Hydrological Impact of Climate Change: A sensitivity study for the Netherlands

In this thesis various aspects of hydrology and the water management system of the Netherlands have been modeled and the sensitivity to climate change, due to increased greenhouse gas concentrations, has been assessed for various system parameters. The use of climate analogies was introduced as a means to obtain boundaries within which the impact of climate change will fall. Besides climate analogies also hypothetical climate scenarios have been applied.

In chapter 1 the context of the topic `Hydrological Impact of Climate Change' was presented. It appeared that nowadays reliable predictions of future climate change cannot be given; especially the prediction of precipitation is problematic. While waiting for more reliable climate scenarios, hydrologists can perform sensitivity studies or they can contribute to the development of macroscale hydrologic models for inclusion in climate models. In this thesis the first option has been chosen.

Chapter 3 dealt with the climate inversion problem. The climate inversion problem is the problem that arises when climate data with a certain space and time scale, must be transformed into data with smaller space and/or time scales. It was shown that up till now hydrologists and water resources engineers use relatively simple methods to transform existing meteorological time series to obtain climate scenarios. In two ways it has been shown that the results of hydrological impact studies are rather sensitive to the method used to transform existing meteorological time series. First, using the probability distribution of annual precipitation depth of De Bilt and the corresponding frequencies of extreme events. Second, using the maximum direct runoff resulting from a triangular storm as calculated for hypothetical catchments. In the latter case, it appeared that the magnitude of the dimensionless characteristic response time, determines to a great extent the sensitivity to the method used for the transformation of the meteorological time series.

The climate analogy method has been applied as a possible approach to the climate inversion problem. Using this method, the climate data of Lisboa, Porto, Bordeaux, Nantes, Plymouth, Southampton, Gdańsk, Göteborg and Stockholm (daily mean temperature and daily precipitation depth in the period 1970–1990) have been used as analogies for the possible future climate in the Netherlands. The climate data of De Bilt have been used as representative for the present average climate in the Netherlands.

In chapter 2 the impact of climate change on open water evaporation and potential evaporation (read evapotranspiration) was studied. Three methods for calculating evaporation have been compared with respect to temperature changes alone, the Thornthwaite-method, the Makkink-method, and the Penman-method. Of the three methods, the Penman-method was considered to be the best method to simulate the impact of climate change on evaporation values because the Penman-method is the most physically-based method; knowledge of changes in variables affecting evaporation can explicitly be taken into account.

Using the Penman-method, the present relationship between temperature and open water evaporation was examined for De Bilt. The effects of changes of other relevant climate variables was also examined. It appeared that the present relationship between temperature and evaporation for the present climate cannot be used as a key for future climate conditions because temperature is correlated with sunshine duration and relative humidity.

Scenarios for evaporation have been obtained using climate analogy data and hypothetical changes in the data for De Bilt. It appeared that the influence on evaporation values of an increase of +5W/m2 in net radiation due to a doubling of the pre-industrial carbon dioxide concentration is small compared to the influence of a change in temperature.

From previous published research it has been concluded that it is not clear what the net response of plants will be as a result of increased carbon dioxide levels. As a result the changes in crop factors, due to the greenhouse effect, are rather uncertain. As a first order approximation it has been assumed that crop factors do not change.

It may be concluded that the calculation of evaporation for changed climate conditions is still a complicated task because information on several influencing factors is not yet available.

Chapter 5 dealt with the impact of climate change on the overflows of combined sewer systems. The sewer systems have been modeled as a reservoir with an existing precipitation record of Lelystad (15 years of precipitation depth at 5 minute time intervals) as input. Overflow variables were defined for the reservoir. The sensitivity of those variables to changes in the reservoir parameters and changes in climate were assessed using the reservoir model. Because the climate analogies comprised only daily precipitation depth series, and not five minutes precipitation depth series which were required for the model, results for the climate analogies could only indirectly be obtained and were restricted to the overflow frequency off. The hypothetical climate scenarios consisted of transformations of the precipitation depth series of Lelystad.

It has been shown that the daily data of the climate analogies may be used to calculate the changes in overflow frequency. The results for the climate analogies give a reasonable upper and lower boundary; the lower boundary for off varies between a decrease of 15% to a decrease of 30% and the upper boundary for off varies between an increase of 35% to and increase of 130%.

It has been shown that for a constant multiplication factor and a seasonal dependent multiplication factor, the relative changes in the overflow variables are very sensitive to changes in the reservoir parameters. Therefore, the impact of a future change in the precipitation series will be different for each sewer system. Furthermore, the overflow variables showed a clear seasonal pattern with minima in the months of December, January, February, March, and April and maxima in the months of June, July, and August. Therefore, it is important to know the exact seasonal pattern of a future climate change, because a seasonal pattern in a future climate change may change the values of the overflow variables considerably; e.g., the effect of a net annual increase in precipitation depth of 20% on the maximum annual overflow intensity may be greatly amplified by a seasonal pattern when this pattern has a peak in the summer month or may be nearly zero when this pattern has a dale in the summer months.

Besides multiplying the precipitation series of Lelystad by a factor also the addition of precipitation was briefly considered. As in chapter 3 the results showed that the method used for transforming an existing precipitation series to obtain an increase or decrease in precipitation depth, is extremely important.

It thus appeared that a future climate change may have important consequences on the overflow frequencies and other overflow variables of sewer systems and, therefore, on the receiving water-ecosystems. Since the life span of sewer systems and the time scale of a climate change are of the same order of magnitude, it may be wise now to take into account the possible effects of climate change in the design of sewer systems.

In general, it may be concluded that for the purpose of assessing the impact of climate change on sewer systems it is useless to construct more complicated sewer system models than the one presented, as long as there is no certainty on how a climate change will manifest itself in the short-term precipitation data. The climate modelers should rather provide much more detailed information, if possible at all, about how future climate change will manifest itself in short-term precipitation records; it is not enough, e.g., to know that the annual increase in precipitation depth will be a certain percentage. Even knowing the monthly increase or decrease is not enough. It must be known whether an increase or decrease in precipitation depth will, e.g., cause more or fewer storms or whether the increase or decrease will be manifested by multiplication or by addition or subtraction of the original series, or
by some combination of these.

The sensitivity of urban groundwater levels in the Netherlands to climate change was studied in chapter 6. The study was restricted to the direct influence of the precipitation excess on groundwater levels. It appeared that the availability of long time series in urban areas of the Netherlands, of sufficient quality, is rather limited. Because detailed information on the drainage system and soil characteristics near the five analysed observation wells in the cities Delft, Groningen and Haarlem was not available, a stochastic/statistic model was constructed to model the time series of groundwater levels (measured twice a month).
Although the considered observation wells may not be representative for the corresponding cities, it is plausible that the results for all observation wells together give a reasonable first order estimate of the direct impact of changes in the precipitation excess on urban groundwater levels.

It was shown that the effects of changes in evaporation on groundwater levels are small compared to the effects of changes in precipitation depth. For the climate analogies (except Lisboa, Porto, Gdańsk and Stockholm), the extremely high groundwater levels (on average once in a two or ten year period) vary roughly between a rise of 0 to 200 mm; the extremely low groundwater levels vary roughly between a decline of 0 to 100 mm For the hypothetical climate scenarios, the extremely high groundwater levels vary between a rise of 121 mm and a decline of 133 mm; the extremely low groundwater levels vary between a decline of 60 mm and a rise of 56 mm. It may again be concluded that it is very important to know the distribution of a climate change over the different seasons.

Chapter 7 dealt with the sensitivity of public water supply to climate change. A nonlinear multiple regression model with thresholds has been developed to model time series of monthly public water supply for several supply areas in the Netherlands and for the Netherlands as a whole. For each supply area the parameters of the model were estimated and the seasonal part of the model was then used to estimate the impact of climate change. For each supply area the models were used to calculate time series of monthly seasonal water supply for the current climate, the climate analogies and the hypothetical climate scenarios. Several statistics of the output were compared.

It was shown that there is a clear relationship between the climate variables and the seasonal water supply, especially in those areas where the soils are sensitive to drought. For the investigated series of monthly water supply there was always a statistically significant dip in the month of July because there is probably less people in the areas in that month due to a negative balance of tourism. The inclusion of thresholds in the model for the open water evaporation, the number of rainy days and the precipitation deficit, improves the fit and reliability of the model. For some provinces, notably Friesland, Groningen and Zeeland, the effect of climate is small or obscured by other effects.

For the climate analogies (except Lisboa and Porto) the annual maximum seasonal water supply (on average once in a ten year period) for the Netherlands as a whole varies roughly between an increase of 11.5% (4.6 l/cap.day) and a decrease of 18.5% (7.3 l/cap.day); for the hypothetical climate scenarios the corresponding values are an increase of 15.9% (6.8 l/cap.day) and a decrease of 10.9% (4.7 l/cap.day). For the supply area Oost Brabant these results vary between an increase of 21.4% (15.2 l/cap.day) and a decrease of 23.9% (15.2 l/cap.day) for the climate analogies and an increase of 21.1% (16.5 l/cap.day) and a decrease of 11.6% (9.0 l/cap.day) for the hypothetical climate scenarios.

For the climate analogies (except Lisboa and Porto) the annual mean seasonal water supply (on average once in a ten year period) for the Netherlands as a whole varies roughly between an increase of 41.1% (4.4 l/cap.day) and a decrease of 1.4% (0.2 l/cap.day); for the hypothetical climate scenarios the corresponding values are an increase of 28.1% (3.3 l/cap.day) and a decrease of 19.3% (2.3 l/cap.day); for Oost Brabant these results vary between an increase of 156.2% (25.7 l/cap.day) and a decrease of 2.0% (0.3 l/cap.day) for the climate analogies and an increase of 36.6% (6.5 l/cap.day) and a decrease of 22.4% (4.0 l/cap.day) for the hypothetical climate scenarios.

The differences between the results for the Netherlands as a whole and the various supply areas indicate that there are strong regional differences in the effects of climate change on seasonal water supply.

Chapter 8 dealt with the agricultural water supply in relation to climate change. The emphasis was not on actual water use but on the additional water demand of crops, which has been defined as the (positive) amount of water that must, by some means, be supplied to the root zone of crops to let them evaporate at potential rates. It has been assumed that evaporation of crops at potential rates results in a maximum crop yield. The analysis was restricted to grassland, as the major part of arable land in the Netherlands consists of grassland and because the evaporation of grassland is a reasonable upper limit for the evaporation of other crops. The unsaturated zone has been modeled as a single reservoir.

It has been assumed, as a first order approximation, that crop factors remain the same. However, to obtain an impression of the possible effects of changes in the crop factor, the crop factor has been varied between 0.65 and 0.95 for De Bilt and the climate analogies.

It was shown that for all climate analogies, except for Göteborg, in most cases dry and wet years are drier as compared to De Bilt. The magnitude of the additional water demand of crops in a dry year (on average once in a ten year period) is, in a simple manner, related to the water storage capacity of the root zone; e.g., an increase in storage capacity of y mm decreases the additional water demand of crops with y mm. For the climate analogies the additional water demand of crops in a dry year ranged between an increase of 285 mm for Lisboa and 0 mm for Göteborg. For Nantes an increase is found of 158 mm and for Southampton 110 mm.

For the climate analogies it was shown that the sensitivity of the additional water demand of crops to changes of the crop factor is climate dependent. The change in the additional water demand of crops per 0.1 change in the crop factor ranges between 44 mm for De Bilt and 77 mm for the analogy Lisboa.

For the hypothetical climate scenarios it appeared that the climate variables considered were about equally important with respect to the changes in additional water demand of crops. The results for the hypothetical scenarios were well within the upper limit of the results for the climate analogies. However, the lower limit is lower than for the climate analogies. For the lower limit it was found that the additional water demand of crops in a dry year is 50 mm lower than for De Bilt. The upper limit is an increase of 73 mm.

The last impact study was reported in chapter 9 and dealt with the discharge of surplus water from agricultural polders in the Netherlands. Surplus water has been defined as the amount of water that must be evacuated from the polders in order not to exceed a specified water level in the system of ditches and canals.

The open water levels in a typical polder system in the Netherlands have been modeled with a combination of three reservoirs, one reservoir for the unsaturated zone, a linear reservoir for the saturated zone, and a reservoir for the hydraulic transport system. The sensitivity of the open water levels to changes in several parameters was studied using this model. A first order estimate of the sensitivity to climate change has been given using climate analogies and hypothetical climate scenarios. Intermediate results for the unsaturated and saturated zone of the land area have also been presented.

It was shown that for De Bilt and the climate analogies, the amount of water that can be stored in the unsaturated zone in the winter period is rather small. In that period the influence of the magnitude of the maximum storage capacity of the unsaturated zone on the discharges into the hydraulic transport system (and thus the open water levels) is negligible. For De Bilt and the climate analogies rather large amounts of water (dependent on the maximum storage capacity of the unsaturated zone) can be stored in the unsaturated zone in the summer period, reducing the effect of possible heavy convective rainstorms on the discharges into the hydraulic transport system in that period. As expected, most of the high groundwater levels and large discharges into the hydraulic transport system (and thus the occurrence of high open water levels) occur in the winter half year for both De Bilt and the climate analogies. For some of the climate analogies (notably Lisboa, Porto and Bordeaux) the changes in groundwater levels may not be acceptable in practice and measures must be taken to reduce the impact of high groundwater levels or to reduce the high groundwater levels themselves.

As expected, the influence of the reservoir constant of the saturated zone on the height of the open water levels was considerable, whereas the influence of the storage capacity of the unsaturated zone is negligible.

For the climate scenarios a discharge capacity was calculated such that the frequencies of high water levels in the hydraulic transport system remained the same as for the present situation (De Bilt). The ratio of this discharge capacity and the original discharge capacity was named dc*. For most of the investigated climate scenarios dc* > 1. The effects of doubling the percentage of open water (for De Bilt only) on dc* are a reduction of dc*, with the reduction being greatest for small values of the present discharge capacity dc (dc < 9 mm/day). The results for dc* depend very much on the present dc of the polder system concerned and on the exceedance frequency one is interested in. It appeared that the results for Porto represent the (rather extreme) upper boundary, whereas the results for Gdańsk and Stockholm represent the lower boundary for the scenarios investigated in this chapter.

Finally, chapter 10 discussed the contents of the thesis in a broader context and gave recommendations for further study. Further research is necessary (1) to obtain more reliable climate scenarios of sufficient duration (especially precipitation scenarios); (2) to obtain a better impression of the spatial variability of the impact of climate change in the Netherlands; (3) to assess the impact of frozen soils and water courses; (4) to obtain insight into the behavior of crop growth and transpiration as a result of increased carbon dioxide levels; and (5) to obtain insight into the importance of man-induced influences as compared to the impact of climate change.