In this paper we present preliminary results of an alternate approach to the conventional top-down assessment of climate change impact on water resources driven by climate models. A robust transfer function linking atmospheric circulation indices and surface climate is used to predict mean areal monthly precipitation and air temperature estimates at a multi-centennial time-scale (1659–2100), including 13 CMIP5 climate models and two RCP scenarios. By integrating “paleoclimate” reconstructions in the current knowledge of likelihood of climate change, climate hazard assessment is treated as a large number of plausible climate changes instead of being solely expressed as individual CMIP5 projections. Catchment sensitivity to climate change is regarded as changes in peak flows and low flows in a sub-catchment of the transnational Meuse catchment. By combining likelihood of climate change with the knowledge of the sensitivity of a given catchment it is possible to develop a more robust decision-making in water management.

As a part of the effort to scientifically inform the development of the adaptation strategy for the Meuse basin, we detail hereafter the generation of integrated climate and hydrological scenarios for the whole basin. We also present the setup of a first coordinated hydraulic modelling from spring to mouth of the river Meuse. The latter has enabled to compute the range of change in inundation hazard under the "wet" transnational hydrological scenario for the time slices 2021-2050 and 2071-2100. A significantly higher impact of climate change has been found in the middle part of the Meuse basin, compared to the upper and the lower parts. These conclusions have been further confirmed by a refined analysis conducted for a 100 km-long stretch of the river Meuse crossing the Belgian-Dutch border.

Since the mid 1970s, the number of days with westerly atmospheric circulations has strongly increased during winter months. As a consequence, rainfall totals, rainfall event duration and intensity have been subject to significant positive trends throughout the Mosel river basin. However, the trends identified through the non-parametrical test named Kendall's tau have shown to be spatially varying. The intensity of the trends appears to be directly linked to orographic obstacles that are well known to have a strong influence on average rainfall totals. A direct consequence of the changes having affected winter rainfall under westerly atmospheric circulations on the one hand and the spatial variability of these changes on the other hand, is a spatially varying positive trend in maximum winter streamflow. Thus, even though a clear large-scale change has affected winter rainfall over the past decades, its intensity is either strongly moderated or enhanced by orographic obstacles. The related changes in streamflow are directly dependent on the spatial variability of the changed rainfall characteristics.

Trends of monthly air temperature extremes were investigated in five meteorological stations of the Grand-Duchy of Luxembourg during the period 1949–1998. The application of an innovative homogenization method based on the concept of relative homogeneity to climatic time series allows identifying multiple break points, as well as correcting data series in an objective and robust statistical way. The rise of maximum temperature (Tmax) has occurred at a rate of 1.5 times that of the minimum temperature (Tmin) in winter (+1.4 °C versus +0.9 °C) and summer (+1.4 °C versus +0.8 °C). No trend in temperature extremes was found in autumn, while spring was affected by a small warming (+0.3 °C) of Tmin and no change in Tmax resulting in a decrease of the diurnal temperature range (DTR) (−0.3 °C). In spring, a strong positive linear relationship between Tmin warming and local terrain slope could be found. Comparison to new-gridded large-scale climatologies indicates generally close agreement to temperature trends during the 1949–1998 period, while a lower local warming was observed in summer during the post-1975 period following the changing-point year of atmospheric circulation over North-western Europe. This study shows that the question of data homogeneity is not trivial and should receive careful attention before quantifying historical temperature trends and identifying their spatial patterns at regional scale.

Hydro-meteorological data of high spatio-temporal resolution have been exploited since the mid1990s for environmental research in the Grand Duchy of Luxembourg. Examples of ongoing field observations are given in this paper, with a special emphasis on the large hydro-meteorological event of January 2003, which generated severe inundations in the floodplains of the main Luxembourgish tributaries (Alzette and Sure rivers) of the upper Mosel river. The large rainfall-runoff event of January 2003 was well documented through a single hydro-meteorological database collected via dense rainand stream-gauge networks, set up by three institutions of the Grand-Duchy of Luxembourg (Public Research Center-Gabriel Lippmann, Ministry of Interior and Ministry of Agriculture). Different maps, derived from ground-based measurements of the January 2003 flood, illustrate the relationship between the spatio-temporal distribution of rainfall intensities, runoff contributing areas, as well as the propagation of flood waves in the channel network. Of particular interest is a better estimation of flood peaks related to heavy rainfall intensities, as well as an enhanced understanding of the influence of the geological substrate on the rainfall-runoff relationship with high antecedent saturated conditions. However, monitoring streamflow during such a large event, remains a difficult task owing to the uncertainty related to the rating-curves for high water stages.

High spatio-temporal resolution monitoring has only been progressively developed in the Rhine-Meuse basins over the last few decades. As a consequence, basic hydrological information can be very scarce in some areas. In regions which are homogeneous from a hydroclimatological and physiogeographical point of view, hydrographs can be reproduced via regionalized hydrological models, provided that climatological observation series are available.The Alzette river basin, monitored since the mid-1990s by a very dense hydroclimatological observation network, had been chosen in the framework of the IRMA-SPONGE project FRHYMAP for transposing the conceptual hydrological models HRM and SOCONT and regionalizing their parameters. The regionalized models were to be used both for extending the currently available runoff series and evaluating runoff in neighbouring non-monitored basins.The 16 monitored sub-basins of the Alzette, reflecting the physiogeographical diversity of the study area, were divided into two subsets, serving for both the calibration and the validation procedures. Once the transposition of the models to the Alzette basin had been successfully assessed, their parameters were linked to the physiogeographical characteristics of the sub-basins. The performance of the thus regionalized models was assessed via a validation on a subset of basins that had not been retained for the elaboration of the regional parameter sets.The transposition of the HRM and SOCONT model to the Alzette river basin was completed successfully. Results overall proved to be satisfying, with the FIRM model performing equally well for low flows and high flows, while the SOCONT model showed best results for high flows and a systematic overestimation of the mean discharge. Both models proved to be adequate for evaluating daily runoff in non-monitored basins of the Grand-Duchy of Luxembourg, helping thus to counterbalance the considerable lack of hydrological observation series in this part of the Rhine basin. Copyright (C) 2004 John Wiley Sons, Ltd.

A continuous rainfall-runoff simulation was performed to assess the potential effect of climate changes on the streamflow regimes and water resources of tributaries of the Alzette river basin extending over 1176 km2, mainly in the Grand Duchy of Luxembourg, characterized by various hydrological patterns. Global climate change scenarios for the 2050 horizon, based on GCM projections from the KNMI and UKHI synoptic runs, were disaggregated into mesoscale daily PET and rainfall series. Seasonal expected PET changes were proportionally applied to present daily values, whereas future hyetographs were empirically constructed according to observed trends in rainfall time series for the study area. The various ways of applying the mesoscale rainfall scenarios exert a significant influence on the magnitude and spatial distribution of streamflow responses. The comparison of future and present hydrographs also shows that the impact of mesoscale climate change is extremely variable with regard to the considered hydrological variable. The spatial variability of streamflow responses is largely conditioned by climatic and physiographical characteristics of the sub-basins. The winter period is most affected by altered climate conditions and some sub-regions appear to be particularly sensitive in terms of changes in low or high flows.

The role of groundwater resurgence on runoff coefficients was investigated in the Alzette river floodplain, upstream of Luxembourg-city (Grand-Duchy of Luxembourg). A threshold in the saturation level was identified through the calculation of a water balance at daily time step. This threshold appeared to coincide with groundwater resurgence in the Alzette river floodplain. Once groundwater resurgence appears, runoff coefficients reach their maximum level of approximately 70%, regardless of individual rainfall event totals. Groundwater level monitoring can thus serve as a tool for evaluating the overall saturation level, as well as the runoff coefficients that are to be expected in case of rainfall. The water balance and groundwater levels having been identified as indicators of the overall soil moisture in the basin prior to a flood, these variables were used to develop an empirical tool for the estimation of peak discharge in the Alzette river basin for rainfall events of a given return period.

The knowledge of rainfall patterns is a key issue for regionalization in hydroclimatic studies. In mountainous areas, the sparsity of the measurement network, and the complexity of relationships between rainfall and topography make an accurate and reliable spatialization of rainfall amounts at the regional scale difficult. The purpose of this paper is to present an objective, analytical and automatic model of quantification and mapping of orographic rainfall applied to the north-eastern part of France but also applicable in other complex tot-rain. PLUVIA distributes point measurements of monthly, annual and climatological rainfall to regularly spaced grid cells through a multiple regression analysis of rainfall versus morpho-topographic parameters derived from a digital elevation model. The use of an omnidirectional parameterization of the topography induced by a windowing technique allows better account to be taken of the synoptic-scale weather systems generating the different rainfall quantities of interest and the spatial scale of orographic effects. It also provides a more physical interpretation of geographical and topographical parameters selected for spatial estimation. The application relics on a network of more than 150 rain gauges spread over 30 000 km(2) and concerns monthly to several yearly amounts of a sequence of 20 years. Advantages and limitations of the PLUVIA system are compared with those of two commonly used methods of multi-variate geostatistics: kriging with external drift and extended collocated co-kriging. Copyright (C) 2002 Royal Meteorological Society.

This paper presents a regionalization methodology and an original representation of the downstream variation of daily streamflow using a conceptual rainfall–runoff model (HRM) and the 3D visualization tools of the GIS ArcView. The regionalization of the parameters of the HRM model was obtained by fitting simultaneously the runoff series from five sub-basins of the Alzette river basin (Grand-Duchy of Luxembourg) according to the permeability of geological formations. After validating the transposability of the regional parameter values on five test basins, streamflow series were simulated with the model at ungauged sites in one medium size geologically contrasted test basin and interpolated assuming a linear increase of streamflow between modelling points. 3D spatio-temporal cartography of mean annual and high raw and specific discharges are illustrated. During a severe flooding, the propagation of the flood waves in the different parts of the stream network shows an important contribution of sub-basins lying on impervious geological formations (direct runoff) compared with those including permeable geological formations which have a more contrasted hydrological response. The effect of spatial variability of rainfall is clearly perceptible.

The Hydrological Recursive Model (HRM), a conceptual rainfall-runoff model, was applied for local and regional simulation of hourly discharges in the transnational Alzette River basin (Luxembourg-France-Beigium). The model was calibrated for a range of various sub-basins with a view to analysing its ability to reproduce the variability of basin responses during flood generation. The regionalization of the model parameters was obtained by fitting simultaneously the runoff series of calibration sub-basins after their spatial discretization in lithological contrasting isochronal zones. The runoff simulations of the model agreed well with the recorded runoff series. Significant correlations with some basin characteristics and, noticeably, the permeability of geological formations, could be found for two of the four free model parameters. The goodness of fit for runoff predictions using the derived regional parameter set was generally satisfactory, particularly for the statistical characteristics of streamflow. A more physically-based modelling approach, or at least an explicit treatment of quick surface runoff, is expected to give better results for high peak discharge.

Good quality hydrological models are available to establish the water balance over large regions. A precise spatial representation of the entries of the water cycle such as the rainfalls, is then a crucial condition to obtain reliable and accurate results. However, it is not always easy to get a representative rainfall map, particularly in rugged terrain where the spatial structure of the rainfall is complex.Two geostatistical methods that mix rainfall and topographical data when interpolating rainfall are applied: Kriging with External Drift and Collocated Cokriging. They are compared to a third method based on a multi-linear regression from a directional parameterisation of the topography. Particular attention is paid to two points: the role played by the pre-processing of the Digital Elevation Model, particularly its smoothing.The statistical analysis of the results obtained from several series of rainfall measurements (monthly to multi-annual data) allows us to draw some preliminary conclusions in the scope of determining optimal and automatic procedures for calculating rainfall fields.

Hydro-meteorological data of high spatio-temporal resolution have been exploited since the mid- 1990s for environmental research in the Grand Duchy of Luxembourg. Examples of ongoing field observations are given in this paper, with a special emphasis on the large hydro-meteorological event of January 2003, which generated severe inundations in the floodplains of the main Luxembourgish tributaries (Alzette and Sûre rivers) of the upper Mosel river. The large rainfall-runoff event of January 2003 was well documented through a single hydro-meteorological database collected via dense rain- and stream-gauge networks, set up by three institutions of the Grand-Duchy of Luxembourg (Public Research Center-Gabriel Lippmann, Ministry of Interior and Ministry of Agriculture). Different maps, derived from ground-based measurements of the January 2003 flood, illustrate the relationship between the spatio-temporal distribution of rainfall intensities, runoff contributing areas, as well as the propagation of flood waves in the channel network. Of particular interest is a better estimation of flood peaks related to heavy rainfall intensities, as well as an enhanced understanding of the influence of the geological substrate on the rainfall-runoff relationship with high antecedent saturated conditions. However, monitoring streamflow during such a large event, remains a difficult task owing to the uncertainty related to the rating-curves for high water stages.

Owing to the non-linearity of the rainfall- infiltration-runoff relationship, soil water content in the river basin represents a key parameter to be monitored for flood management purposes. Remote sensing observations can be used in hydrologic models as a source of time varying hydrologic state data that allows constraining model predictions. The analysis of a series of ERS-1 SAR images showed that the mean backscattering coefficient of selected soil parcels is strongly correlated with a ground-based wetness index the so-called soil saturation index (SSI). This paper shows that SSI values obtained via remote sensing can be used to update the internal saturation states of rainfall-runoff models through the sequential assimilation of the soil moisture information. The assimilation procedure is based on an extended Kalman filter as both simulated and observed saturation states are prone to errors. The magnitude of the correction thus depends on the ratio of errors on the observations and the model. Further research is needed to reduce the uncertainties that remain over the reliability of SAR to provide soil moisture information with a sufficient level of accuracy.