Projected changes in future reference evapotranspiration for Luxembourg derived from a very high resolution regional climate model

Abstract:

1. Introduction
The assessment of climate change impacts on the terrestrial water cycle is still one of the major challenges in hydrological modelling approaches (Bormann, 2011). In its 5th Assessment Report, the Intergovernmental Panel on Climate Change (IPCC) has recently identified "harm or economic losses from inland flooding" as one of the eight major climate risks (IPCC, 2014; Kintisch, 2014). Potential climate change effects for Luxembourg - derived from a multi model ensemble of Regional Climate Models (RCM) - and their effects on vegetation were recently assessed by Goergen et al. (2013). Air temperatures were projected to rise by 1.2 K for the near (2041-2050) and 3.6 K for the far future (2091-2100). Additionally, changes in seasonal precipitation patterns are expected.
According to Prudhomme and Williamson (2013) future changes in evapotranspiration will be equally important as changes in precipitation patterns for determining changes in river flows. Evapotranspiration is the combination of soil evaporation and vegetation transpiration and is affected by meteorological variables, crop characteristics and management, as well as environmental aspects. The evapotranspiration rate from a reference surface is called reference evapotranspiration (ETo) and was introduced to study separately the evaporative demand of the atmosphere, independently of crop type, crop development and management practices (Allen et al., 1998).
There are many possibilities to calculate the potential evapotranspiration, ranging from simple temperature based approaches to the complex Penman equation (Penman, 1948). Although some studies demonstrated that temperature based methods show in some cases comparable or even better results than more complex approaches (Kannan et al., 2007; Wehner et al., 2011) this cannot be generalized for future climate conditions.
We used the ETo Calculator from the Food and Agriculture Organization (FAO) of the United Nations to calculate the reference evapotranspiration based on measurements of the official WMO reference station in Luxembourg and on climate projections of a RCM.
The objectives of our study are a) the assessment of the present day reference evapotranspiration rates in Luxembourg based on measurements and b) analyze the influence of projected climate change conditions on the precipitation and evapotranspiration conditions for two future time-spans.

2. Materials and methods
There are numerous methods based on different concepts to calculate the evapotranspiration. They can be classified based on the meteorological variables used for the calculation. Basic approaches only rely on air temperature (e.g. Thornthwaite, 1948), or radiation (e.g. Turc, 1961); more advanced models use a combination of these parameters Penman (e.g. 1948). We decided to use the reference crop evapotranspiration recommended by the FAO, based on the Penman-Monteith method (Allen et al., 1998).
In our study, we used different data sets of direct measurements, as well as the outputs of a RCM. Since no official national weather service is operated in Luxembourg, long-term observational data sets are rather scarce. The only official WMO station (WMO station ID = 06590) is located at the Findel airport, southeast of the City of Luxembourg (Junk et al., 2014). Measurements of daily mean, minimum and maximum air temperature, mean relative humidity, wind speed, and sunshine duration were retrieved for the period from 1991 until 2000 (reference period).
To assess possible future climate change effects at the regional scale, we used the COSMO-CLM Regional Climate Model with a convection permitting high spatial resolution of 1.3 km. Time series of the corresponding meteorological variables measured at Findel for the two pseudo stations were extracted. These pseudo stations were situated in climatologically characteristic regions in Luxembourg - Reuler in the North of the country and Esch-sur-Alzette in the South (Figure 1, a).
The COSMO-CLM is a non-hydrostatic limited-area atmospheric prediction model developed within the COSMO partnership (Consortium for Small-scale Modeling). It is operationally used by several European weather services (e.g. the German Weather Service or MeteoSwiss), as well as by many academic research groups, who develop and apply it as a regional climate model (Rockel et al. 2008). This study uses a three-step nesting approach. The global model ECHAM5 (forced with A1B scenario) has been downscaled to a resolution of 18 km in a domain covering continental Europe (Hollweg et al., 2008) and further down to a resolution of 4.5 km for a domain covering part of Western Europe (Gutjahr and Heinemann, 2013). The target domain is resolved with a resolution of 1.3 km and the integration time covered three 10-year time-slices 1991 until 2000, 2041 until 2050 and 2091 until 2100. The complete domain covers Luxembourg and the German states of Rhineland-Palatinate and Saarland, as well as parts of northern France and eastern Belgium. For the final run a Runge-Kutta numerical scheme with a time step of 15 s, 40 vertical levels, an advanced Kessler's scheme for microphysics (three category ice scheme including graupel), the Tiedtke scheme for moisture convection as well as the Ritter and Geleyn scheme for radiation were used (Junk et al., 2014).
A known problem is the bias of RCM driven by Global Climate Models compared with observed fields. Because appropriate bias correction methods are not available for wind speed or sunshine duration we decide not to bias correct the final RCM output. Additionally, standard bias correction schemes are not suited for maintaining the physical consistence between all variables used to calculate ETo.
In our study, we used the software tool "ETo calculator" developed by the Land and Water Division of FAO to calculate the grass reference evapotranspiration (ETo) according to FAO standards. A uniform grass field (complete soil coverage, well irrigated and regularly cut) is worldwide considered as a reference surface. The calculations were based on the Penman-Monteith equations (Allen et al., 1998) and closely approximates grass ETo, considering both physiological and aerodynamic parameters.
We used time series of the daily mean, minimum and maximum air temperature, relative humidity, mean wind speed, as well as the sunshine duration as input parameters for the ETo calculations. Quality checking procedures based on lower and upper threshold values for the meteorological input data were also included in the software (Allen et al., 1998; Raes, 2012).

3. Results & Discussion
At the reference station of Findel, monthly sums of ETo show a clear seasonal cycle, strongly correlated to monthly mean temperatures (Figure 1, b). Values typically reach or exceed 150 mm during summer months, and the minimum values are reached in winter, dropping below 25 mm. Generally, precipitation sums are higher than ETo sums in winter, but some exceptions have occurred in our 10-year analysis. On the contrary, during summer months ETo sums generally exceed precipitation sums.
For the climate change signals, an increase in mean ETo values is projected in all seasons for the near future (2041-2050) for both pseudo-sites (Figure 1 c, e). The increase is strongest in summer, with mean values being higher by approximately 10 mm, whereas in winter mean ETo values increase by only a few millimetres. For precipitation the strongest increase is projected in autumn months, by around 30 mm for mean values, and slight decrease in the sum for mean precipitation sums in summer by around 10 mm.
For the far future (2091-2100), summer months show a very clear signal, for both ETo and precipitation (Figure 1 d, f). ETo increases by nearly 30 mm and precipitation decreases by more than 20 mm for mean values. This is in line with the results presented by Baguis et al. (2009) who found shifts towards higher evaporations rates in Belgium, especially during the summer season.
For autumn, ETo values increase by roughly 15 mm, whereas the mean of precipitation sums returns to the values similar to the period 1991-2000, after they have risen for the near future. In winter, an increase of a few millimetres of ETo is projected, along a small decrease in mean precipitation. The spring months feature ETo sums similar to the period 1991-2000, after they have shown the second largest increase during the near future period. This indicates that other parameters than temperature can play a dominant role for the climate change signal, in line with Gong et al. (2006), who showed that the relative humidity is more important than the air temperature for the Penman equation for a case study in China.
For the far future, confidence intervals, based on standard deviation are all positive for ETo, except for spring. Both pseudo-stations that have been analysed show a similar climate change signal as well for ETo, as well as for precipitation, despite a stronger increase in temperatures for the Esch-sur-Alzette station as documented by Junk et al. (2014).

4. Conclusion
Our study demonstrated significant changes in two important components of the water cycle, i.e. air temperature and precipitation for climate change projections for Luxemburg. For the far future time span the confidence intervals for all seasons were positive except for spring. For the summer, we expect an increase of around 20% in ETo while for winter no obvious changes were projected. For all seasons, except summer, changes in ETo and precipitation exhibited a strong non-linear behavior between the near and far future, essentially due to the complex interrelationship between temperature, humidity, wind speed and radiation.
Our analyses show that a physically consistent data set of all meteorological variables is needed to assess the influence of climate change on the water cycle. Therefore, bias correction approaches solely for air temperature and precipitation are insufficient. Additionally, the total lack of ETo measurements in Luxembourg that can be used for model validation must be solved. From the different possibilities to measure ETo from pan evapotranspiration, sap flow measurements, lysimeters and eddy covariance measurements we have chosen the latter to be applied in an upcoming transnational project.

Acknowledgements
We gratefully acknowledge the financial support of the Ministere de l'Enseignement superieur et de la Recherche (MESR) of the Grand-duchy of Luxembourg in the framework of the MOVE and REMOD research programs.

References
Allen, R., Pereira, L., Raes, D., Smith, M., 1998. Crop evapotranspiration - Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper In: FAO (Ed.). FAO, Rome, Italy.
Baguis, P., Roulin, E., Willems, P., Ntegeka, V., 2009. Climate change scenarios for precipitation and potential evapotranspiration over central Belgium. Theoretical and Applied Climatology 99, 273-286.
Bormann, H., 2011. Sensitivity analysis of 18 different potential evapotranspiration models to observed climatic change at German climate stations. Climatic Change 104, 729-753.
Goergen, K., Beersma, J., Hoffmann, L., Junk, J., 2013. ENSEMBLES-based assessment of regional climate effects in Luxembourg and their impact on vegetation. Climatic Change 119, 761-773.
Gong, L., Xu, C.-y., Chen, D., Halldin, S., Chen, Y.D., 2006. Sensitivity of the Penman-Monteith reference evapotranspiration to key climatic variables in the Changjiang (Yangtze River) basin. Journal of Hydrology 329, 620-629.
Gutjahr, O., Heinemann, G., 2013. Comparing precipitation bias correction methods for high-resolution regional climate simulations using COSMO-CLM. Theoretical and Applied Climatology 114, 511-529.
Hollweg, H., Boehm, U., Fast, I., Hennemuth, I., Keuler, K., Keup-Thiel, E., Lautenschlager, M., Legutke, S., Radtke, K., Rockel, B., Schubert, M., Will, A., Woldt, M., Wunram, C., 2008. Ensemble Simulations over Europe with the Regional Climate Model CLM forced with IPCC AR4 Global Scenarios. M & D Technial Report, p. 152.
IPCC, 2014. Summary for Policymakers. In: Field, C.B., Barros, V., Dokken, D.J., Ebi, K.L., Mastrandrea, M.D., Mach, K.J. (Eds.), Climate Change 2014: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Junk, J., Matzarakis, A., Ferrone, A., Krein, A., 2014. Evidence of past and future changes in health-related meteorological variables across Luxembourg. Air Quality, Atmosphere & Health 7, 71-81.
Kannan, N., White, S.M., Worrall, F., Whelan, M.J., 2007. Sensitivity analysis and identification of the best evapotranspiration and runoff options for hydrological modelling in SWAT-2000. Journal of Hydrology 332, 456-466.
Kintisch, E., 2014. In New Report, IPCC Gets More Specific About Warming Risks. Science 344, 21.
Penman, H.L., 1948. Natural Evaporation from Open Water, Bare Soil and Grass. Proc. R. Soc. Lond. A 22, 120-145.
Prudhomme, C., Williamson, J., 2013. Derivation of RCM-driven potential evapotranspiration for hydrological climate change impact analysis in Great Britain: a comparison of methods and associated uncertainty in future projections. Hydrology and Earth System Sciences 17, 1365-1377.
Raes, D., 2012. The ETo Calculator - Evapotranspiration from a reference surface - Reference Manual Ver. 3.2. Food and Agriculture Organization of the United Nations, Rome. Italy, p. 38.
Thornthwaite, C.W., 1948. An Approach towar a Rational Classification of Climate. Geographical Review 38, 55-94.
Turc, L., 1961. Estimation of irrigation water requirements, potential evapotranspiration: a simple climatic formula evolved up to date. Annals of Agronomy 12, 13-49.
Wehner, M., Easterling, D.R., Lawrimore, J.H., Heim, R.R., Vose, R.S., Santer, B.D., 2011. Projections of Future Drought in the Continental United States and Mexico. Journal of Hydrometeorology 12, 1359-1377.