40
General enquiries on this form should be made to: Defra, Science Directorate, Management Support and Finance Team, Telephone No. 020 7238 1612 E-mail: [email protected] SID 5 Research Project Final Report SID 5 (2/05) Page 1 of 40

General enquiries on this form should be made to:randd.defra.gov.uk/Document.aspx?Document=CC0375_3746... · Web viewCENTURY has thus been used as the primary model for simulating

Embed Size (px)

Citation preview

General enquiries on this form should be made to:Defra, Science Directorate, Management Support and Finance Team,Telephone No. 020 7238 1612E-mail: [email protected]

SID 5 Research Project Final Report

SID 5 (2/05) Page 1 of 30

NoteIn line with the Freedom of Information Act 2000, Defra aims to place the results of its completed research projects in the public domain wherever possible. The SID 5 (Research Project Final Report) is designed to capture the information on the results and outputs of Defra-funded research in a format that is easily publishable through the Defra website. A SID 5 must be completed for all projects.

A SID 5A form must be completed where a project is paid on a monthly basis or against quarterly invoices. No SID 5A is required where payments are made at milestone points. When a SID 5A is required, no SID 5 form will be accepted without the accompanying SID 5A.

This form is in Word format and the boxes may be expanded or reduced, as appropriate.

ACCESS TO INFORMATIONThe information collected on this form will be stored electronically and may be sent to any part of Defra, or to individual researchers or organisations outside Defra for the purposes of reviewing the project. Defra may also disclose the information to any outside organisation acting as an agent authorised by Defra to process final research reports on its behalf. Defra intends to publish this form on its website, unless there are strong reasons not to, which fully comply with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.Defra may be required to release information, including personal data and commercial information, on request under the Environmental Information Regulations or the Freedom of Information Act 2000. However, Defra will not permit any unwarranted breach of confidentiality or act in contravention of its obligations under the Data Protection Act 1998. Defra or its appointed agents may use the name, address or other details on your form to contact you in connection with occasional customer research aimed at improving the processes through which Defra works with its contractors.

Project identification

1. Defra Project code CC0375

2. Project title

The development of UK soil properties datasets and tools for supporting climate change impact studies

3. Contractororganisation(s)

National Soil Resources Institute, Cranfield University                         

54. Total Defra project costs £ 123,095 - 00

5. Project: start date................ 01 July 2003

end date................. 13 May 2005

SID 5 (2/05) Page 2 of 30

6. It is Defra’s intention to publish this form. Please confirm your agreement to do so...................................................................................YES NO (a) When preparing SID 5s contractors should bear in mind that Defra intends that they be made public. They

should be written in a clear and concise manner and represent a full account of the research project which someone not closely associated with the project can follow.Defra recognises that in a small minority of cases there may be information, such as intellectual property or commercially confidential data, used in or generated by the research project, which should not be disclosed. In these cases, such information should be detailed in a separate annex (not to be published) so that the SID 5 can be placed in the public domain. Where it is impossible to complete the Final Report without including references to any sensitive or confidential data, the information should be included and section (b) completed. NB: only in exceptional circumstances will Defra expect contractors to give a "No" answer.In all cases, reasons for withholding information must be fully in line with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.

(b) If you have answered NO, please explain why the Final report should not be released into public domain

Executive Summary7. The executive summary must not exceed 2 sides in total of A4 and should be understandable to the

intelligent non-scientist. It should cover the main objectives, methods and findings of the research, together with any other significant events and options for new work.Understanding the likely impact of climate change on crop yield, land use, biodiversity, water resources and quality is a key to developing suitable policies for promoting sustainable agriculture in the future. Modelling plays and important role in improving such understanding. Most models which simulate crop yield, land use, biodiversity, water resources, water quality and fluvial flooding are sensitive to soil chemical and physical property data. It is thus essential that the best or most appropriate soil datasets are used within integrated and fine-resolution modelling, if robust and defensible predictions are to be made. However, the existing measured soil property data that is available for creating national databases to feed these models largely reflect soil conditions up to about 1990. It is questionable if it is appropriate to continue using such data in impact studies for the 2020’s 2050’s and 2080’s because of likely climate-induced changes in soil organic carbon content, soil structure and wetness. The main aims of this research have been to:1. Develop databases of soil properties that characterise soil conditions under arable, managed

grassland, semi-natural grassland and woodland ecosystems across England and Wales as they

All the relevant existing measured soil data in the Land Information System (LandIS) for England and Wales have been used, together with models of soil wetness and organic carbon turnover driven by the new UKCIP02 Climate Scenario datasets, to predict future changes in soil organic carbon content, water regime and associated hydraulic and hydrologic properties under arable, managed permanent grassland, semi-natural grassland and woodland land uses. The CENTURY model was the main tool used to predict climate change-induced changes in organic carbon content of soil horizons. Its predictive ability was tested for a limited range of representative ‘indicator’ soil types under each ecosystem, using measured changes in organic carbon content for from the National Soil Inventory for England & Wales. The results suggest that, for all ecosystems except deciduous woodland its predictions give an acceptable, if not good, fit to the measured changes. Future changes in soil properties such as bulk density and hydraulic characteristics are driven principally by changes in soil organic carbon content and modelling predictions suggest that changes in soil organic carbon and associated hydraulic properties are likely to be greatest in semi-natural grass and woodland soils under the ‘High emissions’ scenario. Predicted patterns of change for all other scenario / land use combinations are not significantly different.

Under arable land use, whereas the measured changes in soil organic carbon from the National Soil Inventory show a general decrease, except where levels are already low, the climate change driven models predict only a brief continuation of this trend before organic carbon contents stabilise or even increase slightly, possibly because of a positive feed-back from increased yields and resulting crop residues. In contrast, for soils under permanent managed or semi-natural grassland, the climate models

SID 5 (2/05) Page 3 of 30

predict that the overall reductions measured from the National Soil Inventory are likely to continue into the future, albeit at a slightly reducing rate. The picture for woodland is uncertain because the models perform least well for this land use and further work is needed to clarify this situation.

Changes in soil water regimes (wetness class) and the Hydrology Of Soil Types (HOST) class are driven by climate-induced changes in soil storage capacity and by climate changes to the patterns of cumulative potential soil moisture deficit (PSMD), the duration of the field capacity period (when PSMD is zero) and the total amount of rainfall during this period. Modelled changes to the climatic monthly water balance suggest that by far the biggest impact of climate change will be to significantly reduce the duration of the field capacity period, with the largest impacts occurring earliest in the English midlands and in areas fringing the highest areas of Wales, south west and north west England. These changes are likely to affect soil moisture regimes in two significant ways. Free draining, relatively permeable soils (wetness class I) in lowland England and Wales will become significantly drier and vegetation growing on them is more likely to experience drought effects. In wetter upland areas, free-draining soils with dark, relatively acid, organic-rich upper layers associated with cool, humid climates (recognised at a global level as Umbrisols ) will be most sensitive to reductions in the field capacity period, which are likely to significantly increase their soil organic matter turnover leading to reductions in their organic carbon content.

In contrast, soils which are wet within 70 cm depth for more than 30 days in most years (soil wetness classes II to VI) are likely to experience significant reductions in their duration of waterlogging within 40 cm depth. Upland ‘climatic’ peaty soils are likely to be most affected, apart from in the core mountain areas of Snowdonia and the Lake District, with up to 40% likely to change from wetness class VI to wetness class V because of a decrease in shallow waterlogging period of at least one month. The other main changes will be associated with slowly permeable, seasonally wet soils (wetness classes IV & V), which occupy approximately 39% of the agricultural land in England & Wales. Overall, up to 45 % of wetness class IV soils and up to 70% of wetness class V soils are predicted to reduce by one wetness class by the 2050’s, depending on the scenario.

The overall impact of these predicted changes in soil water regimes will be to change HOST classes in the slowly permeable and impermeable substrate models, particularly where the soil drainable porosity is relatively large. Depending on the predicted changes in duration of waterlogging within 40 cm depth, HOST classes 24 and 25 are most likely to change to classes 18 and 20. In addition, peaty or humose topped upland soils over more permeable substrates (HOST class 15) are likely to become significantly drier and their peaty or humose nature will become lost. A significant proportion of such soils are most likely to change to HOST classes 17 and 4, depending on the nature of their substrates.

The integrated database and software tool resulting from the project is available for use by the modelling community.

Implications of the workThe results of the modelling carried out to date suggest that the principal impacts of climate change on soil organic carbon and associated hydraulic properties will occur in semi-natural grassland and managed grassland where the measured declining trend in organic carbon content is predicted to continue into the future. The implication of this is that policy directed to carbon sequestration within soils is likely to be most effective if focused on mitigating soil carbon losses in managed and semi-natural grassland, rather than arable ecosystems.

The work has also highlighted the importance of taking into account the impact of potential changes in soil water regimes. In seasonally wet soils under arable & managed grassland effective field drainage increases organic carbon turnover and causes more rapid reduction of topsoil organic carbon. If field drainage becomes less effective in the future, this may reduce or even reverse the trend of declining organic carbon in such soils under grassland, although this may be more than offset by climate-induced reductions in their duration of waterlogging within 40 cm depth, at least in slowly permeable soils with perched water tables. However, the main impact of water regime changes in such soils will be to increase the period when they are accessible to stock or machinery, without causing significant soil structural damage. In the uplands, apart from in the core mountain areas of Snowdonia and the Lake District, the impact of climate change on wet moorland soils appears likely to give significantly reduced duration of waterlogging in the topsoil with continuing significant reduction in organic carbon content. In lowland wet heaths and bogs however, where waterlogged conditions are mainly controlled by non-climatic landscape or management factors there appears to be significant potential for controlling future organic carbon changes by maintaining high soil water tables.

Project Report to Defra8. As a guide this report should be no longer than 20 sides of A4. This report is to provide Defra with

details of the outputs of the research project for internal purposes; to meet the terms of the contract; and to allow Defra to publish details of the outputs to meet Environmental Information Regulation or

SID 5 (2/05) Page 4 of 30

Freedom of Information obligations. This short report to Defra does not preclude contractors from also seeking to publish a full, formal scientific report/paper in an appropriate scientific or other journal/publication. Indeed, Defra actively encourages such publications as part of the contract terms. The report to Defra should include: the scientific objectives as set out in the contract; the extent to which the objectives set out in the contract have been met; details of methods used and the results obtained, including statistical analysis (if appropriate); a discussion of the results and their reliability; the main implications of the findings; possible future work; and any action resulting from the research (e.g. IP, Knowledge Transfer).

1. Background to the ProjectThe spatial resolution of climate change impact studies has become progressively finer as computing capabilities and models improve. The demand from many stakeholders is for even finer spatial resolution predictions at the level of, for example, an individual nature reserve, and this demand for increased resolution is expected to continue. Partly in response to this demand, new UKCIP Climate Change Scenarios were published in April 2002. These scenarios include monthly average weather datasets for a wide range of variables, interpolated to a 5km x 5km grid. A monthly time series of weather data for the period 2011 to 2100 is also available at 5km x 5km resolution.

Most models which simulate crop yield, land use, biodiversity, water resources, water quality and fluvial flooding are sensitive to soil chemical and physical property data. It is thus essential that the best or most appropriate soil datasets are used within integrated and fine-resolution modelling, if robust and defensible predictions are to be made. The existing measured soil property data that is available for creating national databases to feed these models largely reflect soil conditions up to about 1990. It is questionable if it is appropriate to continue using such data in impact studies for the 2020’s 2050’s and 2080’s because of likely climate-induced changes in soil organic carbon content, soil structure and wetness. In addition, the range of models available for simulating differing responses to climate change scenarios usually require soil property data that is integrated over specified depths and spatial resolutions. Unfortunately the depth and spatial resolution of such integration is different for different models, so model-specific soil datasets at different vertical and horizontal resolutions are required. Deriving such datasets on an ad hoc, project-specific basis can be very time-consuming and is not an effective use of limited resources.

The main aim of this project has been to develop an integrated database and interrogative tool for deriving user-specified spatial soil property data for England and Wales that is relevant to the latest UKCIP02 climate change scenarios for the 2020’s, 2050’s and 2080’s. The resulting integrated database and software tool can then be used by the modelling community to derive state-of-the-art soil parameter input data for climate change impact studies based on these scenarios.

2. Scientific ObjectivesThis project had five main scientific objectives:1. To develop a computer-based software tool for deriving user-specified spatial soil input parameters by

integrating existing data on land-use-soil series specific layer properties, spatial distribution of soil series and spatial distribution of land use categories within England and Wales. The methodology implemented in the software will be generic and driven through user-friendly menus.

2. To estimate climate change-induced changes in land use-specific soil horizon organic carbon for the UKCIP02 scenarios for the 2020’s, 2050’s and 2080’s for all soil series in the 1:250,000 scale National Soil Map database. Estimated changes will be derived using a combination of soil organic matter dynamics modelling and existing measured data on changes in soil organic matter over time.

3. To estimate climate change-induced changes in soil wetness class and Hydrology Of Soil Types (HOST) class for all soil series in the 1:250,000 scale National Soil Map database. Estimated changes will be derived using changes in the driving climatic variables computed from the new UKCIP Climate Change Scenarios for the 2020’s, 2050’s and 2080’s.

4. To derive three climate change scenario-specific soil property datasets for all soil series in the 1:250,000 scale National Soil Map database and for four broad land use categories. Datasets will be derived using the results from objectives 2 and 3 together with established pedo-transfer functions for bulk density, soil water retention and saturated hydraulic conductivity

5. To test and de-bug the software tool developed as a result of objective 1 using the datasets developed as a result of objective 4.

The following sections of this report describe how and to what extent these objectives have been achieved and discuss the results in terms of their reliability, implications for policy development and possible future work.

SID 5 (2/05) Page 5 of 30

3. Estimation of climate-change induced changes in land use-specific soil horizon organic carbonChanges in soil organic carbon were estimated using different modelling approaches, depending on whether the simulations were for mineral soils, drained organic soils or undrained organic soils. Current soil organic carbon cycling models can be grouped into two broad categories, ‘ecosystem-level’ models and ‘macro-scale’ models (Paustian et al 1997). The latter have been developed primarily for global-scale or large regional-scale modelling purposes and are thus unsuitable for application to the land use-soil series specific approach required for this project. Most ecosystem-level models have been originally developed and calibrated for specific land use types and usually for free-draining mineral soils.

RothC (Coleman & Jenkinson, 1995), the model used in DEFRA project CC0242, has been validated for a number of existing arable ecosystems in the UK as well as some grassland ecosystems and one forest ecosystem in a cool temperate climate (Smith et al, 1996). However, the model has only been applied to free-draining mineral soils and does not at present have the ability to simulate the impact of increased atmospheric CO2 levels on plant biomass dynamics. The CENTURY model (Metherell et al, 1993; Parton et al, 1994), does have the facility to include the effects of enriched atmospheric CO2 levels on net primary production of biomass through simulating CO2-induced changes to relative production, reduced transpiration, changed carbon to element ratios (e.g. C/N ratio) and changed distribution of C production between roots and shoots. Potentially therefore it is more suitable than RothC for modelling future climate change-induced soil organic carbon changes associated with the new UKCIP climate change scenarios for 2020’s, 2050’s and 2080’s. However the model was originally developed to simulate organic matter dynamics in grassland soils in the USA and has had little application in the UK. Nevertheless it has now been modified to expand its use to a range of agricultural ecosystems (Paustian et al, 1992, Metherell et al, 1995) and a limited number of forest ecosystems (Sandford et al, 1991) and it also has the ability to take into account organic matter dynamics in saturated conditions resulting from ground- or perched water tables and it can also vary the organic matter dynamics across a range of aerobic to anaerobic conditions. This facility has been used to successfully simulate bulk organic matter changes in natural peat bog soil in the US (Chimner et al, 2002). CENTURY has thus been used as the primary model for simulating climate-change induced changes in soil horizon organic carbon for mineral soils under arable, managed grassland, semi-natural grassland and woodland. It has also been used to investigate its potential for simulating changes in organic carbon content of undrained peaty soils. In addition, RothC has been tested for its suitability for in simulating climate-change induced changes in soil horizon organic carbon for mineral soils under arable ecosystems.

For all drained peaty soil types, climate-change induced changes in soil horizon organic carbon have been estimated using an empirical decay-curve model developed as a result of MAFF project LE0203 (Burton, 1995). This approach assumes that the principal organic carbon losses from such drained soils will be from continuing oxidation of organic matter resulting from removal of the original waterlogged conditions and that, although climate change will affect the rate of oxidation, such impacts are likely to be well within the range of uncertainty of the decay curve model.

3.1 Data SourcesSoil variables required by the models were derived from two sources.

Topsoil organic carbon contents for model calibration and evaluation were derived from the National Soil Inventory (Loveland, 1990). This Inventory was made to obtain an unbiased estimate of the distribution of the soils of England and Wales and of the chemistry of the topsoil (0-15 cm). Between 1978 and 1984, samples were collected and soil profiles described at the intersections of an orthogonal 5-km grid over the whole area. This yielded about 6,000 sites of which 5,662 could be sampled for soil. Sufficient sub-sets of the sites were re-sampled at intervals from 12 to 25 years after the original sampling to be able to detect changes in carbon content with 95% confidence. This was done in three phases: in 1994/95 for arable and rotational grassland sites (853 of the original 2,851 sites), in 1995/96 for managed permanent grassland sites (771 of the original 1,559), and in 2003 for non-agricultural sites (bogs, scrub, rough grazing, woodland, etc; (555 of the original 1,281). Approximately 40% of the original sites were re-sampled.

Soil series layer properties for driving the predictive models were derived from data held in the Land Information System, LandIS (Proctor et al, 1999). Data defining the particle-size distribution (texture), organic carbon content, pH, bulk density and hydraulic characteristics of each soil series for each significantly different soil layer down to a depth of 1.5 m or bedrock (whichever is shallower) and under each of four different land uses: arable; ley (short-term ‘rotational’) grass; permanent grass; other (all semi-natural vegetation) has been derived from a variety of sources as described by Hollis et al, 1995 and Hollis, 2004.

Climate variables were also derived from two sources. For model calibration and evaluation, the monthly values of precipitation and temperature required by the models were derived from the UKCIP ‘Baseline’ (1960 – 2000) 5 km grid datasets produced by the Meteorological Office. For future change predictions, the required precipitation and temperature variables were derived from the UKCIP Climate Change Scenarios for the 2020’s, 2050’s and 2080’s published in April 2002 (Hulme et al, 2002). These scenarios provide predictive climate and weather data for the 2020’s, 2050’s and 2080’s at 50km x 50km grid cell resolution. Associated Climate scenarios include monthly average weather datasets for a wide range of variables, interpolated to a 5km x 5km grid and a monthly time series of precipitation and temperature data for the period 2011 to 2100 again available at 5km x 5km

SID 5 (2/05) Page 6 of 30

resolution. Such datasets, which are based on four different CO2 emission scenarios, A1F1, A2, B1 and B2, represent the best expert estimate of future change in climate across England and Wales.

3.2 Mineral soilsThe methodology used to simulate climate-change induced changes in soil organic carbon for mineral soils was as follows:

Step 1 Develop generic management scenarios for arable and managed grassland ecosystems and select appropriate generic ecosystem model variables for semi-natural grassland and woodland ecosystems.

Step 2 Calibrate and evaluate models for each of the selected management/ecosystem combinations.

Step 3 Generate model input data files for the selected model as illustrated in figure 3.2-1.

Step 4 Run model using the monthly time series 2011 to 2100 data to generate climate change-induced organic carbon endpoints for each relevant soil series-management/ecosystem-climate change scenario in each region as illustrated in figure 3.2-2.

Figure 3.2-1 Methodology for generating input data files for CENTURY.

SID 5 (2/05) Page 7 of 30

Figure 3.2-2 Structure of modelled output data for estimated climate change scenario-specific soil organic carbon contents.

Generic management scenarios for arable ecosystems were based on those derived for Defra project SP0523: ‘effect of management changes on soil organic carbon’. Two basic management scenarios were adopted: an ‘intensive arable’ scenario defining management related to a sugar beet, winter wheat and winter barley crop rotation; a ‘less intensive arable’ scenario defining management related to an oilseed rape, winter cereal, and winter beans rotation. Fertilizer inputs, cultivation and harvest data related to these scenarios were combined with atmospheric nitrogen deposition data to characterise each scenario.

For managed grassland ecosystems, a generic management scenario was derived based on an ‘intensive’ grazing regime. CENTURY model parameters relating to a temperate grass clover pasture were selected combined with a management sequence of grazing option GM (low intensity moderate effect on production) in months 4 and 10 and grazing option GH (high intensity quadratic effect on production) in months 6 and 8. Inputs were selected according to a high stocking density of dairy cows with a moderate soil N supply status. Added N was variable depending on the month, whereas P was applied in one spread in early spring (20 kg/ha).

For semi-natural grassland ecosystems CENTURY model parameters relating to C3 grass species were selected and the only inputs, other than via litter were for atmospheric nitrogen deposition. Finally, for woodland simulations, CENTURY model parameters relating to those for the ‘Cowetta Hydrologic Lab, eastern deciduous forest’ and the ‘H.J. Andrews Experimental Forest, temperate coniferous forest’ both from the USA were selected to represent deciduous and coniferous woodland soils in England & Wales, although some parameters were altered to match data held by Forest Research.

The calibration and evaluation step of the methodology was carried out as follows:

Step 1 Select a range of up to 10 ‘indicator’ soil types representative of the range of soils under each of the selected management/ecosystem combinations.

Step 2 Using data from the original National Soil Inventory sampling, derive a mean and standard deviation value of topsoil organic carbon content for each indicator soil type. This represents organic carbon contents of the soils for the period 1978 – 1984.

Step 3 Use the derived mean organic carbon content for each indicator soil type as input to models and fractionate the carbon according to literature values (For CENTURY this was based on Kelly et al, 1990, for RothC it was based on Coleman & Jenkinson, 1995).

Step 4 Using this original fractionation, run the models for 2000 years, with the mean monthly climatic variables for the 1978 – 2000 climate, derived from the UKCIP ‘Baseline’ (1960 – 2000) 5 km grid datasets produced by the Meteorological Office. For each variable, a mean value across all grids in England & Wales was used. Derive the end-point fractionation values.

Step 5 Re-run models, again using the 1978 - 2000 climate data, using the calibrated (end-point) fractionation and compare the predicted organic carbon content with measured values from the National Soil Inventory re-sampling data representing organic carbon contents under arable in 1994-1995, under managed grassland in 1996-1997 and under semi-natural grass and woodland in 2002 – 2003.

SID 5 (2/05) Page 8 of 30

The range of mineral soil types selected as representative of England and Wales is described below and the selected indicator soil types used for model calibration are summarized in table 3.2-1.

Soil type Description1 Shallow free-draining soils over chalk or soft limestone.2 Free draining loams or sands: a) calcareous loams; b) non-calcareous loams; c) non-calcareous

sands; d) acid upland loams with an organic-mineral topsoil3 Calcareous ‘cracking’ clays.4 Podzols: a) free-draining; b) wet lowland heath.5 Slowly permeable, seasonally waterlogged soils a) with mineral topsoil; b) with organic-mineral or

peaty topsoil.6 Shallow acid upland soils over hard rock and with a peaty topsoil

Table 3.2-1 Representative indicator soil types used to calibrate the carbon dynamics models

Management/ecosystem combination Soil type Indicator soils

‘Intensive’ arable 2b2c35a

Arrow; HopsfordNewportHanslopeBeccles

‘Less intensive arable 12a35a

Andover; ElmtonPanholesEveshamDunkeswick; Wickham

Managed grassland 2b5a5b

Denbigh; Manod; RivingtonBrickfield; Cegin; Clifton; Dunkeswick; Salop; WickhamWilcocks

Semi-natural grassland 2b2d4a4b6

5.41; 6.11;6.126.316.4*LH3.11

Deciduous woodland 12a2b2c5a

3.4*5.115,41; 5.71; 5.81/25.51/527.11

Coniferous woodland 2b2c4a5a

5.41; 5.71; 6.115.56.317.11; 7.13

Results of the calibration and evaluation step are evaluated statistically in table 3.2-2 and shown graphically in figures 3.2-3 to 3.2-7.

Table 3.2-2 Statistical evaluation of the model predictions for the calibration and evaluation step

Ecosystem Model used

Mean measured organic carbon

%

Mean predicted organic carbon

%

Model Efficiency of the prediction

Root mean square error of the prediction

Arable CENTURYRothC

2.182.18

2.312.95

0.327-1.828

0.320.67

Managed grassland CENTURY 4.14 4.61 0.129 0.79

Semi-natural grassland CENTURY 9.78 11.17 0.641 2.17

Deciduous woodland CENTURY 4.68 7.15 -15.8 4.73

Coniferous woodland CENTURY 4.83 4.46 0.354 1.20

SID 5 (2/05) Page 9 of 30

Arable

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

4.50

Andov

er

Arrow

Beccle

s

Dunke

swick

Elmton

Evesh

am

Hanslo

pe

Hopsfo

rd

Newpo

rt

Panho

les

Wickha

m

Series

Org

anic

car

bon

%

Mean original O.C. Mean re-sampled O.C. (2003) CENTURY ROTH C

Figure 3.2-3 Predicted topsoil organic carbon contents for the calibrated CENTURY and RothC models in comparison with measured values for representative indicator soil types under arable ecosystems.Yellow bars indicate standard deviation of the re-sampled organic carbon content

Permanent Grassland

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

Brickfi

eldCeg

in

Clifton

Denbig

h

Dunke

swick

Manod

Rivington

Salop

Wickha

m

Wilcoc

ks

Series

Org

anic

car

bon

%

Mean original O.C. Mean re-sampled O.C. (2003) CENTURY

Figure 3.2-4 Predicted topsoil organic carbon contents for the calibrated CENTURY model in comparison with measured values for representative indicator soil types under managed grassland ecosystems. Yellow bars indicate standard deviation of the re-sampled organic carbon content

SID 5 (2/05) Page 10 of 30

Semi Natural Grassland

0.00

5.00

10.00

15.00

20.00

25.00

30.00

5.41 6.11 6.12 6.31 6.4* LH 3.11

Subgroup

Org

anic

car

bon

%

Mean original O.C. Mean re-sampled O.C. (2003) CENTURY

Figure 3.2-5 Predicted topsoil organic carbon contents for the calibrated CENTURY model in comparison with measured values for representative indicator soil types under semi-natural grassland ecosystems. Yellow bars indicate standard deviation of the re-sampled organic carbon content

Deciduous Woodland

0.00

5.00

10.00

15.00

20.00

25.00

3.4* 5.11 5.41 5.51/52 5.71 5.81/2 6.11 7.11

Subgroup

Org

anic

car

bon

%

Mean original O.C. Mean re-sampled O.C. (2003) CENTURY

Figure 3.2-6 Predicted topsoil organic carbon contents for the calibrated CENTURY model in comparison with measured values for representative indicator soil types under deciduous woodland ecosystems. Yellow bars indicate standard deviation of the re-sampled organic carbon content

SID 5 (2/05) Page 11 of 30

Coniferous Woodland

0.00

2.00

4.00

6.00

8.00

10.00

12.00

5.41 5.5 5.71 6.11 6.31 7.11 7.13

Subgroup

Org

anic

car

bon

%

Mean original O.C. Mean re-sampled O.C. (2003) CENTURY

Figure 3.2-7 Predicted topsoil organic carbon contents for the calibrated CENTURY model in comparison with measured values for representative indicator soil types under coniferous woodland ecosystems. Yellow bars indicate standard deviation of the re-sampled organic carbon content

The model efficiency rating used in table 3.2-2 is a statistical evaluation of the overall goodness of the fit of predicted values to measured values. It can range between – very large values and +1, with values of 1 indicating a perfect fit to the measured data. Negative values indicate an unacceptable fit to the measured data, whereas values in excess of 0.6 indicate a good fit. Values between 0 and 0.6 indicate an acceptable but not good fit.

In combination, the figures and table show that:

For arable ecosystems, apart from the free-draining sandy soil (Newport), the CENTURY model performs consistently better than does RothC, which always over-predicts the re-sampled topsoil organic carbon content, often to the extent that is outside the upper bound of the standard deviation of the measured data. It also always predicts an increase in organic carbon content for the 1994/95 re-sampling compared to the original 1978-1984 sampled data, whereas the measured data shows six decreasing trends and five increasing trends. The root mean square error of the RothC predictions is almost 0.7 % organic carbon and the model efficiency suggests an unacceptable overall fit of the predicted to measured values.

In contrast, the CENTURY predictions were almost always within the standard deviation of the measured data and, although it has an overall tendency to over-predict measured mean values, it under-predicts the mean measured organic carbon for three of the eleven representative soil types. The root mean square error of the CENTURY predictions is just over 0.3 % organic carbon, representing 0.14 of the predicted mean value. As with arable ecosystems, the model efficiency suggests an acceptable, if not good, overall fit of the predicted to measured values. However, the model incorrectly predicts the measured trend in organic carbon for 4 of the eleven representative soil types, although there appears to be no consistent pattern for such errors related to either measured trends or initial measured organic carbon content. CENTURY also predicts no change in organic carbon for one of the representative soil types where there is an increasing trend in organic carbon content.

For managed grassland ecosystems, the model efficiency of CENTURY predictions also indicates an acceptable, if not good, overall fit of the predicted to measured values but the root mean square error of CENTURY predictions is larger than for arable and also represents a slightly larger fraction (0.17) of the predicted mean value. Accuracy of prediction of the measured trends in organic carbon changes is similar to that for arable ecosystems with four incorrect predictions and six correct ones but no consistent patterns to the incorrect ones.

SID 5 (2/05) Page 12 of 30

For semi-natural grassland ecosystems, CENTURY appears to be performing very well. The model efficiency rating indicates a good fit of the predicted to measured values and prediction of the measured trend in organic carbon changes is correct for all the indicator soil types. Nevertheless, the root mean square error of CENTURY predictions is larger than for both arable and grassland ecosystems and represents a slightly larger fraction (0.19) of the mean predicted value.

The performance of CENTURY in predicting measured changes of soil organic carbon in woodland ecosystems is worse than for any of the other ecosystems tested. For coniferous woodland, although the model efficiency suggests an acceptable, if not good, overall fit of the predicted to measured values, the prediction of measured trends in organic carbon changes is poor with four incorrect and only three correct predictions and, again, no consistent pattern to the incorrect predictions. For deciduous woodland ecosystems, the model efficiency shows an unacceptable fit of the predicted to measured values though the prediction of trends is better than for coniferous woodland with five correct and three incorrect predictions. The root mean square error of CENTURY predictions is large, representing 0.28 of the mean predicted values for coniferous woodland and 0.66 of the mean predicted values for deciduous woodland

Overall the model calibration and evaluation step indicates that the CENTURY model can be used to predict future changes in soil organic carbon contents with reasonable confidence for semi-natural grassland ecosystems in the UK and with somewhat less confidence for arable and grassland ecosystems and with considerable caution for woodland ecosystems. Root mean square errors associated with model predictions are likely to range from +0.32 % organic carbon for arable, +0.79 % for managed grassland, +1.20 % for coniferous woodland, +2.17 % for semi-natural grassland and +4.73 for deciduous woodland.

In addition this evaluation phase has shown that:

For slowly permeable, seasonally waterlogged soils (soil type 5) under both arable and managed grassland ecosystems, manipulation of the upper layer drainage parameters for CENTURY was necessary to achieve the best simulation of measured changes. Restricting the upper layer drainage from such soils to simulate wet conditions during the winter months resulted in too small a reduction of organic carbon compared to the measured data. By changing the drainage parameters to ensure free drainage of excess water from the upper layers during the winter months a much better simulation was achieved. The implications of this are that artificial field drainage of such soils increases organic carbon turnover and causes more rapid reduction of topsoil organic carbon compared to similar soil types under semi-natural ecosystems.

For wet lowland heath podzols (soil type 4b), the measured changes in organic carbon indicate a consistent increase in soil organic carbon, the only soil type under semi-natural vegetation for which this occurs, although the number of data points available for this soil type / ecosystem combination is limited. The only way that such changes could be simulate using CENTURY was to severely limit the drainage of excess water from the upper soil layer and to use the anaerobicity parameters in the model to reduce decomposition rates due to anaerobic conditions. The parameters are used to scale carbon decomposition rates between fully aerobic and fully anaerobic conditions, depending on the ratio of precipitation to evapotranspiration. Calibration of these parameters has been successfully used to model carbon changes in two fen ecosystems in the USA (Chimner et al, 2002). The wet lowland heath podzols to which the data applies are all from the New Forest area of Hampshire and Dorset and are known to be associated with small areas of ‘valley mires’. The combination of measured data and CENTURY simulations suggests that in areas where waterlogged conditions are controlled principally by local hydro-topographic features rather than climatic factors, changes in organic carbon contents need to be simulated mainly from estimated changes in soil water regime, rather than climate.

3.3 Drained Organic soilsThe empirically derived decay curve model derived for cultivated organic topsoils in MAFF project LE0203 (Burton, 1995) is illustrated in figure 3.3-1 and the exponential relationship for the curve is:

y = a e(-0.0193x)

where: y = organic carbon % in year za = organic carbon % measured at any specific time, t.x = time in years between t & z

(for the purposes of predicting climate-change induced changes in topsoil organic carbon, t is the year 1990).

This empirical decay model was used to estimate the measured change in organic carbon content of all drained arable peat soils in the National Soil Inventory data sets, using the original sampling date of each site as time ‘t’, the measured organic carbon content at this date as ‘a’ and the number of years between the original sampling and the repeat sampling as ‘x’. Only seven NSI points are available for this exercise and the results of the predictions are shown graphically in figure 3.3-2. The model efficiency of the predictions is 0.341, indicating an acceptable, but not good, fit to the measured data. The mean measured organic carbon content of the 7 re-sampled sites is 19.2 % and the mean predicted value is 16.7 %, with a root mean square error of 10.22 % organic carbon. However, because of the small number of samples, these results are heavily influenced by the

SID 5 (2/05) Page 13 of 30

Figure 3.3-1 Decay curve for drained peat soils (after Burton, 1995)

Figure 3.3-2 Predicted topsoil organic carbon contents for drained arable peat soils using the empirical decay curve model of Burton (1995) in comparison with measured values from National Soil Inventory points.

single point where the organic carbon content increases from 26.5 % to 46.8 %. This dramatic increase in organic carbon contrasts strongly with measured carbon change trends at the remaining six sites and may be caused by unusual local factors or represent an error in the database. If this one site is omitted from the analysis, the model efficiency increases to 0.502, the mean measured organic carbon content of the re-sampled sites decreases to 14.6 % and the mean predicted value decreases to 15.8 % with a root mean square error of 4.25 % organic carbon. Based on this analysis, the empirical decay curve model was used to predict future changes in soil organic carbon contents for drained peat soils.

3.4 Undrained organic soils

SID 5 (2/05) Page 14 of 30

For all undrained peaty and peaty topped soil series, the dominant climate change-induced process impacting on organic matter dynamics will be changes to their current waterlogged moisture regime and associated aerobic/anaerobic conditions, especially in the upper soil layers. Changed water regime impacts on organic carbon turnover are likely to dominate any changes resulting from increased or decreased temperatures or climate-induced biological effects. In the CENTURY model there is a facility to simulate both waterlogged conditions in the upper soil layer and to scale carbon decomposition rates between fully aerobic and fully anaerobic conditions, depending on the ratio of precipitation to evapotranspiration. This facility has been successfully used to model carbon changes in two fen ecosystems in the USA (Chimner et al, 2002). By adjusting the drainage and anaerobicity factors in CENTURY, the increase in topsoil organic carbon measured in semi-natural wet heath podzols was successfully simulated (see figure 3.2-5). The CENTURY model was therefore used to simulate climate change-induced changes in soil organic carbon in undrained peat and peaty topped soils as illustrated in figure 3.4-1.

Figure 3.4-1 Methodology for predicting climate change-induced changes in organic carbon content of undrained peat soils.

3.5 Predicted future changesThe calibrated CENTURY model has been used to predict changes in topsoil organic carbon content for the range of indicator soil types representative of each ecosystem. Two examples are shown in figures 3.5-1 and 3.5-2.

The two figures illustrate how the predicted future trends should be interpreted in the light of the measured and modelled data. For the Evesham series under arable (figure 3.5-1) the measured data indicate a decrease in organic carbon, but the model predicted an increase over the measured time period. The modelling is thus incorrectly predicting the measured trend of change. For the future scenarios, however, the model is predicting an increase from the measured re-sampled values (1995) to 2030 but then little change through to 2080. In view of the incorrect prediction of the measured trend of change, the implications of the future modelling are that, for this soil, the measured decrease in topsoil organic carbon is likely to slowly decline over the next 10 to 15 years after which there will be little change, at least up to 2080. Of the four scenarios illustrated the largest changes are associated with the high emissions (H). Predicted changes associated with other scenarios are almost identical

For the Brickfield series under managed grassland (figure 3.5-2) the model correctly predicts the measured trend in organic carbon change but slightly under-estimates it. For the future scenarios, the model predicts a continuing but small decline at least up to 2080. For this soil –land use ecosystem therefore it is very likely that the measured decrease in organic carbon will continue into the future, with largest changes again associated with the High (H) emissions scenario and little difference between other scenarios.

Application of such arguments to the simulated data suggest that: Under arable land use, whereas the measured changes in soil organic carbon from the National Soil Inventory show a general decrease, except where levels are already low, the climate change driven models predict only a brief continuation of this trend before organic carbon contents stabilise or even increase slightly, possibly because of a positive feed-back from increased yields and resulting crop residues. In contrast, for soils under permanent managed or semi-natural grassland, the climate models predict that the overall reductions measured from the National Soil Inventory are likely to continue into the

SID 5 (2/05) Page 15 of 30

Anaerobicity option in

CENTURY

Rainfall:ET in upper layer

OC change in response

to water regime change

calibration

Baseline Rainfall:ET calibrated to anaerobicity parameter

using measured OC NSI data

future, albeit at a slightly reducing rate. The picture for woodland is uncertain because the models perform least well for this land use and further work is needed to clarify this situation.

Evesham Arable 2030-2080

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

2030 2040 2050 2060 2070 2080Time, years

Original Resampled Model (1995) Hi Lo MH ML

Figure 3.5-1 Future change trends in soil organic carbon for the Evesham series under arable.

Brickf ield PG 2030-2080

5.21

4.28

4.62

4.07

3.88

3.74

3.64

3.57

3.52

4.09

3.91

3.79

3.71

3.65

3.60

4.09

3.92

3.80

3.72

3.66

3.62

4.09

3.92

3.80

3.72

3.66

3.62

0.00

1.00

2.00

3.00

4.00

5.00

6.00

2030 2040 2050 2060 2070 2080Time, years

OC

, %

Original Resampled Model (1995) Hi Lo MH ML

Figure 3.5-2 Future change trends in soil organic carbon for the Brickfield series under managed grassland.

SID 5 (2/05) Page 16 of 30

4. Estimation of climate-change induced changes in soil wetness class and Hydrology of Soil Types.Soil wetness and Hydrology of Soil Types (HOST) classes are determined from a number of soil and climatically derived parameters (see figure 5.1). The climatically derived parameters required for classification include climatic field capacity (FC) period, the amount of rainfall during the FC period and potential soil moisture deficit (PSMD). These data are available from the Agroclimatic databank for England and Wales (Jones and Thomasson, 1985), held within the National Soil Resources Institute Land Information System (LandIS). FC period and PSMD were calculated using a simple soil water balance model driven by the difference between precipitation and potential evapotranspiration (PET), the latter being available for the period 1961-1975 at 970 Meteorological Office stations. In order to derive similar data for the future climate scenarios it is therefore necessary to derive PET from the UKCIP future scenario datasets.

4.1 Calculation of PET from the UKCIP dataThe FAO Penman-Monteith method for estimation of PET is considered to be the most robust and requires the following climatic variables for the calculation: minimum and maximum air temperature, solar radiation, wind speed and humidity. These parameters are available in monthly time steps at 50km grid resolution in the UKCIP database. A number of programmes that calculate PET based on Penman-Monteith were tested with daily data for 1998 from the Silsoe weather station, located in Bedfordshire, in one of the driest areas of the country. PSMD was calculated for each method and compared with agroclimatic data for the region (Cambridge). The model with PET estimates closest to the ‘measured’ PET data and resulting PSMD was selected for use with the UKCIP data. However, testing of the chosen method using the UKCIP02 baseline data sets (1960 – 2000) to calculate PET and associated moisture balances for two contrasting UKCIP02 grid squares, where measured weather station data was available for validation, highlighted significant problems. UKCIP Grid squares 430 (SW England) and 374 (Eastern England) correspond to weather stations at Rosewarne (located on the north coast of Cornwall) and Cambridge, respectively. Figure 4.1-1 shows PET and PSMD calculations for grid 430 and comparable weather station data. Using the UKCIP02 baseline data for grid 430, calculated PET never exceeds precipitation resulting in no PSMD developing during the summer. Although this area is wet, local weather station and regional climatic data for earlier periods (which are wetter and cooler than the period 1960 – 2000) suggest that a soil moisture deficit does develop in the summer months.

-80

-60

-40

-20

0

20

40

60

80

100

120

jan feb mar apr may jun jul aug sep oct nov dec

rosewarne PET

UKCIP grid 430PETmemoir PET

rosewarne PSMD

UKCIP grid 430PSMDmemoir PSMD

Figure 4.1-1 PET (positive) and PSMD (negative) (mm) for grid 430 (based on UKCIP 50km monthly data). Rosewarne = weather station PET data from Seismic, memoir- =agroclimatic data from Findlay et al, 1984.

Similarly PET and PSMD calculations in a drier region (grid 374) also indicate PET is underestimated and resulting soil moisture deficits are much lower than would be expected from weather station data (figure 4.1-2). Analysis of the input variables for grid 374 revealed lower monthly values for wind and higher values for relative humidity in the UKCIP02 50km baseline data compared with station averages, which are the main reasons for the lower than expected calculated baseline PET for the grid square. This suggests that using the UKCIP02 baseline data to calculate the PET component of a monthly soil moisture balance would significantly under-estimate the duration of the field capacity period and cannot be used to reliably estimate duration of the field capacity period for future climate scenarios.

SID 5 (2/05) Page 17 of 30

-150

-100

-50

0

50

100

jan feb mar apr may jun jul aug sep oct nov dec

Cambs PETUKCIP grid 374 PETSilsoe98 PETCambs PSMDUKCIP grid 374 PSMDSilsoe98 PSMD

Figure 4.1-2 PET (positive) and PSMD (negative) (mm) for grid 374 (based on UKCIP 50km monthly data). Cambs= weather station PET data from SEISMIC (Hollis et al, 2004), Silsoe98= weather station data PET calculated using same method as grid 374.

Knox et al. (in press) have recently tackled this problem and produced a 5km PET dataset based on UKCIP02 data for the baseline and a selection of scenarios. PET was calculated using Penman-Monteith but because of the limited UKCIP climatic data at 5km the authors have used empirical relationships to estimate the missing parameters (refer to Knox et al., in press for methods). However, The Penman-Monteith calculation method used by Knox et al was not the one selected as giving the best match to the ‘measured’ data at Silsoe. Comparison of the monthly moisture balances calculated using the Knox et al PET data for the UKCIP02 grid relevant to Cambridge with those calculated using the preferred Penman-Monteith method applied to relevant weather station data for Cambridge (figure 4.1.3) suggests that the Knox et al method significantly over-estimates PET and results in an unrealistically large cumulative PSMD. In wetter areas relevant to Rosewarne, a similar pattern is observed.

-700

-600

-500

-400

-300

-200

-100

0jan feb mar apr may jun jul aug sep oct nov dec

baseline2020 MH2020 LO2050 MH2050 HIcambs

Figure 4.1-3 Cumulative PSMD (mm) for grid 545000 260000 (Cambs) using PET from Knox et al. data set. Cambs = data from weather station and comparable to the baseline data set.

The only problem with applying the Knox et al PET dataset is the method used to calculate Penman-Monteith PET and, in order to make this dataset applicable to estimating changes in the FC day period, it is simply necessary to assess of the magnitude of the overestimation of PET and scale it downwards to match the PET data used to produce the original NSRI agroclimatic databank for England and Wales (Jones and Thomasson, 1985). From this original databank, PET data was available for the period 1961-1975 at 970 Meteorological Office stations. This point data was interpolated to the 5km grid corresponding to the Knox et al data set. Inverse distance weighting was used to interpolate point data to a contoured geostatistical surface. The surface was converted to a 5km raster layer with coordinates comparable to the Knox et al data set. The differences between

SID 5 (2/05) Page 18 of 30

no return to field capacity

the two data sets were then calculated for each 5km grid square and each month. As an example of the differences, for the majority of the country for the month of January, PET from the Knox et al data is at least 6 mm greater than that derived from the agroclimatic databank weather stations, which is in excess of 1 standard deviation of the PET values for January in the meteorological weather station data set.

Differences in PET between the Knox et al baseline and NSRI baseline vary monthly with the largest deviations between the winter months. Scaling down of the Knox et al data on a monthly basis was therefore carried out and, in order to retain the spatial structure of PET across England and Wales, linear regressions were used to define a relationship between average annual PET (AAPT) and spatial variables (easting, northing and altitude). This relationship is as follows:

AAPT NSRI = 431.927 + EASTING*4.57710-5 + NORTHING*-1.111 10-4 + ALTITUDE*-0.134 + KNOXAAPT*0.197

R2= 0.88

Scaling relationships between monthly PET and AAPT were then derived from the NSRI dataset with easting and northing. The equations were in the following form:

monthlyPET = b0 + b1EASTING + b2NORTHING + b3AAPT

The specific monthly algorithms are detailed in appendix 1.The regressions were applied to the scaled Knox AAPT to derive scaled monthly PET for the baseline. As Knox used the same methods for the calculation of PET for the baseline and scenario datasets it is assumed that the scaling relationships are also similar. Therefore, the monthly algorithms were applied to the scenario datasets to scale PET.

4.2 Calculating the field capacity period (FCD)For the future change scenarios, a simple monthly soil water balance model was calculated as the difference between monthly precipitation and the derived PET values. For months where precipitation exceeded PET soils are assumed to be at climatic field capacity. In months where PET exceeds precipitation, successive monthly deficits are accumulated in order to derive cumulative potential soil moisture deficit (PSMD). Climatic field capacity period is expressed as the number of days in the year when there is no soil moisture deficit and to calculate this, the days when the field capacity period ends (EFC) in the spring and returns (RFC) in the autumn are required. Resolving EFC day from the monthly data was achieved by assuming a linear relationship between the monthly water balance between the first month where a deficit occurs and the preceding monthly balance. The point at which the water balance is zero is assumed to represent the first day of moisture deficit and hence end of field capacity. The return to field capacity is calculated in a similar manner, where a linear relationship is assumed between the last month with a moisture deficit and the subsequent month’s precipitation excess. The point at which the water balance is zero is assumed to represent the end of moisture deficit and hence the day of return to field capacity (figure 4.2-1).

Figure 4.2-1 Schematic representation of the annual soil water balance.

In some cases the moisture deficit was carried over to January or February and RFC day was calculated accordingly. If climatic return to field capacity was not reached by the end of February, the return to field capacity day was automatically set to the end of February. In contrast, situations without the development of a moisture deficit are assumed to be at field capacity for the whole year (365 FCD). FCD was then calculated for the baseline and each of the scenarios.

SID 5 (2/05) Page 19 of 30

4.3 Estimated changes in the climatic field capacity period for future scenariosThe distribution of estimated changes in the duration of the field capacity period for UKCIP02 future scenarios for 2020’s low and medium-high emissions and 2050’s medium high and high emissions are shown in figure 4.3-1.

Figure 4.3-1 Estimated changes in the field capacity period for different future scenarios

SID 5 (2/05) Page 20 of 30

A summary of the percentage of land affected by the predicted changes is given below.

Scenario Change less than +7 days Increase by at least 7 days Decrease by at least 7 days2020’s low 15 % 3 % 82%2020’s medium high 15% 2% 83%2050’s medium high 8% 0.4% 92%2050’s high 6% 0.2% 94%

When interpreting the data, the limitations of the method used to estimate future scenario FC days need to be taken into account. It is likely that data for grid cells around the coastal areas will be most unreliable and should be discounted. Given this, the data suggest that very few, if any areas of the country are likely to show a significant increase in the field capacity period (most of the grid cells showing significant increases are in coastal areas). By far the biggest impact of climate change is to significantly reduce the duration of the field capacity period with the largest impacts occurring earliest in the English midlands and in areas fringing the highest areas of Wales, south west and north west England. Areas with predicted changes of a week or less (+ 7 days) are considered to show no significant change in FC days. Such areas occur in the uplands of Wales, south west and north west England and the lowlands of western Cambridgeshire & Bedfordshire, the Thames estuary, and Essex & south east Suffolk. These represent the two climatic extremes (wettest and driest areas) of England and Wales. For the wettest areas, soils are currently at field capacity for all of the year (rainfall always exceeds PET) and it requires a significant increase in PET and/or reduction in rainfall to generate any period of moisture deficit. However, at the margins of these areas, where the seasonal rainfall only just balances the seasonal PET, relatively small increases in PET can have a large impact, creating a potential soil moisture deficit for one or two months during the summer and thus significantly reducing the field capacity period. This is well illustrated in figure 4.3-1 where, irrespective of scenario, the ‘wet’ areas showing no significant change in FC days are fringed by areas showing the largest changes. The maps also show that, at least for the higher emission scenarios, this very ‘sensitive’ zone between little change and maximum change shrinks over time to become centred on the mountains of Snowdonia and the Lake District. In the driest areas, the lack of change in FC days is likely to be an ‘artefact’ of the methodology because the current return to field capacity already occurs at or near to the end of February. The end of February ‘automatic return to field capacity’ in the simple moisture balance methodology thus precludes any significant change in their very short field capacity period.

4.4 Impacts on soil moisture regimes and HOSTSoil moisture regimes in England and Wales have been characterised in terms of ‘wetness classes’ based on estimates of the duration of soil waterlogging within 40 and 70 cm of the surface (Hodgson Ed., 1997). Changes in the climatic field capacity period are likely to affect soil moisture regimes in two significant ways.

Free draining, relatively permeable soils that are not ‘wet’ within 70 cm depth for more than 30 days in most years (soil wetness class I) will only experience changes in the periods when their upper layers have a moisture deficit (their ‘dryness sub-class’ as defined by Hodgson Ed., 1997). The estimated changes in the field capacity period indicate that such soils are likely to become ‘drier’ across England and Wales. In the lowlands, vegetation growing on them is more likely to experience drought effects. In wetter upland areas, drought effects are less likely because of the relatively small potential moisture deficits that will develop. However, free draining soils in these areas usually have relatively large amounts of organic matter. They are recognised at a global level as Umbrisols, soils with dark, relatively acid, organic-rich upper layers associated with cool, humid climates where precipitation considerably exceeds evapotranspiration for most of the year (Deckers et al, Eds. 1998). Such soils are most sensitive to reductions of the field capacity period, which are likely to have a significant impact on their organic matter cycle and significantly increase soil organic matter turnover.

Soils which are wet within 70 cm depth for more than 30 days in most years (soil wetness classes II to VI) are likely to experience reductions in the period of time that they are wet. Impacts will vary depending on the cause of wetness with the soil profile.

Upland ‘climatic’ peaty soils which are wet within 40 cm depth for at least 335 days in most years (wetness class VI) because of a large excess of rainfall for most of the year are likely to be most affected, apart from in the core mountain areas of Snowdonia and the Lake District. The estimated changes in the field capacity period illustrated in figure 4.3-1 are greatest in those areas occupied by such peaty soils, which are likely to experience significant changes in the nature and amount of their organic matter because of significantly increased dryness and aeration in their upper layers. The data suggest that, even in the low emission scenario for the 2020’s, of the 534 grid cells which have an FC day period of between 334 and 365 days, 37% (200 grid cells) experience a reduction of at least 30 days and 7% (40 grid cells) experience a reduction of at least 60 days. Such reductions are likely to change a wetness class VI soils to wetness class V.

In slowly permeable soils with ‘perched’ water tables (Surface Water Gley and ‘stagnogleyic’ soil types; Avery, 1980), the climatic field capacity period has been shown to significantly affect the duration of soil ‘waterlogging’ within 40 cm and 70 cm of the surface (Hollis, 1989) and the relationship between the field capacity period and duration of waterlogging differs depending on the presence or absence of field drainage.. Most slowly permeable, seasonally wet (Surface Water Gley) soils (wetness classes III to V) under arable or reasonably intensive permanent grassland production systems had field drainage installed, mainly through the late 1960’s to 1980’s.

SID 5 (2/05) Page 21 of 30

Based on the relationships derived by Hollis (1989), estimated changes in the climatic field capacity period for future scenarios have been used to estimate associated potential changes in the duration of soil waterlogging in drained and undrained Surface Water Gley soils. The distribution of Surface Water Gley soils is shown in figure 4.4-1, together with estimated changes in the duration of waterlogging within 40 cm of the surface in such soils associated with example future scenarios.

Figure 4.4-1 Distribution of Surface Water Gley soils in England & Wales and estimated changes in their duration of waterlogging within 40cm of the surface for future scenarios.

SID 5 (2/05) Page 22 of 30

Surface Water Gley soils cover approximately 23% of all land in England & Wales and approximately 39% of the agricultural land. The estimated changes suggest that, whatever the future climate scenario, all of this land will experience a decrease in duration of waterlogging within 40 cm depth. Reductions are likely to be greatest in the field drained soils, with an average reduction of between 18 and

34 days waterlogged within 40 cm depth depending on scenario as opposed to an average reduction of between 15 and 29 days waterlogged within 40 cm depth for undrained soils. The greatest changes are predicted to occur in the Welsh and Pennine hills and in the heavy textured soils of the midlands and northeast England. Differences in impact between scenarios are not large, with the biggest differences occurring between the High emissions scenario and all three other scenarios. Changes in overall water regime, as defined according to soil wetness class, can be misleading because of the ‘rigid’ nature of class boundaries. Given this proviso, the biggest changes are predicted to occur in wetness classes IV & V, with between 25% and 30% of wetness class IV soils changing to wetness class III by the 2020’s depending on scenario, and 35 to 45% changing to wetness class III by the 2050’s. Wetness class V soils appear to be more sensitive to climate-induced changes, with 18 to 25 % predicted to change to wetness class IV by the 2020’s but 55% to 70% predicted to change to wetness class IV by the 2050’s. The principal difference between scenarios appears to be in the 2020’s where there are significant regional differences in the magnitude of predicted change, depending on the scenario. In the midlands and eastern England the medium high emissions scenario results in less reduction in duration of waterlogging within 40 cm depth, than does the low emissions scenario. Such differences mainly affect soils of wetness classes III and IV. In contrast, in the west of England and Wales the magnitude of predicted change increases from the low emissions scenario to the medium high emissions scenario.

The overall impact of the estimated changes in soil water regimes of Surface Water Gley soils is to increase the period when they are accessible to stock or machinery, without causing significant soil structural damage. It may also mean that, any impacts of deteriorating field drainage systems over time, because of the lack of maintenance, is more than offset by climate-induced reductions in waterlogging.

Duration of the field capacity period also has a weak relationship with duration of waterlogging in more permeable soils affected by seasonally rising groundwater (Hollis, 1989). However, there are many other significant factors affecting the duration of wetness in such soils and in particular they are affected by the hydrological regimes of the streams and rivers with which they are associated. It is not possible at this stage to assess how their water regimes may change in response to future climatic scenarios.

Changes in soil water regimes have an impact on HOST because they increase the period of time when the drainable component of soil is not waterlogged and thus storage is at its optimum. The effect of this is to change HOST classes in the slowly permeable and impermeable substrate classes, particularly where the soil drainable porosity is relatively large. Depending on the predicted changes in duration of waterlogging within 40 cm depth, HOST classes 24 and 25 are most likely to change to classes 18 and 20. In addition, peaty or humose topped upland soils over more permeable substrates (HOST class 15) are likely to become significantly drier and their peaty or humose nature will become lost. A significant proportion of such soils are most likely to change to HOST classes 17 and 4, depending on the nature of their substrates.

5 Derivation of climate change scenario-specific soil property datasets for all soil series in the 1:250,000 scale National Soil Map database.

The soil properties required to characterise soil conditions relating to the various UKCIP02 climate scenarios are:

Soil series-land use specific soil wetness class and Hydrology Of Soil Types (HOST) class. Soil series-land use specific soil layer properties comprising: layer upper depth, layer lower depth,

gravimetric % content of clay (<0.02 mm), silt (0.002 – 0.06 m), total sand (0.06 – 2 mm), fine sand (0.06 – 0.2 mm) and organic carbon; bulk density (g cm-3); % volumetric water content at 0, 5, 10, 40, 200 and 1500 kPa tension, saturated hydraulic conductivity (cm day-1), van Genuchten’s and n parameters.

The overall methodology used to derive scenario-specific soil property data sets is illustrated in figure 5.1 below. Of the required soil layer properties, existing mean mineral particle-size fractions relating to each soil series are assumed not to change significantly for the future scenarios. Methods for estimating the bulk density and hydraulic characteristics of soil layers across England and Wales have been described by Hollis et al (1995) and Hollis (2004). These methods require basic data on the mineral particle-size fractions, the organic carbon content, the pedogenetic soil layer classification and the permeability class of the soil parent material lithology. On a soil series basis, mean mineral particle-size fractions and the parent material lithology are unlikely to change significantly for the future scenarios. The pedogenetic soil layer classification indicates the dominant weathering processes that have created the layer and give rise to its pedological characteristics. Whilst the intensity and interaction of these processes may alter with climate change over the next 100 years, such changes are either unlikely to be of sufficient magnitude, or will not have had sufficient time, to change the pedogenetic characteristics of the layer. The main soil property driving climate change induced changes in soil layer properties is thus organic carbon. Soil layer characteristics for the future change scenarios have thus been estimated from the existing soil series data on particle-size fractions, pedogenetic soil layer sequence and parent

SID 5 (2/05) Page 23 of 30

material lithology held in the Land Information System, LandIS, together with the modelled changes in organic carbon content using the methods described in section 3.

Figure 5.1. Methods used to derive climate-change scenario-specific soil property data sets.

6 Development of the computer-based software tool for deriving user-specified spatial soil input parameters relevant to specific climate change scenarios.

The software tool will provide a basic interface so that the user can easily select the required modelling options. The majority of the software tool will operate in the background performing calculations and writing output files. The software tool is designed using Microsoft’s Visual Basic development environment, and data will be stored in a Microsoft Access database.

6.1 Base dataThe soil profile data sets store an enormous amount of information. For the 434 modelled profiles, parameter data is provided at 1cm intervals down a nominal 140cm profile, and under 4 different land uses. This results in:

243,040 (434 * 140 * 4) data points per climate change soil parameter. This information is stored in a series of tables in a Microsoft Access database.

6.2 Modelled dataIn order for the soil information to be useful for incorporation into climate models a software tool is required that will perform three modelling functions:

Spatial functions: Allow the user to select a method by which the soil profiles are divided up to provide information for the

climate change models. Integrate the soil profile divisions at the base data resolution Allow the user to select a spatial resolution at which the integrated data can be up-scaled for climate

change models.

Non-spatial functions: Provide a means by which the modelled data can be presented by particular generic soil types

SID 5 (2/05) Page 24 of 30

Bulk density

Organic carbon

Particle-size;Pedogenetic layer class:Parent material lithology

Hydraulic parameters

From existing soil series characteristics in LandIS.

Estimated using CENTURY carbon dynamics model and

empirical decay-curve for drained peat.

Models driven by UKCIP02 climate scenarios

Estimated using monthly water balance model driven

by UKCIP02 climate scenarios

FCD;XWR

Depth to slowly

permeable layer

Wetness class;

HOST

Soil profile divisionThe soil profile, nominally 140cm in depth can be divided up in one of two ways:

By soil horizon:Each soil profile is divided up into a variety of distinct soil layers or ‘horizons’ of varying depth depending on the soil series. These horizons form a natural division of the soil profile that separates layers of possible contrasting characteristics and provides a means of linking similar layers from different soil series as illustrated below:

By soil compartment:As an alternative to the natural divisions of the soil horizon, it is possible to divide up the profile according to arbitrarily assigned depths to create soil compartments. This provides information that can be used to investigate depth-specific climate change topics (for example effect on building foundations).

6.3 Soil profile up-scalingIn order to allow data to be easily up-scaled, the base data is held in tables characterising 1km by 1km cells. There are two base-data data sets:

Soil data Land use data

Soil dataThe National Soil Resources Institue’s 1:250,000 NATMAP, held in LandIS was used to derive a 1km by 1km data set for England and Wales (NATMAP1000).

SID 5 (2/05) Page 25 of 30

A

C

B

B

C

A

Soil Series 1 Soil Series 2

Soil Series 1 Soil Series 2

This provides information on the contributing soil series and their percentages within each of the 158,394 1km cells:

For each 1km cell the leading 20 contributing soil series and their percentages are listed. In certain cases water bodies and other non-soil features may represent a leading contributing component, but the software will factor out these features since it is assumed that these features will not affect the soil-based climate change parameters. In general, for each 1km cell there will be a small percentage of unlisted soil series. Where this is the case the leading 20 soil series percentage contributions have been adjusted to take into account the difference of the unlisted soil series.

Land use dataThe CORINE2000 EU land use data set was acquired for this project. This data provides land use data for 46 separate classes on a 250m by 250m cell basis. A 1km by 1km data set was derived from this data by a process of re-projection (from the Lambert Conic European projection to the OS GB projection), generalisation (from 250m to 1km cells) and amalgamation (from 46 classes to 9 classes)

This provides information on the contributing land uses and their percentages within each of the 158,394 1km cells:

For each 1km cell the 9 contributing land uses (4 land uses corresponding to the 4 land use categories of the soil series profile data; woodland, arable, semi-natural grassland and permanent grassland, plus 5 others representing non-soil land uses; urban, suburban, water, bare ground [rock], no data). The accompanying CORINE data notes describe the suburban class as being between 30-80% hard surfaces [urban] with the remainder being soft-surfaces [semi-natural grassland]. Therefore, a decision was made that the suburban class area in each 1km cell is made up of, on average 45% semi-natural grassland. This division is taken into account during the modelling process.

Soil-land use combination:The soil data and the land use data are combined to form a soil-land use combination for each 1km cell:

SID 5 (2/05) Page 26 of 30

A

C

B

A

B

C

30

50

20

SOURCE1:250,000 VECTOR

BASE DATA1km GRID

3

1

2

CELL SOIL %

B,A,C

C,A

A,B,C

50,30,20

70,30

65,30,5

X

Y

Z

50

40

10

SOURCE250m RASTER

BASE DATA1km GRID

X X Y Y

X X X X

X Y Y Y

X Y Z Y

3

1

2

CELL LANDUSE %

X,Y,Z

X,Y

X,Z,Y

50,30,20

70,30

65,30,5

During this process, an assumption is made in the software model that the soil and land use classes are evenly distributed throughout the 1km cell. However, it is known that certain soil-land use combinations do not normally occur throughout England and Wales and, where such soils occur, this null relationship is taken into account when distributing soil types across existing land uses.

Soil profile integrationThe software integrates the soil profile data firstly by soil horizon or soil compartment:

and then over each 1km cell:

In instances (where the user has selected the soil compartment modelling option) where a soil profile does not extend down through the entire soil compartment the following integration is applied:

In instances (where the user has selected the soil compartment modelling option) where a soil profile does not contribute to a soil compartment the following integration rule is applied:

SID 5 (2/05) Page 27 of 30

BASE SOIL DATA1km GRID

BASE LANDUSE DATA

1km GRID

AX

AY

AZ

15

2

CZ

12

3

BASE COMBINATION DATA

1km GRID

X

Y

Z

50

40

10

A

B

C

30

50

20

*

*

For each parameter:e.g. AvgA= [10 + 11 + 13 + 11 + 10 ] / 5

A

1011

1113

10

For each parameter:AvgCELL= (0.3 * AvgA) + (0.5 * AvgB) + (0.2 * AvgC )

A 30%

C 20%

B 50%

A B C

AvgA

AvgB

AvgC

For each parameter:AvgCELL= (0.3 * AvgA) + (0.5 * AvgB) + (0.2 * AvgC)

A 30%

C 20%

B 50%

A B C

AvgA

AvgB

AvgC

Both instances are flagged in a log file accompanying the modelled data.

6.4 Up-scaling grid dataThe user is able to select from a variety of up-scaled resolutions (e.g. 2km, 5km, 10km, 40km, …):

The software integrates the base cell data to the up-scale resolution:

In instances where the up-scale cell does not have a full compliment of contributing base cells (i.e. in coastal areas) the software adjusts the up-scaled integration accordingly:

Generic profile typesRather than providing just one integrated value per parameter per up-scale cell, the software tool provides an option to differentiate this value by generic profile types ranging from shallow organic soils to deep calcareous soils:

6.5 Software outputThe output from the software is in the form of a comma-delimited text file containing, for each parameter, a line of data that includes:

X and Y coordinates of the lower left-hand corner of the up-scaled cell

SID 5 (2/05) Page 28 of 30

For each parameter:AvgBCELL= (0.0) + (0.5 * AvgB) + (0.2 * AvgC)

A 30%

C 20%

B 50%

A B CAv

gB

AvgC

BASE CELL1km

UPSCALE CELL4km

BASE CELL1km GRID

UPSCALE CELL4km GRID

1 2 3

15 16…

BASE CELL1km

UPSCALE CELL4km

1 2 3

15 16

4

8

14

1211

A 30%

C 20%

B 50%

C high in organic matter [0.8]A,B low in organic matter [0.1, 0.2]Without differentiation

one result total 0.29 [100%]With differentiation

three results total 0.29 [100%] high o.c 0.80 [20%]

low o.c 0.16 [80%]

The integrated parameter value for the up-scale cell. A breakdown of the parameter by the generic profile types.

6.6 Availability of the software toolThe tool and associated databases will be available for use by the climate change modelling community in August 2005.

References to published material9. This section should be used to record links (hypertext links where possible) or references to other

published material generated by, or relating to this project.

SID 5 (2/05) Page 29 of 30

Avery, B.W. (1980). Soil Classification in England & Wales: Higher Categories. Soil Survey Technical Monograph 14.Burton, R.G.O. (1995). Evaluating organic matter dynamics in cultivated organic topsoils – the use of historical analytical data. Report for MAFF (Project No. LE0203 - SSLRC Contract No. 81/3830).Chimner, R.A., Coper, D.J. and Parton, W.J. (2002). Modeling carbon accumulation in Rocky Mountain fens. WETLANDS Vol. 22 , No. 1. pp 100-110Coleman, K. and Jenkinson, D.S. (1995). RothC-26.3. A model for the turnover of carbon in soil: Model description and users guide; ISBN 0951 4456 69.Deckers, J.A., Nachtergaele, F.O. & Spaargaren, O.C. (Eds.) (1998).. World Reference Base for Soil Resources: Introduction. Acco, Leuven, Belgium, 165 pages.Findlay, D.C., Colborne, G.J.N., Cope, D.W., Harrod, T.R., Hogan, D.V. and Staines, S.J. (1984). Soils and their use in South West England. Soil Survey of England 7 Wales Bulletin No. 14.Hodgson, J.M. (Ed.). (1996). Soil Survey Field Handbook. Third Edition. Soil Survey Technical Monograph 6.Hollis, J.M. (1989). A methodology for predicting Soil Wetness Class from soil and site properties. SSLRC Research report for MAFF Project c(iii); May, 1989, 83 pp.Hollis, J.M., Thanigasalam, P. Hallett, S.H., Mayr, T.R. & Jarvis, N. (1995). SEISMIC: User manual. Soil Survey and Land Research Centre, Cranfield University, Silsoe, UK. 108 pp.Hollis, J.M., (2004). Major upgrade of SEISMIC. Final Project Report for DEFRA R & D project PL0551. 25pp.

Hulme,M., Jenkins, G.J., Lu, X., Turnpenny, J.R., Mitchell, T.D., Jones, R.G., Lowe, J., Murphy, J.M., Hassell, D., Boorman, P., MacDonald, R. and Hill, S. (2002). Climate Change Scenarios for the United Kingdom: The UKCIP02 Scientific Report, Tyndall Centre for Climate Change Research, Scool of Environmental Sciences, University of East Anglia, Norwich, UK. 120pp.Jones R.J.A. & Thomasson A.J. (1985) An agroclimatic databank for England and Wales. Soil Survey Technical Monograph 16.Kelly R.H., Parton W.J., Crocker G.J., Grace P.R., Klir J., Körschens M., Poulton P.R. & Richter D.D. (1997), Simulating trends in soil organic carbon in long-term experiments using the CENTURY model, Geoderma 81, 75-90Knox J., Weatherhead K. & Hess T. (in press) Mapping the impacts of climate change on soil moisture deficits: implications for irrigation in England and Wales. Climatic Change.Loveland, P.J. (1990). The National Soil Inventory: Survey design and sampling strategies. Pages 73-80 in H. Lieth & B. Markert (eds) Element Concentration Cadasters in Ecosystems. VCH Verlagsgesellschaft, Weinheim, GermanyMetherell, A.K., Harding, L.A., Cole, C.V. and Parton, W.J. (1993). CENTURY. Soil Organic Matter model Environment: Technical Documentation Agroecosystems Version 4.0. Great Plains System research Unit Technical Report No. 4. USDA-ARS, Fort Collins Colorado, USA.Metherell, A.K., Cambardella, C.A., Parton, W.J., Peterson, G.A., Harding, L.A., and Cole, C.V. (1995). Simulation of soil organic matter dynamics in dryland wheat-fallow cropping systems. In: Lal, R., Kimble, J., Levine, E. and Stewart, B.A. (Eds.), Soil Management and the Greenhouse Effect. Adv. Soil Sci., CRC/Lewis Publishers, Boca Raton, FL, pp. 259-270.Parton, W.J., Ojima, D.S., Cole, C.V. and Schimel, D.S. (1994). A general model for soil organic matter dynamics: sensitivity to litter chemistry, texture and management. In: Quantitative Modeling of Soil Forming Processes. SSS Spec. Publ. 39, Madison, WI, pp. 147-167.Paustian, K., Parton, W.J. and Persson, J. (1992). Modeling soil organic matter in organic-amended and nitrogen fertilized long term plots. Soil Sci. Soc. Am J. 56, 476-488.Sandford, R.L. Jr., Parton, W.J., Ojima, D.S. and Lodge, D.J. (1991). Hurricane effects on soil organic matter dynamics in forest production in Luquillo Experimantal Forest, Puerto Rico: results of simulation modelling. Biotropica 23, 364-372.Smith, P., Smith, J. U. and Powlson, D.S. (1996). GCTE Task 3.3.1 Soil Organic matter network (SOMNET): 1996 Model and Experimental Metadata.

SID 5 (2/05) Page 30 of 30