling of the dry/rainy periods and the periods with scarce/excessive water volume.

2.2. High-resolution meteorological and hydrological datasets


2.2.1. High resolution spatial and temporal interpolation of meteorological data

Temperature and precipitation gridded datasets were constructed within the CLIMHYDEX project for evaluating the performance of regional climate models (RCMs), for developing Statistical Downscaling (SD) models needed in WP4 (4.2) and as input data for hydrological models (WP5). The time interval is 1961-2010 for temperature and 1975-2010 for precipitation.

For interpolating 6-hour temperatures and precipitation, a two-step multivariate gridding approach was used. First, the hourly normal maps, considered as multiannual average (1961-2010) of air temperature for each hour (4 meteorological terms) of a standard year (366 days) were interpolated. In case of precipitation the monthly normal maps were used for the period 1975-2010. In this step, the Residual Kriging method was used with potential predictors derived from the SRTM (Shuttle Radar Topography Mission) DEM (Digital Elevation Model) and CORINE Land Cover product. For interpolating the residuals of the regression model, three gridding methods were tested: Multiquadratic (MQ), Ordinary Kriging (OK) and 3D Kriging (using time as a third dimension). It was found that the MQ method is the most appropriate. The cross validation was used to establish the model skill. In the second step, in case of temperature, the anomalies of each hour, day, and year for the period 1961-2010 were computed. The anomalies were interpolated using the same methods applied for gridding regression residuals. The final hourly surface air temperature maps were obtained by summing the maps from first step with the anomaly maps. For precipitation, the rations against the 1975-2010 climatology were used instead of anomalies and the final hourly precipitation maps were obtained by multiplying the maps from the first step with the ration maps. The final outcome of this work is a gridded temperature and precipitation dataset, at a six-hour time step, available in high spatial resolution (1kmx1km)

A 10 minutes gridded dataset was also constructed for the air temperature over a shorter period of time (2007-2012), using approximately the same methodology. Because the air temperature data was available for 1-hour time interval, first the temperature surfaces were interpolated at this time resolution, after that, the final 10-minutes grids were obtained by applying a linear interpolation algorithm in time domain.

The new precipitation gridded datasets were compared, both quantitatively and qualitatively, with two different sets spanning over the same period of time: CarpatClim ( and E-OBS ( The two sets of gridded data are constructed on a daily temporal resolution, at different spatial resolutions. The CarpatClim daily grids have a spatial resolution of 0.1°×0.1°, covering the area between latitudes 44°N and 50°N, and longitudes 17°E and 27°E. The E-OBS data files contain daily gridded data covering the area between 25°N - 75°N and 40°W - 75°E, available on a 0.25° regular lat-lon grid. For the pattern analysis, the grids of multi-annual precipitation mean (1975-2010), aggregated form 6-hour grids (CLIMHYDEX), were compared with grids of the same parameter calculated from CarpatClim, E-OBS and precipitation grid interpolated by RK having as target variable multi-annual mean calculated at stations (multiannual mean) (Figure 1). The map of the multiannual values computed at stations was considered in this analysis the reference dataset (Figure 1 top left). The CLIMHYDEX dataset reflects the impact of topographic features in spatial distribution of the target variable, as well as the reference (multiannual mean) and the CarpatClim data. The mountains' influence on rainfall is not well emphasized on the E-OBS map; the low spatial resolution does seem to have a negative influence on interpolations, several intervals with large amounts of precipitations being underestimated at higher altitudes. On the south-eastern part of Romania, CLIMHYDEX and E-OBS have a similar pattern as the reference data, lower values of rainfall covering about the same area on the three maps.


As an application, three extreme precipitation patterns are calculated at grid pixel within the 1975-2010 period as follows:

  • R10MM Annual count of days when precipitation ≥ 10mm;
  • RX5DAY, Annual maximum 5-day precipitation;
  • R95P% Precipitation fraction of annual total precipitation due to daily precipitation > 95th percentile.

The results related to precipitation are included in a peer-reviewed paper (Dumitrescu et al., 2014, under review).These patterns were used as reference patterns to validate the EURO-CORDEX RCMs (see 4.3).


Figure 1 Precipitation amounts (mm) - multiannual mean (1975-2010): interpolated by RK using as target variable the multi-annual means calculated at stations (top left); aggregated from hourly grids (top right); CarpatClim (low left); E-OBS (low right).


2.2.2. Remote sensing data

Remote sensing data (D2.3, completed) has been commonly agreed with NIHWM, and prepared by NMA, as follows: snow cover extent, at 8-day synthesis derived from MODIS at 500 m spatial resolution for Barlad, Crisul Alb, and Tinoasa-Ciurea river basins, for the interval January-April of the years 2005, 2006, and 2012, respectively; land cover, extracted from Corine Land Cover (CLC1990, CLC2000, CLC2006) for the selected river basins, at a spatial resolution of < 50 m for CLC1990, and <25 m for CLC2000 and CLC 2006; MODIS products of reflectance, 8-day synthesis, 1 km spatial resolution, used for computing datasets of vegetation indexes (years 2003, 2005, 2007 and 2008), namely: Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Normalized Difference Infrared Index (NDII).


2.2.3. Estimation of precipitation (6 minutes temporal resolution) from radar information

Weather radar-based rainfall accumulation is widely used by both meteorologists and hydrologists firstly to detect and monitor severe weather, and when the case to issue severe weather warnings, and in quantitative precipitation estimation (QPE) and flood modeling, respectively. Atmospheric precipitation varies both in space and time; therefore, certain hydro-meteorological applications require high spatiotemporal resolution data to accurately describe the atmospheric conditions associated with the rainfall events. Within this research, QPE improvement in the area of a test river basin, namely Barlad River Basin, was aimed through adjusting the radar-based rainfall estimations using the ground measurements. To achieve this, the mean field bias approach was used on the data provided by WSR-98D weather radar system that is operational in the close proximity of the test basin, at Barnova. The radar data has a spatial resolution of 1x1 km, being sampled at every approximately 6 minutes, while the surface observations (rainfall accumulations) are those made at the weather stations at every 10 minutes. The datasets spans over the 2006–2012 period, the study being focused only on the convective season (April–October interval).

The general analysis involved four main steps: a) conversion of radar reflectivity into rain rate, b) calculation of 6 minutes radar-based accumulations, c) mean field bias calculation, and d) adjusting the radar-based rainfall estimations. Therefore, after the conversion of the reflectivity, the rain rate was integrated in time to estimate the radar accumulations over a given interval. Surface rainfall observations at ground were afterwards used to derive the mean field bias of the radar estimations, bias used to obtain the final grids of radar rainfall field at the level of the Barlad River Basin. The final result of this approach was a 1x1 km and 6 minutes resolutions grid of the spatial rainfall distribution at the basin level. The assessment of the final results revealed that the radar-based rainfall accumulations slightly underestimate the precipitation in the pilot basin, but after the adjustment was performed the results were significantly improved. The adjusted radar estimations were much closer to the values registered at the ground than in the case of unadjusted radar estimations. Consequently, the method used was proven to be useful, and with great potential to be implemented operationally. The final grids were made available to the Partner no. 2 (INHGA) to be used as input into hydrological models. The results of this research were presented at 3 scientific events, from which one international (EGU-2013) and two national (MeteoRomania scientific annual conference and INHGA scientific conference


2.2.4. High spatial resolution data used as input in hydrological models

The fine resolution spatial data, used as inputs for hydrological models (D2.6, finalized), were calculated by the NIHWM for two medium size river basins (Crişul Alb and Bârlad) and two representative river basins (Moneasa and Tinoasa-Ciurea) and refers to the performance of certain thematic layers (hydrographic network, reservoirs, hydrometric stations, etc.), production of Digital Terrain Models, generation of morphometric parameters of the hydrometric station catchments, as well as of the following parameters: field capacity, runoff coefficient, evaporation capacity, using the CN (Curve Number) index. The map with the maximum retention capacity was achieved. Soil moisture in the unsaturated area was determined based on the relations described in the Vidra hydrological model.


2.3. Outputs from regional (RCMs) and global (GCMs) climate models (D2.4)


The daily data of maximum / minimum air temperature and precipitation amount resulted from the simulations of 7 RCMs (two more than the planned minimum of 5) elaborated within the FP6 ENSEMBLES project ( for the period 1951-2100, were downloaded by the CO. These data will be used in the next stages of the project in order to compute the climatic extremes indices mentioned at section 2.1 for the actual period, and changes in their regime in the future for 2020-2050 and 2070-2100 time intervals, under the A1B emission scenario. Taking into account the fact that in the last years other GCM/RCM simulations become available through the CORDEX experiment, three such RCMs have been considered under various RCP scenarios. Some of these simulations have already been analysed in the reported period (see 4.3).


WP3 – Mechanisms controlling the variability of climatic extremes


Four activities have been planned for this WP and they have been mainly finalised in the reported period, apart from the preparation of some papers for publications summarising the results carried out in this WP. The WP3 comprises the analysis of a wide range of indices describing the climate extremes (mainly the seasonal ones presented in section 2.1) through a comprehensive approach using advanced and complex statistical techniques. That is why, the most published papers in the reported period comprise the WP3 activities; in 2014, 8 such articles were published/accepted in ISI journals. The WP3 results have also been presented in many scientific conferences.


3.1. Characteristics of spatial-temporal variability of seasonal climatic / hydrological extremes (D3.1a)


3.1.1. Climate extremes

The linear trends and shifts in the mean have been analysed for all the 21 seasonal indices presented in WP2 using the Mann-Kendall and Pettitt tests, respectively. Additional seasonal time series regarding the frequency of Cumulonimbus (Cb) clouds and rainfall days have been also analysed. The EOF (empirical orthogonal functions) has been used to identify the main modes of spatial and temporal variability. The results can be summarised as follows:


a) Significant increasing trends for the six temperature extremes (Frtmax90, Frtmin90, Dtmax90, Dtmin90, Tmax, and Tmin) have been detected in all seasons, except for autumn and Dtmax90-spring. The increase rate is more enhanced in summer and less enhanced in spring. Regarding the spatial pattern of the trend magnitude, some differences can be revealed between winter and summer. In wintertime, the increasing trend is higher over the extra-Carpathian regions (more pronounced in southern and south-eastern regions) and is not significant over some small intra-Carpathian regions. In summer, the trend is significant over the entire country, showing an almost homogeneous magnitude. A multi-decadal variability can be revealed, especially in summertime, when longer time series are examined (e.g. Frtmax90 at Bucuresti-Filaret station, 1901-2010) that could be associated to AMO (Atlantic Multi-decadal Oscillation), as many previous papers have shown for the European climate extremes. These results are presented in Busuioc et al. (2014a). A similar climate signal has been found for other more complex temperature indices (e,g thermal stress indices) as presented by Dobrinescu et al. (2014) (under review).

b) Regarding precipitation extremes (Frpp90, Dmaxpp90, Dmaxpp0), the climate signal is not as clear as for temperature extremes. Significant increasing trends over some areas in the Frpp90 during autumn, ppmaxd during summer and autumn, Dmaxpp0 during summer (mainly recorded over south-western, central and western areas) were detected. In the remainder of cases, the linear trends are not significant (or are significant only for small areas).  However, a strong decadal /multi-decadal component could be revealed, more evident for longer time series (1901-2010) that is in agreement with the findings presented on the European scale by other papers and could be associated to NAO (North Atlantic



Figure 2. The standardized values of the Frtmax90 (left)  and Dmaxpp0 (right) calculated for  Bucuresti-Filaret station (1901-2010), in comparison with the standardised values of AMO, for  summer (Busuioc et al., 2014a).

Oscillation) in wintertime and AMO in summertime; these results are presented in Busuioc et al. (2014a).

As an example, Figure 2 shows the standardized values of the Frtmax90 and Dmaxpp0 calculated for Bucuresti-Filaret station (1901-2010), in comparison with the standardised values of AMO in summer and Figure 3 shows the linear trend of Frtmax90 and Dmaxpp0 for winter and summer.


Similar climate signal has been detected for other precipitation indices such as standardised precipitation index (SPI, Cheval et al., 2014). Even if the increase in maximum daily precipitation is significant at the 5% level only on a limited number of stations in summer, the rate of change in maximum daily precipitation in Romania in response to changes in near-surface temperature is increasing at most of the locations and is much larger than that in the daily mean precipitation (which is around zero). This result suggests the relative contribution of thermodynamics -related to Clausius-Clapeyron (CC) Equation -to changes in precipitation characteristics in summer, as previous theoretical and application studies highlighted; this aspect is thoroughly analysed in other paper at 9 representative stations (Busuioc et al., 2014b, under review)., showing a similar climate signal at all stations: the 90th percentile shows a dependence close to the CC relation, the 99th one - close to CC only for temperatures below to 120C and then a super CC scaling (about twice CC relation), while the 99.9th one exhibits a twofold CC scaling.


Frtmax90 winter

Frtmax90 summer


Dmaxpp0 winter


Dmaxpp0 summer


Figure 3. Linear trend (number of days over 1962-2010) of Frtmax90 and Dmaxpp0 for winter and summer. In case of the Dmaxpp0, the cross pattern areas show trends at the 5% significance level. In case of the Frtmax90, the trends are significant at the 5% level over the entire country. (Busuioc et al,, 2014a).

c) Regarding the pedological drought, expressed by soil moisture reserves over the 1970-2012 period at 34 agrometeorological stations with complete data set, the main conclusions are dependent on the type of crop and developing phase. During peak water demand in the not irrigated corn crop, the moisture reserves at most analyzed stations show a high variability, depending on recorded rainfall and their distribution, maximum air temperatures and number of days during the heat of July and August; not significant linear trend can be revealed for these time series. The prevailing values of soil water content are generally located in low limits. At the irrigated corn crop, a slight increase tendency for the number of months with pedological drought at all agrometeorological stations is noticed (April to September). For the winter wheat, during seedbed preparation for culture setting up and beginning of sowing optimal time (September), the moisture content in 0-20 cm soil layer was generally located in low and very low limits over an extended period, showing extreme, strong and moderate pedological drought, especially in southern and southeastern areas. However, an upward shift around 1993 to sufficient provision of moisture reserve at all stations, can be revealed. Similar characteristics have been found for the time series related to peak water demand period (May and June) of the winter wheat, with the difference that an upward shift around 2003 to sufficient provision of moisture reserve can be revealed only for the western stations.


d) The frequency of rain showers in Romania exhibits a significant increasing trend in spring and summer over almost the entire country that is in agreement with the increase in the frequency of Cumulonimbus clouds. However, not significant changes in the frequency of rainy days have been found showing a shift in the nature of precipitation towards more showers, despite no significant changes in the seasonal amount and the daily extremes (Busuioc et all., 2014b, under review).


e) Using the EOF technique, it was found that all indices presented at a) and b), show similar principal modes of spatial variability (i.e. the same sign over the entire country, with a higher magnitude over the southern and eastern areas in wintertime and generally over the southwestern-western areas in summertime) but their fraction of explained variance is higher for temperature extremes. The second mode exhibits a dipole structure with differences between winter and summer: in winter the Carpathian Chain influence is noted, with an opposite variability between the intra-Carpathian and extra-Carpathian regions, while in summer a northeast-southwest gradient can be revealed. The multifield EOF analysis applied to the combination of several climate extremes reveals, as principal mode, the same sign of simultaneous variability for all thermal extremes and the opposite variability between thermal and precipitation extremes, except for Dmaxpp0, which shows opposite sign against the other precipitation extremes but generally in- phase with temperature extremes.

3.1.2. Hydrological extremes    

To highlight the dry periods and theirs severity on the Romanian watersheds, the standardized flow index was calculated (Standardized Flow Index - SFI), expressing the hydrological drought. The SFI index is based on a similar procedure as for standardized precipitation index (SPI) but the mean monthly flow series in natural regime are used, instead of monthly total precipitation amounts. The change points in the seasonal SFI have been analysed using the Pettitt test. Significant downward shifts (5% level) in summer and autumn around ‘80 years have been identified for the Jiu, Vedea and Ialomita river basins. For other river basins (Cris, Mures Timis, Olt, Arges and Siret), a 10% statistical significant downward shift between 1963 and 1968 was found. The results of this study have been presented at various conferences and are published by Ionita et al. (2014), Borcan, et al. (2014) and Retegan & Borcan (2014).


3.1.3. Homogeneous areas with respect to extreme variability

  1. Climate extremes

Considering four seasonal precipitation extreme indices (Frpp90, Dmaxpp0, Dmaxpp90, Ppmax24) and total seasonal precipitation amount at 98 meteorological stations over the 1961-2010 period, the time series of their standardised values have been computed (e.g. seasonal anomalies divided on standard deviation). These time series have been further used to identify homogeneous regions for each of these parameters, in terms of their variability characteristics. The cluster analysis has been used to reach this objective based on the STATISTICA 8 software. More clustering methods have been tested and, finally, the Ward-Manhattan method has been selected. For each of the five precipitation indices, 7 homogeneous regions have been identified, which are quite similar each other, such as: intra-Carpathian region, southwestern, southern, southeastern, northeastern, northwestern regions.

This result shows that physical reasons (e.g. atmospheric circulation) corroborated with local factors (e.g. Carpathians, Black Sea, etc) are responsible for this behaviour. To prove this, the correlation between several circulation indices (NAO, AMO, EA, EAWR) and regional averages of precipitation indices (presented above) has been computed. In wintertime, a significant negative correlation with NAO and EAWR (stronger for NAO) has been found for all indices, (except for Dmaxpp0 when positive correlation was found), with some differences from one index to another. In summertime, significant correlations with AMO and EA was found for southern and southwestern regions only for the Dmaxpp0.


  1. Hydrological extremes

Based on the characteristics of four hydro-climatic indices, a new index MEOFIA (Ensemble indices of development multivariate EOF) is proposed. Homogeneous areas are identified based on distribution of monthly discharges recorded at 27 gauging stations in the period 1931-1998 as well as on rotated EOFs for a set of Palmer type indices. In this way, it is highlighted the capacity of each index to capture the spatial-temporal distribution features of hydro-climatic extreme events over the Romanian territory, as well as the ability of their ensemble to maximize the overall specific information for each season.

Finally, the most important genetic factor of extreme events is analysed, mainly the sequence of circulation types on Atlantic-European sector (-50V, 40E, 30N, 65N), in connection with the hydrological-precipitation regime over the Romanian territory, by using a Markov chain model with hidden states (NHMM). The synoptic analysis of typical situations for drought and excessive wet periods (rainfall generating high flow) revealed a dipolar pattern of the sea level pressure (SLP) influencing the hydro-climatic regime in South-Eastern Europe. By conditional analogy method, the SLP patterns associated to drought and non-drought states in Romania have been selected. These results have been published by Mares et al. (2014).


3.2 Connection between the variability of climate extremes în Romania and simultaneous variability of large-scale climate anomalies (CCA method)


By CCA, the optimum linear combination for two multi-dimensional vectors (predictand – the spatial vector of climate extremes and predictor – the spatial vector of certain large-scale variables) and pairs of patterns are selected so that their associated time series be maximum correlated. By construction, the CCA method allows a physical interpretation of the mechanism controlling the variability of the regional climate parameters. As a novelty in the CLIMHYDEX project, the combination between more climate extremes (temperature and precipitation) has been considered as predictand. As large-scale predictors, the combination between dynamic and thermodynamic factors has been used. New predictors such as thermodynamic instability and total precipitable water have been also considered. Three types of mechanisms have been analysed by the CO as presented in the following.


3.2.1. Mechanisms controlling the simultaneous variability of temperature and precipitation extremes


The CCA applied between combined large-scale predictors and combined selected climate extremes in Romania (Frtmax90, Frpp90, Dmaxpp0) was noted to be a skilful tool to find plausible explanations from the physical point of view of the large-scale mechanisms responsible for the characteristics of the climate


CCA1 (SLP +T850)

R=0.87        var=42%


CCA1 SH700       

Var= 42%


CCA1 (Frtmax90+Frpp90)













Figure 4.  Patterns of the first CCA pair of the combined predictors (top, SLP (contour)+T850 (shaded)-(a), SH700-(b)) and combined Frtmax90 (contour) and Frpp90 (shaded)(bottom, (c)). The explained variances and canonical correlation coefficients are displayed.

extremes’ variability, especially the simultaneous variability of several climate extremes, as presented below. To our knowledge, this technique has been applied for the first time, at least in the climate extreme analysis.  These results have been published by Busuioc et al. (2014a) and can be summarised as follows:

- For winter, it was found that the thermodynamic factor (represented here by air temperature anomalies at 850 hPa) mainly controls the temperature extremes in Romania, while the dynamic one (represented here by the surface circulation, e.g. sea level pressure anomalies) roughly controls the magnitude pattern of the temperature extremes. Regarding the precipitation extremes, the role of the two factors is reversed. The specific humidity completes the picture of this mechanism. Therefore, the significant increase in the temperature extremes and not statistically significant trend  in  precipitation extremes in wintertime could be explained by the combination of two plausible mechanisms given by the first two CCA pairs, with very strongly correlated associated time series (0.88, 0.82), as follows: more frequent positive T850 anomalies over Romania simultaneous with more frequent anticyclonic structures and negative SH700 anomalies are associated to a significant increase in the frequency of very warm days and the length of dry intervals, as well as decrease in frequency of very wet days (CCA1); the predictand CCA1 patterns are similar to their trend patterns; more frequent cyclonic structures over Romania simultaneously with positive T850 anomalies and positive SH700 anomalies are associated to a significant increase in the frequency of very warm and wet days and shorter dry intervals (CCA2);


-  In summertime, the thermodynamic factor is dominant for both temperature and precipitation extremes analysed in this paper. For temperature extremes, the T850 alone could explain their variability characteristics, while for precipitation extremes (frequency and duration) the SH700 has the dominant role, except for the maximum duration of dry intervals (Dmaxpp0) that is controlled by a combination between T850 and SH700 anomalies. This result is in agreement with the previous studies related to the convective nature of summer precipitation in Romania. Therefore, the simultaneous significant increase in the summer frequency of very warm days and in the maximum length of dry intervals (higher anomalies in the southeastern, central and southwestern areas) can be mainly explained by the increase in frequency of strong positive T850 anomalies over the entire Romania simultaneously with an increasing frequency of negative SH700 anomalies (equivalent with the SH700 decrease).  The predictand patterns are similar with their trend patterns.


            Finally, we could conclude that the analysis of the simultaneous variability of several climate extremes in Romania, in connection with the simultaneous variability of physically connected large-scale climate variables, provides a useful tool to find more plausible physical mechanisms explaining the observed changes in the regime of climate extremes in Romania. The connections found in this study are strong and explain a great of the total observed variance, showing that these results can be used in a future study to build skilful statistical downscaling models, simultaneously for several seasonal climate extremes, giving the results more physical coherence, which could add value to these models.


3.2.2. Mechanisms controlling the variability of the rain shower frequency în Romnaia


The lftx4-Best Lifted Index (4-layer) has been selected as main predictor in this case. Lifted Index LI provides an estimation of the thermodynamic instability in the free atmosphere and represents the difference between environmental temperature at 500mb and the temperature which a parcel will achieve if it is lifted dry-adiabatically from the surface to its lifted condensation level (LCL) and then moist-adiabatically to 500mb. The lftx4 refers to the lowest value of LI, computed using parcels from four different layers near the ground surface. Other predictors refer to large-scale sea level pressure (SLP) and precipitable water (PW) considered as ingredients in generating convective rains. The seasonal (spring and summer) frequencies of the rain showers (Fr-RS) recorded at 81 meteorological stations over 1961-2010, uniformly distributed over Romania, have been used as predictands.


The CCA supplies the optimum large-scale mechanisms responsible for changes in the frequency of rain showers in Romania, represented by the first CCA pair, showing a strong connection and accounting for an important fraction of the observed variance of the rain shower frequency. The lftx4 pattern exhibits a dipole structure with a nucleus of thermodynamic instability centred over Romania. The combination of more ingredients (lftx4, SLP, PW) accompanying the rain showers' generation, leads to a stronger connection. Results reveal that the dynamic factor given by the surface circulation is an important additional ingredient for spring, while the precipitable water is important for summer, showing the mixed character of rain showers in spring (large-scale combined with convective) and the strong convective character in summer. The time series associated with this CCA pair exhibit a significant increasing trend, justifying from physical point of view the observed changes in the frequency of rain showers in Romania. These results can further be used in developing skilful statistical downscaling models to project future changes in the rain shower frequency at local scale in Romania, which is not possible from the direct RCM simulations. These findings are presented by Busuioc et al. (2014b) in a peer-reviewed journal (under review).


3.2.3. Mechanisms controlling the variability of thermal stress indices in Romania


Other indices quantifying the direct discomfort felt by the human body, such as the heat index based on air temperature and relative humidity (THI) and cold index (WCT) based on air temperature and wind speed, usually called thermal stress indices, have so far been less analysed from the climatological point of view (in terms of trends and mechanisms controlling their variability) and they have a high impact on human health. In this section we summarise the mechanisms explaining their changes presented in section 3.1.

It was found that the main large-scale mechanism responsible for the spatial and temporal behaviour of the two stress indices is given by the first CCA pair derived from the connection between the combination of T850 with SLP (thermodynamic and dynamic factors) for WCT and the combination T850-SH700 (thermodynamic factors) for THI. The connection is strong, showing that these results can further be used in developing skilful statistical downscaling models (SDMs) to project future changes in the thermal stress indices in Romania (including their extremes) at local scale, producing high resolution information useful in impact studies on human health and other domains. It was found that, for both indices, their increasing trend is mainly explained by the increasing trend of the temperature at 850 hPa covering Romania simultaneously with an increasing trend in zonal circulation (WCT) and an increasing trend in specific humidity at 700 hPa (THI).

 On a decadal/multi-decadal time scale, the WCT/THI variations are modulated by NAO (WCT) and AMO (THI). These results are presented by Dobrinescu et al. (2014) in a peer-reviewed journal journal (under review).

3.3. Connection between the variability of climate extremes in Romania and large-scale blocking.circulation indices

The work associated with this issue was carried out mainly by the UB-FF partner. An additional technique (compared to those presented in 3.2) to justify the increasing trend of the frequency of winter extreme high temperatures over Romania (Frtmax90) was used by analysing the connection between the time series associated to the Frpp90 EOF1 (PC1) and indices associated to various already established large-scale circulation patterns such as: North Atlantic Oscillation (NAO), West Pacific (WP), East Atlantic (EA), Scandinavian patterns (SCA) and blocking activity in the (20°W-70°E) sector.

It was shown that the East Atlantic Oscillation controls a significant part of inter-annual extreme high temperature variability over Romania via advection of warm air from the west. In addition, a strong relationship between blocking activity and frequency of extreme high temperature events in Romania was found. High blocking activity in the (20°W-70°E) sector is related with relatively strong advection of cold air over the country during winter. On the other hand, low blocking activity in the same sector is related with weak advection of relatively cold air in the region. Moreover, the blocking frequency in this sector is modulated mainly by the East Atlantic Oscillation. However, the dominant pattern of Frtmax90 index from Romania also showed weak connection with the NAO, SCA, and WP. This is not so surprisingly, especially in the context of recent findings of other paper, which show that NAO centers of action are influenced by the phase of the EA and SCA patterns. These results are presented by Rimbu et al. (2014) in a recent article published in a peer-reviewed journal. Similar work was carried to explain the mechanism controlling the variability of summer Frpp90 (frequency of very wet days). In this respect, it was found that enhanced blocking activity over the 0°E-40°E sector is associated with high frequency of summer extreme precipitation events due to enhanced synoptic scale activity associated to advection of relatively high potential vorticity from northeast toward Romanian region as well as with enhanced synoptic scale activity in the Mediterranean region. Enhanced blocking activity in the 50°E-70°E sector favors an eastward extension of Atlantic jet and the associated instabilities are related to extreme precipitation events over large parts of central and Eastern Europe. It was argued that these two distinct blocking patterns explain a substantial part of extreme precipitation variability in Romania during summer. This is an additional explanation to those presented above (3.2.1) and published by Busuioc et al. (2014a).These results are presented by Rimbu et al. (2014b) in a peer-reviewed journal (under review).


Other study was carried out to establish a link between heat waves and large-scale mechanisms which are responsible for their occurence. To acomplish this, 2144 cases of heat waves were analysed that were recorded at 105 synoptic stations in Romania over the time interval between 1983 and 2012. To define the heat wave, the territory of Romania was splited in four regions, each with its own climate influenced by the Carpathian Mountains and the Black Sea; for each region the statistic distributions of the maximum temperatures' average and those of the monthly absolute maximum temperatures for the climatologicaly period of 30 years (1961-1990) were computed. By taking into account the 90-percentile, thresholds were established for a heat wave in each region. The threshold for Region 2 is 35°C and 33°C for Region 1, 3 and 4.  By this rule, 2144 heat waves were identified with a duration of at least 3 days. The daily circulation types, which were established with two circulation Catalogues – GWT (with 18 types) and WLK (with 40 types), could be asociated to each day within a heat wave. Then, the frequencies of the circulation types during each heat wave were computed. This way, the main synoptic patterns were spotted that were responsible for the heat waves in Romania. The longest heat wave (18-days) within the entire of 30 years (1983 – 2012) study period was registered at the Oradea synoptic station in 1994, during July, 25th  and August, 11th.

The dominant circulation types, as they were established with the GWT Catalogue are those that are northeastern and Undefined anticyclonic, while those established with the WLK Catalogue are those that are southwestern and northwestern in anticyclonic regime - both in humid and dry air. În conclusion, we can state that the main baric system responsible for the occurence of heat waves in Romania is the ridge of the Azores Anticyclone, which extends over the central Europe close to Romania. The movement of this Anticyclone towards the North of Africa leads to its ridge's extension over southeastern Europe, and to a southwestern circulation in the middle troposphere that favours warm air advection from northern Africa. This ridge eventually envelopes the South of the continent and modifies the southwestern circulation into northwestern. Also significant are the situations in which an anticyclone is centered over Romania and plays a major part in influencing the occurence of heat waves. These results have been published by Barbu et al. (2014b).

3.4. Mechanisms generating short time heavy rainfall using radar information

The main aim of this activity (carried out by CO) was to investigate the dynamic and thermodynamic, both synoptic and mesoscale, configurations associated with severe weather in eastern Romania. Three different sets of data were used in this study. One of the dataset comprises the rainfall accumulations performed every 12 hours at the weather stations located within the Barlad Biver Basin. A second set of data is the Doppler weather radar measurements performed with the S-band radar system located near the basin. Both precipitation and radar data were supplied by the National Meteorological Administration database. The third set of data is represented by reanalysis data: NCEP-DOE Reanalysis II data provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their Web site at Thus, using the empirical orthogonal functions analysis (EOF), simultaneously applied to daily average geopotential field at different pressure levels, a number of severe weather events were clustered into classes. At European scale, the analysis showed the existence of three eigenvectors which describe up to 73% of the data. Using the projections of these vectors on the dataset, eight classes of configurations resulted. For each class the main dynamic characteristics were highlighted. Repeating the empirical orthogonal functions analysis on the same fields, but using reanalysis data at six hours interval, it was revealed that each class has a limited number of eigenvectors which can be used to explain the dynamic similarity between different baroclinic evolutions. The mesoscale feature analysis was performed using Doppler weather radar data. For all the classes, thermodynamic instability and dynamic organization factors increase from south to north in the studied area. The occurrence of certain classes takes place in different times of the convective season, while the most persistent classes are positioned in the middle of the convective season. An important finding is that by projecting the three eigenvectors on the numerical weather prediction models outputs, one can observe the most likely average dynamic behavior to expect in a synoptic time interval. If a tendency stands out, then one can go to smaller time scale (i.e., 6 hours) to check if a certain archive of severe case for the respective class has similar dynamics. The climatological evolution of the projection on the data of each class, for the 1980-2013 interval, shows a change in the baroclinic dynamic configuration associated with heavy rainfall. This dynamic behavior can be associated with the decrease of the amplitude of the baroclinic wave to the eastern and the western parts of the European scale domain, and its amplification in the central part. These results are presented by Carbunaru and Burcea (2014, under review).


WP4. Climate change scenarios

Four activities have been planned in this WP that will be finalised in 2015. The main results achieved until now are presented in the following.

4.1. Development of statistical downscaling models for seasonal climate extreme indices

This activity, carried out by the CO (ANM), refers to the statistical downscaling models (SDMs) mainly based on CCA (canonical correlation analysis). As a novelty, in this project, this type of SDM is applied to the combination of various climate extremes as predictand by using the results presented in section 3.2.1 (Busuioc et al. (2014a). In this way, as a test, the combination between the winter Frpp90, Dmax90 and Dmaxpp0 over a homogeneous region placed in south-western Romania has been considered. As large-scale predictors, the SLP, T850 and SH700 have been used, either alone or in combination. It was found that the SLP is the best predictor for the simultaneous variability of the analysed precipitation extremes, the combination with the T850 gives a lower SDM skill but it is still skillful. The highest skill was obtained for Frpp90 for the entire region and the lowest one for the Dmaxpp90, with a skillful SDM only for some stations.

Another test was carried out for the thermal stress indices (ITU and IR) over the entire country. It was found that for both indices the CCA model is skillful over the entire country. T850 is a good predictor for both indices. The combination T850-SLP (for IR) and T850-SH700 (for ITU) gives a higher skill only for IR. These results have been presented at two scientific conferences (Dobrinescu and Busuioc, 2014a,b).

For the first time in Romania, a SDM based on artificial neural networks from the Multi-Layer Perceptron family was tested for various seasonal temperature and precipitation indices. Even if reasonable results have been achieved for the temperature indices, this type of SDM is mainly recommended to be used for daily/sub-daily time series, due to quite short seasonal time series used for the model calibration (see 4.2).

4.2. Statistical downscaling models at high spatial and temporal resolution

This activity was carried out by the CO with the main objective to develop SDMs for high spatial resolution gridded data set (1km) and a temporal step of 6 hours. The application planned for 2014 is made for the Birlad River Basin (19200 grid points) and will continue in 2015 for the Crisul Alb Basin. Three types of SDMs were developed: conditional stochastic model for generation of precipitation, SDM based on CCA for estimation of temperature and SDM based on artificial neural networks (ANN) for estimation of temperature (at station scale only).

4.2.1. Conditional stochastic model

This model is a combination between the statistical downscaling model based on CCA and a Markov chain model of the first order. In CLIMHYDEX project, an improved version of this model was developed compared to those developed in the FP6 EU ENSEMBLES project: high spatial and temporal resolution; generation of precipitation time series at all grid points in a single model run. The precipitation either occurs or it does not (the two states) and the conditional probability of precipitation occurrence depends only on the occurrence on the previous 6- hours interval. There are two parameters describing the precipitation occurrence process: the transition probability p01, the probability of a wet interval following a dry interval, and p11, the probability of a wet interval following a wet interval. As a wet interval, the case of 6-hour precipitation amount > 0.1 mm is used in this study. The variation of precipitation amount on wet 6-hour interval is described by the gamma distribution which has two parameters: the shape parameter (k) and the scale parameter ( representing the mean precipittaion over the wet 6-hour intervals). Therefore, the conditional stochastic model (CSM) is dependent on the four parameters (p01, p11, k, ). The CSM performance is assessed in two steps: firstly, the performance of the CCA model (expressed as explained variance by reconstructed values from the total observed variance and correlation coefficient between observed and reconstructed values) in estimating the four parameters and secondly, the stochastic model performance in reproducing the observed precipitation statistical parameters:  mean precipitation amount and other extreme indices. The CCA model is developed for each of the four parameters, simultaneous at all grid points (19200), and each season (winter, spring, summer, autumn). Finaly, 16 CSMs have been developed. The CSMs have been calibrated over the interval 1976-2000 and validated over the independent data set 2001-2010. As large-scale predictors, the combination of various dynamic (SLP) and thermodynamic (T850, T500, SH1000, Sh850, SH700, SH500) variables has been considered. It was found a good performance for the CCA models for all the four parameters and all seasons, the highest skill being obtained for p01 and p11 and lowest one for k. The seasonal precipitation amount computed from the 6- hour generated amounts have been compared to similar values derived from observations. It was found a very coherent temporal evolution between the spatial average of the two types of values (simulated and observed) over the validation interval 2001-2010.

This model was used to generate 6 - hour precipitation time series at the 19200 grid points for the two future periods (2021-2050, 2071-2100) using the predictors simulated by the CNCM33 GCM  (ENSEMBLES experiment, stream2, run1) (see 4.4).

4.2.2. SDM based on CCA for temperature

Taking into acount the fact that the time series of 6 - hour temperatures are not statistically independ to be further used in development a linear SDM based on CCA, the model has been developed for each 6- hour interval (1,7,13,19) and two seasons: warm (May-October), cold (Novemver-April). The predictors are 2m temperatures ( NCEP/NCAR reanalysis,2.5 x 2.5 spatial resolution) over the area 20-30E, 40-50N. The model inputs are anomalies projected on their first two EOFs, explaing more than 98% for predictands and 92% for predictors. Finaly, 8 SDMs have been developed and the skill is very high with the explained variance of the estimated values over the independent data set (1991-2010) greater than 83%.

4.2.3. Artificial neural networks

In this phase, several statistical downscaling models based on artificial neural networks from the Multi-Layer Perceptron family were tested. Data series of mean air temperature at 6-hour time step from eight meteorological stations covering the area of the Bârlad river basin (Eastern Romania) were used as predictands. The large-scale predictor is represented by the temperature at 850 hPa, from the domain 20-30E, 40-50N, at 2.5×2.5° spatial resolution extracted from NCEP/NCAR

The computations were made with the Statistica Automated Neural Networks (SANN) module of Statistica v. 8, developed by StatSoft. The module includes a tool that can automatically evaluate a large number of different neural network architectures of varying complexities, and select the best set of specific architectures for a given problem. Several network types, network sizes and activation functions were tested in order to select the appropriate variables and then optimizing the network architecture by heuristic search. The best results were achieved with a MLP network with 25 predictors and 10 hidden layers, although models with 5-8 hidden layers provided comparable results. A simple two-fold cross-validation scheme was used for evaluating the performance of each model, where the data was partitioned into contiguous sets for the periods 1961-1990 (used for training) and 1991-2010 (used for validation). The model performed very well for six out of eight stations (with linear correlation coefficients above 0.97), and provided good results for the other two. Later, model was applied over the entire country, in order to obtain a sub-daily gridded dataset for the Bârlad river basin.

4.3. RCM /GCM validation

In the reported period, the validation of three RCMs (RACMO22E, HIRHAM5, REMO2009) of the EURO-CORDEX experiment, with respect to their capability to reproduce the characteristics of various precipitation extremes, has been carried out. This activity will be finalysed in 2015. Three precipitation indices have been considered: annual frequency of daily amount greater than 10mm; annual maximum 5-day precipitation; precipitation fraction of annual total precipitation due to daily precipitation > 95th percentile. The results show that the three RCMs overestimate long term average of these indices (1971-2000).

4.4. Climate change scenarios at various spatial and temporal scale

The SDMs developed within the activities 4.1 and 4.2 have been used to project the future changes (2021-2050, 2071-2100) of temperature and precipitation indices at seasonal or sub-daily (6 hours) scale.

Firstly, the SDMs presented at 4.2.1 and 4.2.2 (CSM and CCA SDM) have been used to achieve changes in gridded 6-hour temperature and precipitation over the Birlad Basin using the predictors simulated by the CNCM33 GCM (ENSEMBLES experiment, stream2, run1). It was found that over the 2021-20150 period, it is expected a temperature increase between 0.90C -1.30C, with the higher increase in the middle of the day in warm season compared to middle of the night. For the 2071-2100 period, the increase is higher ranging between 2.00C and 2.8 0C with a higher gradient between the middle of the day and middle of the night. For precipitation, the preliminary results show a decrease of seasonal amount of about 30 mm/season for summer and autumn

Secondly, preliminary results for the projections of some seasonal extreme indices have been achieved. It is the case of stress indices (ITU and IR). For the period 2021-2050, it was found for both indices a higher increase over the eastern part of Romania, with higher values for IR (1.16-1.50 units).

Some preliminary results have also been obtained using the direct RCM outputs (CMIP5 experiment) under various emission scenarios. In this way, a decrease of the PDSI drought index over the Birlad River Basin was found, highest under the RCP 8.5 scenario. This result is in agreement with changes found in other extreme precipitation indices over Romania such as:  number of wet days, very heavy rain and continuous wet days. The conclusions will be finalized in 2015.


WP5. Climate change impact on hydrology


The activities planned for this WP are carried out by the P1 (INHGA).The results achieved until now for each activity are summarised in the following.


5.1. Extrapolation of spatial characteristics for extreme hydrological events


In addition to climatic factors, the maximum flow is influenced as well by the altitude of the drained area, the surface and the shape of the catchment, moisture and permeability of the soil, land use, forest cover, influence of human activities etc. It was identified a relationship that highlights more accurately the formation of absolute maximum monthly discharge, and which depends on the catchment area, the medium altitude of the basin and lithological class. Specific absolute maximum monthly flow was also analysed. It varies between 8 and 48 l/s/km2. This parameter shows an extremely high heterogeneity in Bârlad river basin unlike other basins in Romania. The analysis revealed Tecucel basin and Vaslui basin at Satu Nou gauging station, where the discharges, especially the specific flow, are much higher in relation to physical and geographical conditions.

For Ciurea-Tinoasa representative basin located in the northern part of Bârlad plateau with an area of 4.02 km2 the SWAT model was calibrated in order to identify at detailed scale the complex mechanisms that control the variability of hydrological extremes. The calibration period was 1992-2010. Since both the maximum and monthly flows show a significant dependence to the time of concentration, a methodology has been developed to estimate this parameter on GRID data. This information allows the estimation of extreme discharges values for small or unmonitored rivers.

The hydrological modelling is applied to investigate the hydrological responses to land cover changes. Based on the current deciduous forest coverage (77%) and analyzing scientific and practical needs of the addressed problem, we established two new work scenarios: 42% reduction of forest cover and complete reduction by replacing with secondary grassland.

5.2. Identification and assessment of the spatial and temporal characteristics of flood and drought events in the selected basins


To highlight the dry periods from a hydrological point of view the procedure for calculating the SFI to the monthly discharge time series of three hydrometric stations (Vaslui, Barlad and Tecuci) situated on the Barlad River was applied. The results showed that the highest variability between SFI and SPI is identified for the summer season, where the response time between the meteorological and hydrological drought at different time scales is up to 3 months. On the annual time scale, the response time is at least 6 months on the Barlad River.

Based on the temporal evolution of the SFI6 index, the hydrological drought periods are clearly highlighted: moderate (-1.5 ÷ -1), severe (-2 ÷ -1.5) and extreme (≤ -2), but also periods of time with high discharges. In this way, the periods characterized by the hydrological drought have been recorded for the years: 1963 - 1964, 1968 - 1969, 1987, 1990, 1994 - 1995, 2001, 2007 and 2009. Also, it is shown that starting 1986, SFI6 presents an increased variability of dry periods.

The results were presented at the Annual Scientific Conference INHGA 2013. It was also published an article in “Hidrotehnica Review” (Chelcea & Adler, 2013).

5.3. Analysis of variability of hydrological characteristics under the influence of climate change


To achieve this objective the extreme values theory (EVT) was applied to average monthly flow values series from Gurahonţ hydrometric station on Crisul Alb River. The following results were obtained: EVT analysis (theory of extreme values) showed that the discharge values between 130 and 470 m3/s have the return periods between 5 and 100 years for the winter and spring months. For summer and autumn months, discharges reaching up to 1300 m3/s have the return periods between 5 and 1000 years. The results obtained in this work were presented at the Annual Scientific Conference of INHGA (2013) and published in two articles ISI and BDI respectively (Ionită et al., 2013; Chelcea & Ionita, 2013).

5.4. Analysis of groundwater resources and exchange between surface water and groundwater


Analysis of geological and hydrogeological conditions of the groundwater in the Crisul Alb basin was based on:

• studying of the geological and hydrogeological maps, inventory data on lithology and hydrogeological characteristics;

• fieldwork - hydrostatic level measurements campaign in 18 hydrogeological wells and 48 domestic fountains, located between Vărşand and Gurahonţ.

The following activities were carried out:

 - 13 lithological columns of representative wells of national hydrogeological network;

 - 6 hydrogeological cross-sections through the Vărşand, Cil, Zărand, Ineu, Bocsig, Chişineu Cris alignments;

 - Piezometric map of the shallow aquifer;

 - Data concerning the evolution of hydrostatic levels in the wells of National Hydrogeological Network.

Generally, there was a downward trend in average multiannual hydrostatic levels and a clear trend of increasing hydrostatic levels in the first 4 months, then fall until September, followed by an increase until the end for most wells analyzed. Corroborating these data and analyzing the relationship between surface water and groundwater, it is found that, in most part of the study area, Crisul Alb drains the shallow aquifer. The exception is only a relatively small sector, upstream of Ineu, where the river recharges the shallow aquifer.

5.5. Hydrogeological model calibration and quantification of climatic factors on groundwater


The conceptual model of a hydro-structure represents the support for both quantitative assessment of groundwater flow and water resources management. In this case conceptual model of the groundwater along the Crisul Alb River is important. Based on this conceptual model we will be achieved also, the quantification of climatic factors on groundwater using MIKE SHE model.

In this activity has been elaborated an article, which has been accepted for presentation at the conference "The air and the water components of the environment" organized by Babes - Bolyai University, Cluj Napoca (20 to 22 March, 2015). The paper (Radu et al., 2014) is going to be published into the Conference Proceedings, an indexed BDI volume.

5.6. Model with distributed parameters (NOAH) to quantify the impact of climate change on extreme flow in selected representative catchments and Barlad river basins


NOAH is a hydrological model with distributed parameters that can be used for detailed simulation of hydrological processes in catchments. This model is composed of several sub-models that simulate the energy and water balance in the soil, evapotranspiration, snow cover evolution, surface hypodermic and basic flow, runoff propagation in river beds, etc. In order to assess the impact of climate changes on the hydrological regime in Barlad, river basin, by this detailed hydrological modelling, the NOAH model calibration was necessary.

Historical meteorological data (rainfall and temperature at fine and very fine resolution, obtained in the frame of the CLIMHYDEX Project and other meteorological parameters, taken from "WATCH-forcing-Data-ERA-Interim” data sets) were used for NOAH model calibration. For each of these types of data, a conversion program, in the format needed as model input, was developed. In addition to the Tinoasa-Ciurea representative basin, the following catchments were selected: Barlad River at Bacesti station, Tutova River at Rădeni station and Berheci River at Feldioara station, also considered as representative for Barlad river basin. After the calibration process, optimal values of the parameters were obtained, which were further used for hydrological processes simulation in the selected representative catchments.

To estimate the impact of climate variability and changes on extreme discharges Barlad river basin, two simulations were performed for the following future periods 2021-2050 and 2071-2100. The results were presented at two conferences and have developed two articles, one already published in Die Bodenkultur - Journal for Land Management, Food and Environment (Matreata et al, 2013), and another which will be published in the Proceedings (CD) of INHGA Annual Scientific Conference (Mătreaţă & Mătreaţă, 2014).


5.7. Model with concentrated parameters (CONSUL) to quantify the impact of climate change on extreme discharges in Barlad river basin


The hydrological model CONSUL is a conceptual model based on semi-distributed parameters depending on the physical characteristics of the river basin. According to the schematic representation (physiographic modelling) of how water flows and collects in a river basin, the model computes the discharge hydrographs on sub-basins and then performs their routing and composition on the main river and tributaries. CONSUL model was used for the simulation of the largest floods in the Bârlad river basin from the period 1975-2010. Continuous flow simulation from the period 1984 - 1988 was also used to calibrate the hydrological model parameters and from the years 2006 and 2010 for validation. Calculation of average precipitation and air temperature (hydrological model input data) for each sub-basin was performed using a pre-processing program of meteorological data from original rectangular grid nodes corresponding to Bârlad river basin, averaging being achieved as weighted values based on the representativeness of these nodes for each analyzed sub-basin. Calibration of model parameters was performed in two stages: (i) individual (based on the 25 rainfall-runoff events) and (ii) globally (by continuous flow simulation on considered calibration period). The flow simulation results with the CONSUL model in the Bârlad river basin, during both the calibration and validation periods, showed that the model gives the best results, in particular in the case of floods generated by precipitation evenly distributed in space.

In order to estimate the impact of climate variability and changes on extreme discharges of Barlad river basin, two long term simulations, at step 6 hours, were performed, using as input data series of precipitation and temperature from simulations of climate evolution.

The results of this work were presented at the EGU General Assembly 2014 Vienna (Mic et al., 2014a). Also, two articles were developed: Corbus et al., (2014) (CD, INHGA 2014, Annual Scientific Conference); Mic, et al, (2014, under review, ISI journal).


5.8. Comparative analysis of both affluent and defluente solid flow in Tungujei reservoir, as result of hydro-climatic regime variation.

The analysis of suspended solid and liquid flow transited through Tungujei reservoir during the period 1988 - 2012 shows that the solid flow is determined either by the type of dominant hydrological regime in the area or the operating mode of accumulation. It was observed that in the rainy periods extended on longer intervals of time (> 20 hours) and having a high intensity (> 80 l/m2) the alluvial material dislocation occurs on slopes, which is discharged into the river basin with increasing of water turbidity including coarser sediments transport downstream.

Comparison of the sediment contribution in Tungujei Reservoir with the diffluent flow shows a diminished alluvial in downstream in almost all years analyzed (1988-2013). But there are years when the diffluent flow is higher than the solid volumetric contribution in reservoir, reaching to 340% (2007).

To get a more accurate knowledge of alluvial volumes into Tungujei reservoir, it was considered useful to include in the calculation of solid balance, the second component of alluvial flow - solid flow dragged.

The results show that the evolution of deposits obtained by solid balance in the period 1988 - 2011 (557,000 m3) is reasonably close to the value determined topo-bathymetric in the basin (374,000 m3). The difference of about 180,000 m3, resulting in addition from the solid balance calculation in 2011, is due to either the informative data included in the calculation of solid flow or the processing way in situ of the topo-bathymetric data.

In terms of silting-up rate, evaluated differently for the periods 1988 - 1999 (29,989 m3 / year) and 2000-2013 (14476 m3 / year), the result shows very clearly the impact of hydroclimatic deficit, reported especially in the last decade. In the next steep (2015) the future changes of silting-up rate in Tungujei reservoir will be estimated using the climate changes scenarios for the periods 2021 – 2050.


Project Director,

Dr. Aristita Busuioc




ANNEX 1. Published papers under the CLIMHYDEX framework (including «under review» ones)

A1. International peer-reviewed journals

1. Barbu N., Cuculeanu V., Stefan S., 2014a: Modeling the precipitation amounts dynamics for different time scales in Romania using multiple regression approach. Romanian Journal of Physics vol.59, Nos. 9-10- 1127-1149 (

2. Barbu N., Georgescu F., Stefanescu V.E., Stefan S., 2014b: Large- scale mechanisms responsible to heat waves occurrence in Romania. Romanian Journal of Physics vol. 59, Nos. 9-10, 1109-1126,

3. Birsan M-V., Dumitrescu A., 2014. Snow variability in Romania in connection to large-scale atmospheric circulation. Int. J. Climatol. DOI: 10.1002/joc.3671 (open access)

4. Busuioc A., Dobrinescu A., Birsan, M-V., Dumitrescu, A., Orzan, A., 2014a: Spatial and temporal variability of climate extremes in Romania and associated large-scale mechanisms. Int. J. Climatol. DOI:10.1002/joc.4054 (open access).

5. Busuioc, A. Birsan M-V, Carbunaru D., Baciu M. and  Orzan, A., 2014b; Changes in the thermodynamic instability on the North-Atlantic European scale and connection with changes in regional rain showers. Int. J. Climatol.  (under review)

6. Carbunaru, D, Burcea, S., 2014:Thermodynamic configurations associated with heavy rainfall in Eastern Romania, . Int. J. Climatol. (under review)

7. Cheval S., Busuioc A., Dumitrescu A., Birsan, M-V, 2014: Spatiotemporal variability of the meteorological drought in Romania using the Standardized Precipitation Index (SPI). Climate Research vol 60:235-348.  DOI: 10.3354/cr01245

8. Dobrinescu, A., Busuioc, A., Birsan M-V., Dumitrescu, A. and Orzan, A., 2014: Changes in thermal discomfort indices in Romania and responsible large-scale mechanisms. Climatic Research (under review)

9. Dumitrescu A, Bojariu R, Birsan MV, Marin L, Manea A, 2014a: Recent climatic changes in Romania from observational data (1961-2013). Theoretical and Applied Climatology. DOI: 10.1007/s00704-014-1290-0

10. Dumitrescu A., Birsan, M, Manea, A., 2014b.  Spatio-temporal interpolation of the hourly precipitation values over Romania for 1975-2010,  Int. J. Climatol. (under review).

11. Ionita, M., N. Rimbu, S. Chelcea and S. Patrut, 2013: Multidecadal variability of summer temperature over Romania and its relation with Atlantic Multidecadal Oscillation, Theor. Appl. Climatol., 113(1-2): 305-315, doi: 10.1007/s00704-012-0786-8

12. Ionita, M., Chelcea, S., Rimbu, N., Adler, M-J., 2014: Spatial and temporal variability of winter streamflow over Romania and its relationship to large-scale atmospheric circulation, Journal of Hydrology, doi:

13. Mares, C., Adler, M.J., Mares, I., Chelcea, S., Branescu, E., 2014: Discharge variability in Romania using Palmer indices and a simple atmospheric index of large scale circulation, Hydrological Sciences Journal - Manuscript ID HSJ-2014-0349, (accepted)

14. Mic, R.P., Corbuş, C., Mătreaţă, M., 2014: Flow Simulation in Bârlad River Basin Using Romanian Hydrological Model CONSUL, Romanian Reports in Physics (under review).

15. Rimbu, N., Stefan, S. , Necula, C. ,2014a: The variability of winter high temperature extremes in Romania and its relationship with large-scale atmospheric circulation. Theoretical and Applied Climatology, DOI:10.1007/s00704-014-1219-7

16. Rimbu, N., S. Stefan, A. Busuioc , F. Georgescu, 2014b: Blocking circulation and precipitation extreme in Romania during summer. Int. J. Climatol l (under review).

17. Stefanescu, V., Stefan, S., Georgescu, F., 2014: Spatial distribution of heavy precipitation in Romania between 1980 and 2009. Meteorological Application 21:684-694. DOI:10.1002/met.1391

A2. Books/chapters

1. Birsan MV, 2013: Statistical downscaling of sub-daily temperature by means of artificial neural networks. In V Barsan (Ed) Hands-on science, innovative teaching, gates to research. Știința, Chișinău. ISBN 978-9975-67-873-5. 

2. Silvia Chelcea, Monica Ionita, Mary-Jeanne Adler, 2014: Identification of dry periods in the Dobrogea region, în Carmen Maftei (Ed),  Extreme Weather and Impacts of Climate Change on Water Resources in the Dobrogea Region,  IGI Global (în curs de apariție)

3. Monica Ionita, Silvia Chelcea, 2014: Spatial and temporal variability of seasonal drought over Dobrogea region and its relationship to large-scale atmospheric circulation and North Atlantic Ocean Sea Surface Temperature, în Carmen Maftei (Ed),  Extreme Weather and Impacts of Climate Change on Water Resources in the Dobrogea Region,  IGI Global (în curs de apariție)


A3. BDI publications

1. Borcan, M., Neculau, G., Retegan, M., 2014: The influence determined by the changing of mark climatic parameters upon the hydrological regime in southern Romania,14th GeoConference on Water Resources. Forest, Marine and Ocean Ecosystems”, 17-26 iunie 2014, Albena, Bulgaria, în Proc. I, (BDI) pp. 759 – 766, DOI: 10.5593/SGEM2014/B31/S12.097

2. Chelcea, S., Ionita, M., 2013: Extreme value analysis of the Barlad river time series, Ovidius” University Annals of Constanta, Year XV- Issue 15 - Series CIVIL ENGINEERING journal (BDI), ISSN 1584-5990, pp 273-280

3. Matreata, S., Baciu, O., Apostu D., Matreata M., 2013: Evaluation of the Romanian flash flood forecasting system – case study in the Calnau river basin. Die Bodenkultur - Journal for Land Management, Food and Environment. 64. Band/Heft 3-4/ISSN 006-5471, pp67-72.


4. Retegan, M., Borcan, M., 2014: Assessment of the potential impact of climate change upon surface water resources in the Ialomița River Basin from Romania, „14th GeoConference on Water Resources. Forest, Marine and Ocean Ecosystems”, 17-26 iunie 2014, Albena, Bulgaria, în Proc., I, (BDI) pp. 89 – 96.
DOI: 10.5593/SGEM2014/B31/S12.012

A4. Others publications

1. Adler, M.J., 2013. Climate Change and Its Impact in Water Resources, Belgrade AIHS Volume, Conference Proceedings of Climate Variability and Change – Hydrological Impacts, Belgrade, October 2013.

2. Adler, M.J., Corbus, C., Mic, R., Chelcea, S., 2013. Climate Change and Its Impact in extremes, Proceedings of the Annual Conference of the University of Bucharest, Phaculty of Physics, 21 June, 2013.

3. Adler M-J, Chelcea, S., 2014: Climate change and its impact in water resources in Romania, Proceedings of XXVI Conference of the Danubian Countries on Hydrological Forecasting and Hydrological Bases of Water Management, 22-24 September 2014, Deggendorf, Germany

4. Chelcea, S., Adler, M.J., 2013: Legătura dintre seceta meteorologică şi seceta hidrologică pe râul Bârlad, revista Hidrotehnica, vol. 58, 4-5, pag. 22, ISSN 0439-0962.

5. Corbuş, C., Mic, R.P., Mătreaţă, M., 2014: Calibrarea parametrilor modelului hidrologic CONSUL prin simularea scurgerii în bazinul hidrografic Bârlad, CD cu Lucrările Conferinţei Știinţifice Anuale a INHGA 2014, (în curs de publicare)

6. Mătreaţă, M., Mătreaţă, S., 2014: Calibration particularities for the NOAH distributed hydrological model parameters. Application for the Bârlad river basin, CD cu Lucrările Conferinţei Știinţifice Anuale a INHGA 2014, (în curs de publicare)


B. PhD thesis


1. Dumitrescu Alexandru: Spatializarea parametrilor meteorologici și climatici prin tehnici SIG. Sustinuta public in data de 19.09.2012, Facultatea de Geografie, Universitatea Bucuresti.

2. Victor Stefanescu:  Elaborarea unui sistem informational pentru reducerea impactului fenomenelor meteorologice severe. Sustinuta public in data de 20.09.2012, Facultatea de Fizica, Universitatea Bucuresti.