Climate change , re-/ afforestation , and urbanisation impacts on evapotranspiration and streamflow in Europe

Since the 1950s, Europe has undergone large shifts in climate and land cover. Previous assessments of past and future changes in evapotranspiration or streamflow have either focussed on land use/cover or climate contributions, or on individual catchments under specific climate conditions but not on all aspects at larger scales. Here, we aim to understand how decadal changes in climate (e.g., precipitation, temperature) and land use (e.g., de-/afforestation, urbanization) have impacted the amount and distribution of water resources availability (both evapotranspiration and streamflow) across Europe 5 since the 1950s. To this end, we simulate the distribution of average evapotranspiration and streamflow at high-resolution (1 km) by combining a) a steady-state Budyko model for water balance partitioning constrained by long-term (lysimeter) observations across different land-use types, b) a novel decadal high-resolution historical land use reconstruction, and c) gridded observations of key meteorological variables. The continental-scale patterns in the simulations agree well with coarser-scale observation-based estimates of evapotranspiration, and also with observed changes in streamflow from small basins across 10 Europe. We find that strong shifts in the continental-scale patterns of evapotranspiration and streamflow have occured between the period around 1960 and 2010. In much of central-western Europe, our results show an increase in evapotranspiration in the order of 5–15% between 1955– 1965 and 2005–2015, whereas much of the Scandinavian peninsula shows increases exceeding 15%. The Iberian peninsula and other parts of the Meditteranean show a decrease in the order of 5–15%. A similar north-south gradient was found for changes in 15 streamflow, although changes in central-western Europe were generally small. Strong decreases and increases exceeding 45% were found in parts of the Iberian and Scandinavian peninsulas, respectively. In Sweden, for example, increased precipitation is a larger driver than large scale reand afforestation, leading to increases in both streamflow and evapotranspiration. In most of the Mediterranean, decreased precipitation combines with increased forest cover and potential evapotranspiration to reduce streamflow. In spite of considerable local and regional scale complexity, the response of net actual evapotranspiration to changes 20 in land use, precipitation, and potential evaporation is remarkably uniform across Europe, increasing ∼35–60 km/y, equivalent to the discharge of a large river. For streamflow, effects of changes in precipitation (∼95 km/y) dominate land use and potential


Interactive comment
Printer-friendly version Discussion paper Since the forest cover effect is hardcoded in the Budyko model, it will simulate changes in ET.However, this remains an extrapolation, which needs a better validation than what is now presented.The authors mention that the modelled ET average agrees well with the patters of GLEAM.Here I would like to ask the the authors to report statistics of this comparison.Then they also report a correlation with streamflow changes of r = 0.34, which corresponds to an explained variance of 12%, leaving 88% unexplained!Please show the scatterplot.Since there is a need for a validation of the model, I think that the model should be able to predict the observed streamflow changes better than a reference, for example using the changes in precipitation and maybe PET.Only if the Budyko model shows a higher skill I see justification to use that model and its change of the landuse parameterisation for the whole of Europe with confidence.
Land-use change is modelled by changing the parameter in the Budyko model using data from lysimeters.This is quite a central methodological step and ignores differences in scale of a lysimeter with that of a heterogeneous landscape.It also ignores that the parameter in the Budyko model can be different due to climatic variation, in particular the seasonality of rainfall to that of evaporative demand and the rainfall frequency.Jaramillo et al., 2018 HESS showed that there are increases in evaporative fraction, not explained by climate for many catchments in Sweden.Yet the link to changes in forest properties was rather weak.In contrast this study prescribes a distinct effect of forest age, hence there is a strong tendency that this study assesses the upper range of changes in water balance (if the HILDA database actually reflects the changes).
The choice of Thornthwaite method for PET is not acceptable for various reasons: a) It underestimates the evaporative demand (PET or Rn/L, see also van der Schrier 2011 or Maes et al., 2018).An annual average of PET of 700mm/yr for Southern Europe is far to low.That is why the authors need to scale it by an arbitrary factor aPET in the Budyko curve.b) Since it is a function of temperature only, it will be overly sensitive to warming trends which is arguably pretty strong for the considered period.It also misses changes in shortwave solar radiation, see e.g.Wild et al., 2007. c) The authors argue for Thornthwaite because of data availability.However, there is data on sunshine duration / cloud cover.Furthermore, the diurnal temperature range correlates with solar radiation and has been used as a proxy for this, e.g.Wild et al., 2007, Makowski et al. 2008.Apart from these major issues I enjoyed reading the paper.It is very well written, is well structured and has appealing figures.The topic is of high relevance for HESS.However, I believe that the validity of the Budyko approach needs to be demonstrated and therefore I recommend major revisions.

Minor Remarks:
Introduction, L20ff: it is argued that there are no sufficient studies which treat both landuse change and climate change on streamflow / ET.However, there are studies which indeed try to accomplish this, which I want to bring to the attention of the authors.For example Jaramillo et al., 2018   The choice of rectangular sub-regions seems arbitrary to me.Why not use relevant river basins, where data is available to see if your prediction is indeed pointing in the right direction.For example on P9L10 it is mentioned that Scotland shows dramatic increases in streamflow, is this finding supported by observed changes?Table 3: The units in the caption should be kmˆ3/yr and not km/yr.In any case I would assessed changes in multiple catchments in Sweden.Renner et al., 2014 assessed observed changes of streamflow in East Germany.Lopez-Moreno et al., 2011 for catchments in Spain.

Figure 3 :
Figure 3: color of missing values (NA) should not be white, as indicated in the legend Figures 6,7: there should be a color legend, a 3D color scheme on a map is a beautiful drawing but really difficult to grasp.What is the meaning of grey here?Similar magnitude of all drivers or a missing value?To what reference are the data scaled 2-98%, all of Europe?