OR/14/018 Results

Mansour, M M, Hughes, A G. 2014. Land Use, Climate Change and Water Availability: Preliminary modelling of impacts of climate change and land use change on groundwater recharge for England and Wales. British Geological Survey Internal Report, OR/14/018.

Introduction

The following section presents the results produced for the three sets of models runs: historical simulation, climate change using the FFGWL hydrological datasets and land use change. The historical simulation is used as a 'basecase' to which the results to the CC and land use simulations are compared. For simplicity long-term average (LTA) potential recharge is used for comparison. The model is run for the full time period (January 1962 to December 1992 for the historical simulation) and then average recharge for this time period produced for each node.

Two ways of presenting the results are used: the first are maps of LTA recharge for the whole of England and Wales and the second are box-whisker plots. The latter is used to summarise differences in behaviour between catchments.

Box and whisker plots are a convenient way of graphically displaying the statistical characteristics of numerical data. A whisker plot is defined mainly by five values:

• The mean of the data which sits in the centre of the box;
• the lower and upper limits of the box which are also called the lower quartile (Q1) and the upper quartile (Q3); and
• the two bars outside the box which are the minimum and maximum values that are not outliers. Outliers below the lower whisker are all the values that are less than and those above the upper whisker are all the values that are greater than with IQR defined as the inter quartile range, which is the distance between Q1 and Q3. Outliers are rare values but can happen.

Historical simulation

Figure 8 shows the LTA potential recharge for the various runs undertaken including the overall historical simulation for England and Wales (top left). The recharge gradient is mainly west to east with potential recharge decreasing from >1200 mm/a in western Wales to <100 mm/a in north Norfolk. The influence of higher rainfall due to orographic effect in Wales and north-west England can be clearly seen. Other influences such as soil type can be observed in the Thames and Wealden basins. The LTA recharge clearly shows the combined influence of spatial distribution of rainfall, PE, land use, soil and geology at outcrop. It is the interaction between these factors and changes to the driving data (FFGWL climate data) and land use which is presented below.

Figure 8    Long-term average recharge for the basecase.

Figure 9    LTA recharge for FFGWL climate change runs.

Climate change

Long-term averages for England and Wales

Figure 9 presents the LTA results for the 11 RCMs for each time slice. The first column shows absolute values for the LTA for Future Flows historical simulation, the remaining columns display differences. Column two shows the difference between the FFGWL and the historical simulation, columns three to five show the difference between the FFGWL time slices and the FFGWL historical simulation. The difference between the FFGWL historical simulation and that of the actual historical simulation is necessary as the perturbations in the initial conditions for each FFGWL simulation results in different time series of recharge.

Table 4 shows the summary of the differences for the average recharge for the model simulation. The differences in the average have been summarised by the bars in the right hand column of the table, with blue representing an increase and red a decrease. Bars are produced for the difference between each time slice (2010–39, 2040–2069, 2070–99) and the simulated historic.

Since each RCM has different starting conditions so as to achieve the variability in the future predictions (three timeslices: 2010–39, 2040–2069, 2070–99). These variations between the RCMs also affect the simulation of the historical period which can be compared against recharge calculated for observed data. Therefore, to understand how the different RCMs perform against know conditions the results are compared with those computed from observed data. These are presented in Table 4 in column 2 and presented pictorially in column 6. The latter is shown to illustrate the difference between the simulated historic (resulting from the RCM) and the historical simulation based on gridded observed data.

Examining the difference between the simulated historic and the historic (Table 4; column 2 and 6) shows that the majority of the RCMs are dryer than the observed (afgcx, afixa, afixi, afixj, afixk, afixm and afixq) with the remainder being wetter (afixc, afixh and afixl). Comparing these with the future predictions (Table 4; columns 3 to 5 and 7 for a pictorial representation) allows the examination of whether this pattern is followed in the results for the timeslices. Generally there is greater recharge in the historical simulation than the future predictions. Only simulations afixi and afixk are dryer in both the historical simulation and the future predictions. This suggests that the predictions using the RCMs underestimate recharge for the future predictions.

The following summarises the variation between the future predictions based on the RCMs:

afgcx: the historical simulation results in slightly lower recharge with increasing recharge over the subsequent time slices.

afixa: This historical simulation produces the lowest recharge with recharge increasing over the time slices, but starting from a reduced situation.

afixc: The historical simulation produces slightly increased recharge compared to the historical simulation. Recharge increases over the time slices with the increase in the 2080s being twice that of the 2020s and 2050s.

afixh: This shows the greatest increase in recharge from the historical summation to the simulated historic produced by the RCM. This set of runs produces the greatest increase with the 2080s showing the biggest increase.

afixi: A slight reduction in recharge is observed for the simulated historic. The time slices show an increase in recharge from initially negative value.

afixj: Similarly to afixi, a slight reduction in recharge is observed for the simulated historic. The results for the timeslices are generally lower the 2050s showing the greatest decrease.

afixk: Similarly to afixj, a slight reduction in recharge is observed for the simulated historic. The results for the timeslices are generally lower the 2050s showing the greatest decrease.

afixl: The historical simulation produces significantly increased recharge compared to the historical simulation. The timeslices show an initial increase in recharge but then shows a reduction.

afixm: There is a decrease in recharge compared to the historical simulation, but recharge increases with the greatest increase being for the 2050s.

afio: Recharge is slightly greater for the historical simulation with a reduction for the 2020s and a significant one for the 2050s. There is slightly increased recharge for the 2080s.

afixq: There is a decrease in recharge compared to the historical simulation, but recharge increases with the greatest increase being for the 2080s.

Overall the results show that for the 2050s then recharge is generally lower and further out for the 2080s then recharge is generally higher. The results are mixed for the recent time slice 2020s, with equal numbers of increases and decreases.

Table 4    Summary of differences in average LTA recharge for each RCM timescale (mM/a).

Note: Average differences are shown as coloured squares one for each column of data; blue is a positive difference and red is a negative one.

Catchment summaries using Box-Whisker plots

The following sections describe the variations in recharge values calculated over England and Wales as a whole, and also as sampled for each of the focus CAMS catchments. Recharge values presented in these sections are given in Table 5 and also shown in Figures 10 and 11. The average and maximum LTA recharge values calculated using the historic rainfall and evaporation data are shown in the first and second columns. The third and fourth columns give the maximum and minimum values of the 11 averages of LTA recharge values calculated for the 11 future runs of the first time horizon 2010–2039 (2020s). The fifth and sixth columns contain the second time horizon 2040–2069 (2050s), and the seventh and eighth columns hold the values for the third time horizon 2070–2099 (2070s).

Table 5    Summary of long term average historic and future recharge value characteristics

Catchment	Historic		2010–2039		2040–2069		2070–2099
	Average LTA recharge	Maximum LTA recharge	Highest average LTA recharge of 11 runs	Lowest average LTA recharge of 11 runs	Highest average LTA recharge of 11 runs	Lowest average LTA recharge of 11 runs	Highest average LTA recharge of 11 runs	Lowest average LTA recharge of 11 runs
Dee	0.716	2.911	H: 0.783	J: 0.664	H: 0.768	K: 0.621	H: 0.809	K: 0.613
ElyOuse	0.235	0.682	H: 0.284	A: 0.168	H: 0.31	O: 0.185	H: 0.36	J: 0.216
HampAvon	0.773	1.308	H: 0.907	A: 0.632	H: 0.944	J: 0.653	H: 1.041	M: 0.681
Stour	0.531	1.246	L: 1.31	A: 0.349	H: 0.427	J: 0.337	I: 0.507	M: 0.363
Tees	0.36	2.132	H: 0.412	O: 0.34	H: 0.382	A: 0.346	H: 0.418	K: 0.343
Thames	0.4	1.39	H: 0.46	A: 0.3	H: 0.49	J: 0.34	H: 0.52	J: 0.31
Trent	0.393	2.277	H: 0.453	A: 0.343	H: 0.481	K: 0.339	H: 0.498	J,K: 0.367
Usk	1.374	3.463	H: 1.52	GCX: 1.29	H: 1.474	K: 1.287	H: 1.591	K: 1.315
England and Wales	0.612	7.69	H: 0.695	A: 0.57	H: 0.7	K: 0.56	H: 0.753	K: 0.595

The Dee catchment

The Dee catchment is located to the north of Wales, west of England and Wales (Figure 2). The historic LTA recharge values calculated over the Dee catchment have an average of 0.72 mm/day and a maximum of 2.91 mm/day. The LTA average recharge values calculated from the 11 rainfall and evaporation future projection values vary between -15 and 12 % of the historical LTA average recharge value with highest values calculated as 0.78, 0.77 and 0.81 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 0.66, 0.62 and 0.61 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10A shows the Whisker plots for the historic and future recharge values calculated over the Dee catchment for the three time horizons. The highest LTA average recharge value calculated as 0.81 mm/day from projection H for the 2080s. The lowest LTA average recharge value calculated as 0.61 mm/day from projection K for the 2080s. A Whisker plot for the differences between the future and historic recharge values are shown in Figure 11A.

The Ely-Ouse catchment

The Ely-Ouse catchment is located to the east of England (Figure 2). The historic LTA recharge values calculated over this catchment have an average of 0.24 mm/day and a maximum of 0.68 mm/day. The LTA average recharge values calculated from the 11 rainfall and evaporation future projection values vary between -30 and 50 % of the historical LTA average recharge value with highest values calculated as 0.28, 0.31 and 0.36 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 0.17, 0.19 and 0.22 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10B shows the Whisker plots for the historic and future recharge values calculated over the Ely-Ouse catchment for the three time horizons. The highest LTA average recharge value calculated as 0.36 mm/day from projection H for the 2080s. The lowest LTA average recharge value calculated as 0.17 mm/day from projection A for the 2020s. The differences between the future and historic recharge values can be clearly seen in this figure as well as in the corresponding Whisker plots shown in Figure 11B.

The Hampshire Avon catchment

The Hampshire Avon catchment is located to the south of England (Figure 2). The historic LTA recharge values calculated over this catchment have an average of 0.77 mm/day and a maximum of 1.31 mm/day. The LTA average recharge values calculated from the 11 rainfall and evaporation future projection values vary between -18 and 35 % of the historical LTA average recharge value with highest values calculated as 0.91, 0.94 and 1.04 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 0.63, 0.65 and 0.68 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10C shows the Whisker plots for the historic and future recharge values calculated over the Hampshire Avon catchment for the three time horizons. The highest LTA average recharge value calculated as 1.04 mm/day from projection H for the 2080s. The lowest LTA average recharge value calculated as 0.63 mm/day from projection A for the 2050s. As for the Ely-Ouse catchment, the differences between the future and historic recharge values can be clearly seen in Figure 11C.

The Stour catchment

The Stour catchment is located to the south east of England (Figure 2). The historic LTA recharge values calculated over this catchment have an average of 0.53 mm/day and a maximum of 1.25 mm/day. Calculation of recharge using the 11 rainfall and evaporation future projection values produces LTA average recharge values that are lower than the historical LTA recharge values. The future LTA average recharge values vary between -37 and -4 % of the historical LTA average recharge value with highest values calculated as 0.45, 0.43 and 0.51 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 0.35, 0.34 and 0.36 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10D shows the Whisker plots for the historic and future recharge values calculated over the Dee catchment for the three time horizons. The highest LTA average recharge value calculated as 0.51 mm/day from projection I for the 2080s. The lowest LTA average recharge value calculated as 0.34 mm/day from projection J for the 2050s. The Whisker plots of the differences between the future and historic recharge values in Figure 11D clearly shows that on average the predicted future values are lower than the historical LTA recharge values.

The Tees catchment

The Tees catchment is located to the north of England (Figure 2). The historic LTA recharge values calculated over this catchment have an average of 0.36 mm/day and a maximum of 2.13 mm/day. The LTA average recharge values calculated from the 11 rainfall and evaporation future projection values vary between -5 and 17% of the historical LTA average recharge value with highest values calculated as 0.41, 0.38 and 0.42 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 0.34, 0.35 and 0.34 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10E shows the Whisker plots for the historic and future recharge values calculated over the Tees catchment for the three time horizons. The highest LTA average recharge value calculated as 0.42 mm/day from projection H for the 2080s. The lowest LTA average recharge value calculated as 0.34 mm/day from projection O for the 2050s. However, there are no significant differences between the future recharge values calculated using the different projections as shown in Figure 11E.

The Thames catchment

The Thames catchment is located to the south east of England (Figure 2). The historic LTA recharge values calculated over this catchment have an average of 0.4 mm/day and a maximum of 1.29 mm/day. The LTA average recharge values calculated from the 11 rainfall and evaporation future projection values vary between -25 and 30% of the historical LTA average recharge value with highest values calculated as 0.46, 0.49 and 0.52 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 0.3, 0.34 and 0.31 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10F shows the Whisker plots for the historic and future recharge values calculated over the Thames catchment for the three time horizons. The highest LTA average recharge value calculated as 0.52 mm/day from projection H for the 2080s. The lowest LTA average recharge value calculated as 0.3 mm/day from projection A for the 2020s. Figure 11F how noticeable differences between the future recharge values calculated using the 11 different projections.

The Trent catchment

The Trent catchment is located in the centre of England (Figure 2). The historic LTA recharge values calculated over this catchment have an average of 0.39 mm/day and a maximum of 2.27 mm/day. The LTA average recharge values calculated from the 11 rainfall and evaporation future projection values vary between -13 and 28% of the historical LTA average recharge value with highest values calculated as 0.45, 0.48 and 0.5 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 0.34, 0.34 and 0.37 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10G shows the Whisker plots for the historic and future recharge values calculated over the Trent catchment for the three time horizons. The highest LTA average recharge value calculated as 0.5 mm/day from projection H for the 2080s. The lowest LTA average recharge value calculated as 0.34 mm/day from projection K for the 2050s. The differences between the future and historic recharge values can be clearly seen the corresponding Whisker plots shown in Figure 11G.

The Usk catchment

The Usk catchment is located south of Wales, west of England and Wales (Figure 2). The historic LTA recharge values calculated over this catchment have an average of 0.1.37 mm/day and a maximum of 3.46 mm/day. The LTA average recharge values calculated from the 11 rainfall and evaporation future projection values vary between -6 and 16% of the historical LTA average recharge value with highest values calculated as 1.52, 1.74 and 1.59 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 1.29, 1.29 and 1.32 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10H shows the Whisker plots for the historic and future recharge values calculated over the Usk catchment for the three time horizons. The highest LTA average recharge value calculated as 1.59 mm/day from projection H for the 2080s. The lowest LTA average recharge value calculated as 1.29 mm/day from projection K for the 2050s. The differences between the future and historic recharge values can be clearly seen the corresponding Whisker plots shown in Figure 11H.

England and Wales

The calculated historic long term average (LTA) recharge values vary spatially between near zero to approximately 7.7 mm/day with an average of 0.61 mm/day. The LTA average recharge values calculated from the 11 rainfall and evaporation future projection values did not vary significantly from the historical value with the highest values calculated as 0.69, 0.7 and 0.75 for the three time horizons 2020s, 2050s, and 2080s respectively. The lowest LTA average recharge values are 0.57, 0.56 and 0.6 for the three time horizons 2020s, 2050s, and 2080s respectively.

Figure 10I shows the Whisker plots for the historic and future recharge values calculated over England and Wales for the three time horizons. This figure also reflect the small variations in the calculated recharge values with the highest LTA average recharge value calculated as 0.75mm/day from projection H for the 2080s. The lowest LTA average recharge value calculated as 0.56 mm/day from projection K for the 2050s.

Additional analysis on recharge values calculated over selected strips across England and Wales

During the analysis of the recharge model output it was observed that there was greater variability in the Thames Basin in compared to the others. It was postulated that a possible cause of this difference was the size, orientation and position of the catchment. The Thames Basin is elongated in the east-west axis and covers a significant proportion of the distance from the coast to coast. This could mean that it is unduly affected by the west-east nature of the UK’s climate. Therefore, a number of runs were undertaken on strips running north-south and east-west. Four strips are selected at the locations and orientations shown in Figure 12. Additional statistical analyses have been performed on these four areas to investigate how the recharge values vary with the location and orientation of the catchment area being investigated.

The Box-Whisker plots of the future LTA recharge values are shown in Figure 13. This figure shows that the differences between the 11 projections LTA recharge calculated over the north south strip across Wales (Figure 13A) are not as clear as the those calculated over the north south strip across England (Figure 13B). It also shows that differences between the 11 projection LTA recharge values calculated over the east west strip at north of England (Figure 13C) are not as clear as those between the recharge values calculated over the east west strip at the south of England (Figure 13D).

Discussion of RCM variability

The results described above demonstrate that there is a significant variability between LTA recharge produced for each RCM for each timeslice. The RCM which consistently produces the greatest recharge is projection H (afixh). This is wetter for the historic simulation as well as the future predictions suggesting consistency between the historic simulation and future prediction. For the dryer, low recharge case then the results are more mixed, but projection A (afixa) appears to produce the lowest recharge, albeit for the 2020s. For the later timeslices (50s and 80s) then J and K (afixj and afixk) predominate. The latter are dryer for the historic simulation and the future predictions, again suggesting consistency between the historic simulation and future prediction.

Figure 10    Plots of historic and future recharge values.

Figure 11    Plots of differences between simulated future LTA recharge values and historic LTA recharge values.

Figure 12    Position of east-west and north-west stripes.

Figure 13    Plots of future recharge values over selected stripes.

Land use change

Long-term averages for England and Wales

Two figures have been produced to illustrate the spatial changes in LTA recharge produced by modifying land use (Figure 14 and 15). Both figures show the LTA recharge from the historical simulation (top left of the diagram). Figure 14 presents the change resulting from modifying the land cover mapping from LCM2000 to LCM2007. The LTA recharge produced using LCM2007 is presented in the bottom left and the differences shown in the centre of the figure. The results from modifying land use to either woodland, arable or grass are presented in a column on the right-hand side of Figure 14. Figure 15 presents the results from modifying land use at the appropriate spatial location. Here the results are presented in two columns and show the LTA recharge where 50% land use is modified from one type to another.

Examining Figure 14 shows that comparing the recharge produced by using LCM2007 vs LCM2000 provides overall very little difference, but locally these are significant changes. These changes are mostly prevalent in the West of England and Wales and represent a reduction in recharge. For the more radical changes to land use, the following can be observed:

• Woodland: covering the country in trees significantly reduces potential recharge (see Houghton-Carr et al., 2013) — as trees generally use more water than other crops (maximum root constant specified as 2 m), but there are subtleties (e.g. Roberts et a., 2005)
• Arable: covering the country in crops (a representative crop type that has a maximum root depth of 0.75 m and a crop depletion factor of 0.8 is used) increased recharge in urban areas and reduces it over the Welsh hills (change in routing depth)
• Grass: covering the country in grasslands significantly increases potential recharge (significantly reduced crop coefficients with maximum root constant of 0.12 m).

A more subtle approach involves changing one land use type with another at the grid cell where it occurs (Figure 15). This is undertaken for 50% of the overall land use being converted from one type to another. There are 10 landuse types specified in the model using 10 arrays of data. These arrays have the same size and their values represent the percentages of landuse types so at each location the sum of the ten values from these arrays must add up to 100. In the subsequent runs, a 50% of a landuse type is replaced by another landuse type buy halving its percentage value and increasing the percentage value of the replacement landuse type by the same amount. The three land use type (arable, grass and woodland) are paired up with each other to undertake these changes. Of these pairs, the most significant changes are as follows:

• Arable to woodland: significant reduction in the east of England 'bread basket effect'
• Grass to woodland: reduction in potential recharge over the whole country but predominantly in the western half
• Grass to arable: reduction in Wales and western England where managed grassland and semi-natural grass predominates

Catchment summaries using Box-Whisker plots

Figure 16 shows potential impact of complete (i.e. countrywide) land use change to either arable, grass, or woodlands on the calculated recharge values by using Box-Whisker plots. The differences between the LTA recharge values calculated using these land use types and the LTA recharges calculated using the dominant LCM 2000 land use are used to produce Box-Whisker plots. A whisker plot for the differences between the LTA recharge values calculated using the dominant LCM2000 and those calculated using the LCM2007 is also shown in Figure 16. All the plots share a common expected trend, and confirm the observations noted in Figure 14, that is the change of land use to woodlands results in significant reduction in recharge values and the change of land use to grass causes increase in recharge values compared to the values calculated using the dominant LCM2000. The use of land use arable has the lowest impact on the recharge values. This is because arable root depth falls between that of the grass and woodlands root depths, which consequently produces almost identical average recharge values. In general and on average the changes in land use from year 2000 to year 2007 did not cause significant impact on the calculated recharge values with the Dee and Usk catchments the only catchments showing wide range between the upper and lower limits of the Whisker plot. All plots show a number of outliers in the calculated differences. The maximum absolute change in recharge values calculated by replacing LCM2000 by LCM2007 over England and Wales is 0.6 mm/day.

The land use impact on the calculated recharge values is also investigated by varying the percentage land use classes of the percentage LCM2000 data by replacing 50% of one class by another class at a time. Figure 17 shows the Whisker plots of the differences between the recharge values calculated from these runs and the run using the percentage LCM2000 for all the catchments. This figure indicates that changing the land use from grass to forest causes the most significant reduction in recharge. On average, the reduction in recharge values is 0.26 mm/day using the recharge values calculated over England and Wales. However, this figure also shows that on average replacing 50% of arable by grass causes more recharge than replacing 50% of forest by grass. This depends on the extent of the area covered by the different land use types.

On average the increase of recharge caused by replacing 50% of arable by grass is 0.022 mm/day but the maximum calculated increase in this case is 0.12 mm/day using the recharge values calculated over England and Wales. The increase in recharge values caused by replacing forest by grass is 0.012 mm/day but the maximum calculated increase in this case is 0.22 mm/day.

National maps of annual long term average recharge totals and differences, in mm/a simulated by ZOODRM (for various dominant land use runs, all assuming historic climate)

Figure 14    LTA recharge for changes to land use: LCM and single type coverage.

National maps of annual long term average recharge totals and differences, in mm/a simulated by ZOODRM (for various percentage land use runs, all assuming historic climate)

Figure 15    LTA recharge for changes to land use: like for like changes.

Figure 16    Plots of differences between recharge values calculated with selected dominant land use types and those calculated with actual dominant LCM 2000.

Figure 17    Plots of differences between recharge values calculated with selected variations in percentage LCM 2000 and those calculated with actual percentage LCM2000.