Water scarcity is becoming an increasing problem worldwide, especially in arid regions. A lack of fresh water has been considered a serious cause of current and future conflicts both regionally and internationally [1
]. In humid regions, the problems are less socially and politically severe. As a result, publications about regional water scarcity in such regions are scarce. However, in humid regions, scarcity does also arise during parts of the year because the water demand is temporarily larger than the available amount of water which can be used without causing undesirable effects.
It may seem odd that in the Netherlands, water scarcity caused a conflict, since this country (35,000 km2
, Figure 1
) is situated in the delta of the rivers Rhine and Meuse and because it has a humid climate. Moreover, groundwater levels in most of the country are shallow (less than 1–2 m below soil surface), so that in the growing season (summer), capillary flow from the groundwater table to the rooting zone is an important source of water supply to plants [3
]. Why then, with this apparent surplus of fresh water, should there be water scarcity? The answer lies in the fact that water tables in The Netherlands are controlled as much as possible to meet the demands of land-use functions, particularly of agriculture.
After World War II, Europe developed a common agriculture policy to abandon hunger and starvation [4
]. In line with that, the Dutch government stimulated land improvement to increase crop production. The Netherlands succeeded like no other country, with the result that, nowadays, after the United States, it is the second largest exporter of agricultural products in the world (www.cbs.nl/en-gb
). To achieve this, in most areas, the water management system was considerably adapted, especially to get rid of excess water, by digging and re-dimensioning ditches and by installing drainage pipes, weirs, and pumps. As a result, groundwater levels dropped and crop production increased, as intended. Though this operation brought wealth to farmers and the rest of the Dutch population, there were also unintended negative side-effects: heavy pollution of soil, water and air [5
], and decline of groundwater levels in nature areas adjacent to farmland, which resulted in a serious loss of conservation values. In 1990, the Dutch parliament recognized this loss as a national problem that should be tackled [8
]. The basis for this decision was a number of research reports, among which was the analysis by Rolf [9
] of trends of groundwater levels observed from 1950 till 1989 in a large number of piezometers installed outside the influence of groundwater abstraction wells in the sandy Eastern and Southern ‘high’ part of the country (Figure 1
). Rolf [9
] found that groundwater levels in these four decades had dropped, on average, 0.3 m more than he could explain using a time series of precipitation and reference evaporation. This gap of 0.3 m he called ‘achtergrondsverdroging’, which can be translated as ‘background decline’. A background decline of similar magnitude was also observed by others who used numerical or analytical groundwater models. Since the publication of Rolf [9
], numerous reports and papers appeared about this hydrological black hole, unfortunately all in Dutch (see Supplementary Material
): Hydrologists analyzed groundwater level time series in a different manner, used additional information like soil maps, and tried to find flaws in computation methods and in measurement techniques, etc. In the last decade, the debate among Dutch scientists intensified, and in 2013, the association of Dutch hydrologists started a study commission on ‘background decline’. In 2016, the editorial staff of the most important professional journal for Dutch hydrologists, Stromingen
, announced a complete ban on the matter: Publications about background decline would no longer be accepted as further discussion on the topic would not provide new insights.
The debate about the causes of the background decline is not only of interest to hydrologists but might also have serious juridical and financial consequences, especially because Dutch farmers are entitled to financial compensation for crop damage caused by groundwater abstractions. Because of this, drinking water companies pay millions of euros to farmers every year. More expensive but harder to quantify are the salaries of a whole army of civil servants, lawyers, judges, specialists, and arbitration commissions who are involved in disputes over the effects of groundwater extractions on crop yields.
There have been many changes since 1950 that could have caused the background decline of groundwater levels in the Netherlands. The history of the Dutch landscape needs to be considered to study these causes, since many landscape changes have hydrological consequences: Wet areas have been extensively drained; cities have expanded; abstraction wells were installed for drinking water supply, industry, and agriculture; infrastructure was constructed, sometimes disrupting impermeable layers; polders were reclaimed from the sea and lakes; in clay and peat areas, drainage caused considerable soil subsidence; etcetera. Without careful investigation of these causes, the measured decline in groundwater levels cannot be explained or is wrongly fully attributed to the only causes considered.
In this paper, we will focus on one possible cause of background decline not considered in previous studies, namely the anthropogenic changes in the groundwater recharge since 1950. Groundwater recharge is defined here as the amount of precipitation that does not evaporate, run off superficially, or disappear in the sewer system, but eventually percolates to the groundwater. Crop yields in agriculture have risen sharply over the last half century, and because water use of crops is proportionate to crop production, this must have led to more crop evapotranspiration. In addition, grasses, shrubs, and trees became more abundant in nature areas, partly under the influence of atmospheric nitrogen deposition and cuts on nature management. Those plants evaporate more than the original vegetation. Urban expansion, too, may have contributed to the reduction of groundwater recharge due to the fact that in urban areas a large part of the precipitation water flows into the sewer system and no longer reaches the groundwater.
Our study is an example of ‘forensic hydrology’, a term that was introduced in the scientific community in 2007 by the Southwest Hydrology magazine. In the preface on a volume dedicated to the topic, the publisher wrote [10
]: “Southwest Hydrology’s approach is to look at the hydrologic tools available to determine the history of an event—such as water contamination, recharge, or groundwater capture—that matters to some entity, for example, a manufacturer, well owner, or municipality. This issue’s authors also touch upon ways to make forensic investigations successful in and out of the courtroom”.
In our forensic study, we will analyze how changes in land use and in crop yield have affected groundwater recharge and groundwater table. We focus on changes in groundwater recharge and will not quantify the possible contribution of other changes in the landscape to the background decline of the groundwater table. We will compare two historical periods: the period around 1950 and around 2010 (i.e., 1947–1953 and 2007–2013, respectively). The reason for this approximate time denotation is that some of the available historical statistics cannot be attributed to a particular year. For both periods, further referred to as 1950 and 2010, we will investigate both land use and the yields that were achieved in agriculture. We will demonstrate that these historical changes had an impact on the groundwater table, which cannot be neglected. We will focus on the province of Noord-Brabant (Figure 1
), but the results of our study are applicable to the entire high part of the Netherlands, which mainly consists of Pleistocene cover-sand, intersected by brooks. Dominant land-use is agriculture (especially dairy farming and bio industry).
4.1. Limitations of Our Study
In this study, we dealt with the consequences of changes in land use and in crop yield on the groundwater levels in the province of Noord-Brabant. We did not take the land improvements that were carried out on a large scale after World War II into account explicitly. The main aim of these improvements was to prevent flooding and lower the groundwater table in the spring so that the soil could be cultivated with heavier machines, and the soil temperature and moisture content allowed better growth earlier in the year. The effect of land improvement is implicitly included in our results as far as it leads to higher crop yields.
In 1950, there was no overhead irrigation while the current irrigation in the province amounts to 70 million m3·y−1, equaling a water depth of 14 mm, which is 6% of the annual precipitation surplus (P − Eref). Virtually all irrigation water stems from groundwater and is applied to the fields using sprinkler installations. The irrigation allows for extra plant growth, and the corresponding transpiration is accounted for in the calculation. The irrigation return flow does recharge the groundwater again, so the net loss of groundwater due to the irrigation is equal to the extra evapotranspiration. The transpiration part is accounted for in the recharge reduction calculated from the crop yield. The calculations do neglect the extra evaporation which occurs due to the sprinkling, but the remaining effects of the irrigation are taken into account implicitly.
The consequences of the expansion of paved surfaces and of the increase of shrubs and trees in nature reserves could not be properly quantified in this study. The recharge in urban areas is uncertain, because little is known about the evaporation of this category [30
Our calculation of groundwater recharge for arable land neglects the increase in the efficiency of crop production. The harvest losses were larger in 1950 and there were more weeds on agricultural fields. By way of uncertainty analysis we therefore also analyzed a scenario in which the increase in transpiration of arable land between 1950 and 2010 is half as large as we deduced from the increase in crop production. This resulted in an average recharge in 1950 of R
= 370 mm·y−1
(instead of 294 mm·y−1
). This scenario was thoroughly discussed with the eight members of the study commission on ‘background decline’, mentioned in Section 1
. They agreed the resulting recharge in 1950 should be considered as an upper limit (details in [13
]). The corresponding average decline in groundwater table in the sandy regions of was 23 cm (instead of 33 cm). In conclusion, our study suggests that the effect of changes in land use and crop production in 60 years’ time lies approximately between 0.2 and 0.3 m.
4.2. Comparison with Previous Work
Van Bakel and De Wit [31
] studied the effect of increased crop yield on a potato field in a fertile agricultural polder in the Netherlands. The controlled water level in the polder was optimized to crop growth. Using a model for water flow and crop growth, SWACROP [32
], they simulated the effect on evapotranspiration of an observed 40% increase in potato yield in the period of 1955–1987 (33 y). They used three scenarios, which resulted in an increase in actual evapotranspiration of 76, 74, and 43 mm·y−1
. The results of the first two scenarios were comparable to the 69 mm·y−1
they obtained from a water balance evaluation. When we linearly extrapolate this figure to the 60-year period of our study, we arrive at an evapotranspiration increase for the potato field of 125 mm·y−1
. In our study, we found an increase in the evapotranspiration of potato of 203 mm·y−1
(Equation (4)). The larger increase in evapotranspiration for Noord-Brabant is reasonable, since the potato field of Van Bakel and De Wit [31
] was designed in an optimal manner from the start, while the water management in the province of Noord-Brabant has been improved substantially for agriculture.
The average simulated decline of 0.2–0.3 m in 60 years’ time (1950–2010) in the sandy regions of the province of Noord-Brabant is comparable to the decline since 1950 that Rolf [9
] found in the analysis of piezometric time series in the Netherlands, away from groundwater extractions.
4.3. General Implications
We performed our analysis for the Dutch province of Noord-Brabant. We accounted for urbanization, increased agricultural crop yield, and vegetation changes in nature areas. These processes do occur in many other regions around the globe and will have a similar effect everywhere.
Our conclusion can also be generalized to say that the drawdown due to groundwater extractions may change over time and is ambiguous as long as the drainage situation and groundwater recharge are not properly defined. Generally, the effect of groundwater extraction increases with decreasing groundwater recharge. The non-linearity of groundwater systems becomes more important when longer time periods are considered. This means the effects of influences are interdependent and the order in time needs to be considered for long-term hydrological studies. This result may be of importance to environmental impact assessment studies. Should, for instance, the compensation budget to farmers of a groundwater abstraction be based on the land use and crop yield in the year that the abstraction started or on the current land use and crop yield? This is a question with possibly large technical, juridical, and financial consequences.
To our knowledge, our study is the first example of forensic hydrology into anthropogenic causes of possible water scarcity in a humid region. We foresee that the importance of forensic hydrology will increase substantially in the coming decades. Facing climate change, a growing human population, and more welfare, the demand for water especially for agriculture, households, and industries will increase and, with that, the risk of conflicts as well. Some changes will be very gradual, as we demonstrated with the gradual increase in crop production and land-use in the province of Noord-Brabant. This emphasizes the need to properly register changes in the landscape and to perform sufficient hydrological measurements. Unlike our forensic study, current and coming changes in landscape patterns nowadays can be observed quite easily with the aid of satellites.