Research Note 2 15th May 2008
Adapting to the effects of climate change on water supply - a preliminary assessment M.B.CHARLTON* and N.W.ARNELL Walker Institute for Climate System Research, University of Reading
(Received 19th December 2007; in final form 14th May 2008)
Abstract In general terms, there are four types of barrier on the ability to adapt to climate change: physical, financial, socio-political, and institutional. This paper presents a preliminary exploration of these four barriers, and the factors which determine their characteristics, using a case study of water supply in southern England, where climate change is just one of several pressures on water resources. Simulations of water availability are conducted under different climate scenarios. Modelled climate change impact on river flow can be substantial depending on climate scenario. The results are used to approximate future resource availability for average, dry and wet scenarios and indicate the need for explicit consideration of climate change in adpatation strategies. Different adaptation strategies are then identified. There are very few physical constraints, and the key barriers are financial (specifically cost) and socio-political (specifically associated with different attitudes towards different adaptation options amongst different groups of stakeholders). The water supply sector in England has a framework and a practical methodology for addressing climate change, but there are some characteristics of the broader institutional context – specifically relating to the implied preferences for different types of actions over different time scales – that constrain ability to adapt to an uncertain climate change.
* Corresponding author. Email:
[email protected] MBCharlton.com Research Notes http://www.publications.mbcharlton.com/ResearchNotes.html
2
M.B.Charlton and N.W.Arnell
1. Introduction Climate change is expected to produce temperature increases, drier summers and wetter winters across southern England. Reductions in water availability are expected as a consequence (Arnell, 2004) with direct abstractions becoming less reliable during summer and more seasonal, higher intensity rainfall producing high runoff and less water able to percolate into aquifers (Environment Agency, 2005). In an area already facing water deficits and supply failures (Environment Agency, 2007), and with increasing population demands, adaptation in the short-term (to 2030) is necessary. In general, the water industry is well placed to adapt to changes in water resources but there is considerable uncertainty in the efficacy of a range of adaptation options in the context of uncertain climate change in the future. In general terms, there are four types of barrier on the ability to adapt to climate change: physical, financial, socio-political, and institutional. In this research we explore whether there are practical limits to our ability to adapt to climate change. Specifically, we investigate what are the physical, economic, social and political constraints on adaptation to future water shortage in southern England and we seek to construct and validate a conceptual model characterising the practical constraints on adaptation, which can be tested in other settings. The basic approach is to define a set of scenarios for future resource availability in a catchment, and ask of stakeholders in the water environment: if this were to happen, what could, should, or would, be done? This paper uses water resources management in the Medway Catchment in southern England as an empirical case study to help characterise factors influencing potential limits on adaptation without seeking to represent this area precisely or to produce a real adaptation plan. This paper is an early draft version of a recent conference paper (Arnell and Charlton, 2008) that has also been submitted for an associated edited volume. However, the current paper provides greater detail on the modelling procedure used and early results from this particular approach. It finishes with an outline of the research hypotheses / propoaitions. Arnell and Charlton (2008) provide a more detailed discussion of the conceptual framework developed in the research.
2. Methodology A conventional climate impact methodology was used to develop future resource scenarios coupled to an assessment of adaptation options using literature review in the following stages: 1. Simulate the effects of climate change scenarios on resource availability in the Medway catchment: a. simulate change in runoff / recharge using the Mac-PDM hydrological model outlined by Arnell (1999); modified for application to a single catchment using daily time series of input data (precipitation, temperature and potential evaporation). The model was optimised in a three-stage tuning process following Arnell (1999) using the observed flow data (from gauging station 40003 at Teston) for the period 1980-1983 and validated over the period 1984-1989. Perturbed input data were created using the UKCIP02 (Hulme et al., 2002) climate change scenarios (low, medium-low, medium-high, high scenarios for the 2020s, 2050s and 2080s) applied to the entire 1961-1990 time series. Model runs were also conducted using five alternative scenarios: ECHAM4/OPYC, CGCM2, CSIRO MKII, GFDC_R30, and CCSR/NIES2.
MBCharlton.com Research Note 2 15th May 2008
Adapting to the effects of climate change on water supply
3
b. Apply change in water availability to current estimates of deployable output (DO) to assess change in DO and the supply-demand balance. 2. Prepare narrative descriptions describing the change in the gap between deployable output and demand to the 2030s under three climate scenarios (UKCIP02 medium-low, and two scenarios representing the “dry” and “wet” extremes). This paper uses a demand projection from the Water Resources in the South East (WRSE, 2006) assessment. 3. Using a review of the available literature, realistic potential adaptation options for the study catchment are assessed in terms of advantages, disadvantages and barriers to each option.
3. Impact of climate change on Medway flow 3.1. Parameterisation and validation Before application the model was optimised and validated (see Table 1). In the first stage (L1), the cropland land cover class was selected with a non-grass percentage cover of 26.1% and the predominant loam–clay soil class was selected (see Arnell (1999) for initial parameter values). Next (L2), the soil parameters were optimised by generating parameter sets consisting of combinations of values of field capacity (57.93%), saturation capacity (88.20%) and a parameter describing the distribution of soil moisture capacity (b: 0.13) between a range of +/- 75% of the initial value at 25% increments, producing 245 parameter combinations (once all instances of field capacity exceeding saturation capacity are removed). This was followed by an additional 49 runs (L3) to optimise the flow routing parameters (grout and srout).
Run Type
Calibration Validation Baseline
Stage
L1 L2 / L3 a b
Average Annual Flow (mma-1) Daily Estimated Observed Difference Percentage NashChange Sutcliffe 387.30 351.60 310.66 364.00 351.08
280.56 280.56 278.40 259.02 259.02
106.74 71.04 32.26 104.98 92.06
38.05 25.32 11.59 40.53 35.54
0.35 0.39 0.44 0.38 0.42
r
Monthly NashSutcliffe
r
0.69 0.66 0.66 0.67 0.69
0.41 0.42 0.68 0.44 0.54
0.92 0.87 0.86 0.84 0.87
Table 1. Comparison of the observed and estimated average annual flow results for the optimisation, validation, and baseline model runs. The Nash-Sutcliffe efficiency and r values are also indicated. 1
Table 1 shows that optimisation of the flow routing parameters resulted in limited improvement in either estimated Qa or NS so the L2 parameters (listed above) were used for the remainder of the modelling. After optimisation, the model overestimates average annual flow (Qa) values by 71.04 mma-1 and model fit is poor, as indicated by the Nash-Sutcliffe criterion (NS) and associated r 1
Note: For the calibration runs Level 3 (L3) tuning are not shown because there was little improvement over the Level 2 results shown. Baseline (a) refers to comparison between estimated flow and observed flow data th with the leap year Feb 29 stripped out. Baseline (b) is as for (a) except that the missing data values are stripped from both the observed and estimated daily flow results.
MBCharlton.com Research Note 2 15th May 2008
4
M.B.Charlton and N.W.Arnell
value. The model is relatively sensitive to the soil parameters and consequently model estimations are improved considerably during the optimisation. Table 1 also shows the results of the validation exercise. Again, although the NS value is poor for the daily values, these results are encouraging with the chosen parameter set providing a Qa estimate 32.26 mma-1 above the observed results. Furthermore, there is improvement in monthly NS and r values.
3.2. Simulation of baseline conditions Table 2 shows the result of running the optimised model on the baseline data. Qa is overestimated at 364 mma-1. Some of this error can be accounted for by leap year values (7) and missing values (256) in the observed record, which, if excluded, reduce the magnitude of overestimation and improve the model fit. Although there is a large discrepancy between estimated and observed flows (>100 mma-1) the estimated value is consistent with earlier studies (e.g. Arnell, 1999) conducted over much larger geographic domains. This offers the opportunity to investigate indicative change against a baseline using perturbed climate inputs.
Year
UKCIP02 Scenario
1980 2020
Baseline Low Medium-Low Medium-High High Low Medium-Low Medium-High High Low Medium-Low Medium-High High
2050
2080
Qa estimated Difference from baseline % Difference from baseline -1 -1 (mma ) (mma ) 364.00 0.00 0.00 324.70 39.30 -10.80 319.90 44.10 -12.12 319.90 44.10 -12.12 317.10 46.90 -12.88 295.20 68.80 -18.90 283.60 80.40 -22.09 275.30 88.70 -24.37 260.10 103.90 -28.54 269.90 94.10 -25.85 256.10 107.90 -29.64 221.30 142.70 -39.20 202.90 161.10 -44.26
Table 2a. Estimated average annual flow results for UKCIP02 climate perturbations.
3.3. Impact of future climate change on average annual flow Table 2 shows the estimated flow and the change relative to the baseline for each of the scenarios. Figures 1 and 2 show modelled percentage change in Qa relative to the baseline for the UKCIP02 and DDC scenarios, respectively. For UKCIP02, there are two main patterns: 1. Flow is reduced the further into the future we go. The rate of this reduction generally decreases with increased time from baseline. 2. Flow is reduced as we move from the low to high scenarios. The magnitude of this difference increases with increased time from the baseline scenario (reflecting increased uncertainty in climate predictions in the future).
MBCharlton.com Research Note 2 15th May 2008
Adapting to the effects of climate change on water supply Year
UKCIP02 Scenario
1980 2020
Baseline ECHAM4/OPYC CGCM2 CSIRO MKII GFDL_R30 CCSR/NIES2 ECHAM4/OPYC CGCM2 CSIRO MKII GFDL_R30 CCSR/NIES2 ECHAM4/OPYC CGCM2 CSIRO MKII GFDL_R30 CCSR/NIES2
2050
2080
5
Qa estimated Difference from baseline % Difference from baseline -1 -1 (mma ) (mma ) 364.00 0.00 0.00 299.50 64.50 -17.72 373.60 -9.60 2.64 372.20 -8.20 2.25 346.90 17.10 -4.70 415.00 -51.00 14.01 306.60 57.40 -15.77 343.70 20.30 -5.58 379.30 -15.30 4.20 346.40 17.60 -4.84 513.50 -149.50 41.07 276.80 87.20 -23.96 342.30 21.70 -5.96 456.20 -92.20 25.33 294.10 69.90 -19.20 559.10 -195.10 53.60
Table 2b. Estimated average annual flow results for DDC climate perturbations.
Given the marginal nature of water resources in the catchment, even the estimated value for the 2020s low scenario (11% lower than baseline) is likely to result in a significant impact on available water resources although the range between 2020s scenarios is only about 2%. The DDC scenarios demonstrate a range of impacts including substantial increases and decreases in Qa reflecting sensitivity to climate model uncertainty.
0 -10 -20 -30 -40 -50 1980
2000 Low
2020
2040
Medium-Low
2060 Medium-High
2080
2100
High
Figure 1. Percentage change in average annual flow relative to baseline for each UKCIP02 scenario.
MBCharlton.com Research Note 2 15th May 2008
6
M.B.Charlton and N.W.Arnell 60 40 20 0 -20 -40 1980
2000
2020
2040
ECHAM4/OPYC
CGCM2
GFDL_R30
CCSR/NIES2
2060
2080
2100
CSIRO MKII
Figure 2. Percentage change in average annual flow relative to baseline for each DDC scenario.
4. Future Resource Availability Scenarios Using this information and incorporating the demand projection three scenarios have been constructed for the Medway catchment. Using values from Campaign to Protect Rural England's Water Resource Strategy for Kent (CPRE, 2006) unless otherwise specified, DO for 2030 is calculated as a first approximation by: 1. taking the current (2006) deployable drought output (765 Ml/d) and subtracting a reduction enabling meeting Water Framework and Habitat Directive targets (50 Ml/d). 2. then subtract the current surface DO (estimated as 25% of current DO because only about 25% of the supply is from surface water) multiplied by the percentage change in Qa from baseline for each scenario. The total forecast demand for Kent to 2026 taken from CPRE (2006) is 805 Ml/d, which incorporates the current 'critical period' 1-in-10 dry-year demand (725 Ml/d), public supply growth including headroom (49 Ml/d), and demand growth arising from climate change (23 Ml/d) and agricultural irrigation (8 Ml/d). The supply-demand balance is then calculated by subtracting demand from the DO estimation. The resulting water availability scenarios are as follows: Average scenario: the UKCIP02 medium-low scenario indicates a 12.12% reduction in available surface water (river flow) by 2030 relative to the baseline. This indicates a critical dry year DO of 691.82 Ml/d and a deficit of 113.18 Ml/d (reduced to 63.18 Ml/d if the Directive targets are excluded). Thus a substantial regional deficit exists as a direct result of climate change and population growth by 2030. Dry scenario: The ECHAM4/OPYC 2020 scenario produces the largest reduction in river flow: 17.72% relative to the baseline estimation. This indicates a critical dry year DO of 681.11 Ml/d and a deficit of 123.89 Ml/d (reduced to 73. 98 Ml/d if the Directive targets are excluded). Thus under MBCharlton.com Research Note 2 15th May 2008
Adapting to the effects of climate change on water supply
7
the most extreme drying scenario modelled for the 2030s a substantial regional deficit exists as a direct result of climate change and population growth by 2030. Wet scenario: the CCSR/NIES2 2020 scenario contradicts the impacts suggested by the average and dry scenarios indicating a 14.01% increase in river flows producing a critical dry year DO of 741.79 Ml/d . Despite these flow increases, a relatively large deficit of 63.21 Ml/d is projected to 2030 although the pressure on resources as a result of meeting Directive targets is more important in this scenario (the deficit is reduced to 13.21 Ml/d if the targets are excluded). These results show that even under conditions of increased river flows, the Kent region will be in deficit by 2030 as a result of a number of pressures of which climate change is an important component. The results should be interpreted with caution because the approach outlined above is only an indicative first approximation and does not pertain to present actual values. The validity of applying indicative flow changes from the Medway to the whole of Kent is not realistic given the considerable variability in catchment characteristics in the region. The uncertainty arising from different climate scenarios is also worthy of further attention, in addition to the uncertainty arising from the application of the hydrological model and the assumptions inherent in the current calculation of the supply-demand balance (including no change in ground water sources).
5. Adaptation Options The modelling exercise, whilst only a first approximation, indicates that to maintain water supply, adaptation strategies need to make explicit consideration of the impact of climate change (although it is only one of a multitude of resource pressures). Table 3 details a range of specific resource schemes and more generic options proposed for securing water supply for the Medway catchment and the Kent region arranged in concordance with the twin-track approach, derived from water company water resources management plans (WRMPs) produced during the last planning period, company web sites, an Environment Agency (2004a) summary document and the Medway Catchment Abstraction Management Strategy (Environment Agency, 2005), which details catchment-specific means of maintaining security of supply. The complex responsibility for water resources in the catchment means it is necessary to consider schemes across the Kent region. Many of these options have been incorporated into the plans of the four water companies responsible for water resources in the Medway catchment: Mid Kent Water (MKW), Southern Water (SW), South East Water (SEW) and Sutton and East Sussex Water (SESW). Mid Kent Water and South East Water have recently merged. As part of the next planning phase, water companies are scheduled to release new draft WRMPs in April and May 2008. This will necessitate a re-evaluation of the adaptation options as presented here and in Arnell and Charlton (2008). It is noticeable from Table 3 that supply-side schemes are more specific, whilst those on the demand-side are largely generic. The South East England Regional Assembly (SEERA, 2005) reiterate that new water resources are likely to include a range of measures, including new transfers and pipelines, new or enlarged reservoirs, and technologies new to the UK, potentially involving de-salination plants and effluent re-use. However, the Environment Agency (2004a and 2004b) has criticised companies for being too reliant on developing infrastructure and not sufficiently considering demand-side schemes or shared strategic resources. A criticism shared by the CPRE. CPRE (Warren, 2007) propose an adaptation approach that relies less heavily on such strategic MBCharlton.com Research Note 2 15th May 2008
8
M.B.Charlton and N.W.Arnell
resource development, consisting of a three-part strategy based on water efficiency measures, waste water re-use and raw water transfers. Demand-side options include a range of water efficiency schemes in addition to new technologies, education and revised licensing approaches.
Adaptation option
New Reservoirs Reservoir Enlargement De-Salination Flood Storage Reservoirs Winter Storage Groundwater Developments
Surface Water Developments
Aquifer Storage and Recovery Indirect waste water reuse / Effluent Re-use
Bulk transfers from outside region
Increased connectivity of resources in region / sharing supplies
Altered operating procedures for existing schemes
Specific Example / Scheme
Gains in supply or savings in demand 2 (Ml/d) 40 18 20 -
Implementation date
Cost
2019/20 2014/15 2014/15 AMP4 -
-
-
1. Maidstone / Tonbridge 2. Enhance outputs 3. Two new schemes 4. Further enhancement in northern zone 5. Further enhancement in Kent 6. New scheme 7. Phased reduction in Deployable Output for EWFD and HD compliance 1. Increased abstraction from Medway 2. Increased abstraction from Thames 3. New abstraction in Sussex North RZ 4. Increased abstraction from Bough Beech Reservoir st 1. 1 phase 2. Two further phases 1. Kent Recycling Scheme 2. Re-use of water from Margate / Broadstairs scheme 3. River Medway Water Improvements Links from sources in north Wales, Midlands, North East - use London Ring Main for knock on transfer into Kent. 1. Bewl Water to Ashford Trunk Main
-
2011-2026 2006 Beyond 2010
-
9 50 (reduction in supply) -
2010 2010-2014
-
2012 2008 2010 2021
-
5 up to 20
2010 2027 -
-
-
-
£20M -
6.3
2030
2. Transfer main from Canterbury into Ashford 3. Transfer from Burham zone into Stansted zone 4. Transfer from Bewl to Darwell 5. Exedown to Otford 6. East Peckham to Pembury 7. Bewl / Kippin's Cross to Pembury 8. Bewl to Best Beech MKW SEW merger
4
2010
£24.4 M -
5.2
2030
-
-
AMP4 AMP4 AMP4 AMP4
-
Broadoak Clay Hill Bewl Newhaven Upper Eden, Upper Medway, Beult
-
Table 3a: Supply Side Adaptation Options for Water Companies covering The River Medway Catchment
2
Gains in supply are for the option and not specifically for the catchment. Ml/d unless otherwise specified.
MBCharlton.com Research Note 2 15th May 2008
Adapting to the effects of climate change on water supply Adaptation option
Reduce leakage
Tariff and metering structures to reduce per capita water usage
Increased use of water-efficient devices in existing properties Increased use of water-efficient devices in new development Public education
Increase use of restrictions during drought conditions Tradeable permits Abstraction Licenses
Increased use of grey water/reuse in existing properties Increased use of grey water/reuse in new development Reduce rate of new development
Specific Example / Scheme
9
Gains in supply or savings in demand 3 (Ml/d) 0 29 ~ 25 l/h/d (10-12 %)
2025/26 2025/26 2030
-
Projected penetration: MKW: 75% SEW: 70% SESW:
