nutrient balance

UK TAPAS Action - Soil Nutrient Balances

UK TAPAS Action Soil Nutrient Balances Final ReportDate: 5th May 2009

Authors:

David Fernall, Defra [email protected] Alistair Murray, Fera [email protected]

-1-


Contents

1. Executive summary ............................................................................................................................ 5 1.1 Main aims of the work plan .......................................................................................................... 5 1.2 Key project phases........................................................................................................................ 5 1.3 Conclusions ................................................................................................................................... 5 1.3.1 Approach/methodology ........................................................................................................ 5 1.3.2 Data issues ............................................................................................................................. 6 1.3.3Quality of specific data components ...................................................................................... 6 1.3.4 Geographic scale ................................................................................................................... 7 1.3.5 System issues ......................................................................................................................... 7 1.4 Results - Project developments and outputs ............................................................................... 7 1.4.1 National level balance sheet system and estimates ............................................................. 7 1.4.2 NUTS1 level balance sheet system and estimates ................................................................ 8 1.4.3 NUTS3 level balance sheet system and estimates ................................................................ 8 2. Introduction to the TAPAS work plan ................................................................................................ 9 2.1 Aims of the work plan................................................................................................................... 9 2.2 Methods ....................................................................................................................................... 9 2.3 Key project phases........................................................................................................................ 9 3. Introduction/background to nutrient balances................................................................................ 10 4. Evaluation of Gross Nutrient Balance approach ............................................................................. 11 4.1 Understanding the policy needs for Gross Nutrient Balances ................................................... 11 4.2 Evaluating the OECD Gross Nutrient Balance approach ............................................................ 13 4.2.1 Strengths ............................................................................................................................. 13 4.2.2 Weaknesses ......................................................................................................................... 14 5. Develop and appraise options at National level.............................................................................. 15 5.1 Current OECD system ................................................................................................................. 15 This spreadsheet system is summarised in a flow chart in Annex 1. ................................................... 16 Details of how the current system can be developed and alternative approaches are described in .. 16 5.2. Data issues ................................................................................................................................. 16 5.2.1. Physical data ....................................................................................................................... 16 5.2.2. Coefficients ......................................................................................................................... 22 5.3 Sensitivity analysis ...................................................................................................................... 23 6. Develop and appraise options at NUTS1 level ................................................................................. 24 6.1 Data Issues .................................................................................................................................. 24 6.2 System Issues.............................................................................................................................. 24 6.3 Example of a simplified approach .............................................................................................. 25 7. Develop and appraise options at NUTS3 level ................................................................................ 25 7.1 Data issues .................................................................................................................................. 25 7.2 System issues .............................................................................................................................. 26 7.3 Proposal for a simplified method ............................................................................................... 26 8. IT systems for calculating balances .................................................................................................. 28 8.1 Requirements of IT system ......................................................................................................... 28 8.2 National level data...................................................................................................................... 28 8.3 NUTS1 level data ........................................................................................................................ 28 -2-


8.4 NUTS3 level results ..................................................................................................................... 29 Worksheet 1 NUTS3 level data on crops and livestock ..................................................................... 30 Worksheet 2 NUTS3 proportions of NUTS1 totals ............................................................................ 30 Worksheet 3 Balance sheet items for each NUTS1 region ................................................................ 30 Worksheet 4 Balance sheet items for each NUTS3 region ................................................................ 31 Worksheet 5 Summary of balances and loading per hectare for NUTS3 regions ............................. 31 8.5 Producing estimates at any geographic scale ............................................................................ 31 9. Output and interpretation of results ............................................................................................... 31 9.1 Use of results at different spatial scales..................................................................................... 31 National level ................................................................................................................................ 32 NUTS1 ........................................................................................................................................... 32 NUTS3 ........................................................................................................................................... 33 9.2 Use of GIS & mapping ................................................................................................................. 33 9.3 Modelling policy scenarios ......................................................................................................... 33 9.4 Other models/tools for analysis ................................................................................................. 34 Details of the main models and systems are given in Annex 7. ........................................................... 34 10. Areas for further work .................................................................................................................... 34 10.1 Improved estimates offtake from pasture ............................................................................... 34 10.2 Regional level fertiliser use data .............................................................................................. 35 10.3 Improved estimates of straw use ............................................................................................. 35 10.4 Regional level coefficients of atmospheric deposition ............................................................ 35 11. Summary of key conclusions .......................................................................................................... 35 11.1 Approach/methodology ........................................................................................................... 36 11.2 Data issues ................................................................................................................................ 36 11.3 Quality of specific data components ........................................................................................ 36 11.4 Geographic scale ...................................................................................................................... 37 11.5 System issues ............................................................................................................................ 37 12. Summary of key project developments and outputs ..................................................................... 37 12.1 National level balance sheet system and estimates ................................................................ 37 And resulted in the following outputs; ................................................................................................ 38 12.2 NUTS1 level balance sheet system and estimates ................................................................... 38 Developments at the NUTS 1 level: ...................................................................................................... 38 Outputs at the NUTS 1 level: ................................................................................................................ 38 12.3 NUTS3 level balance sheet system and estimates ................................................................... 38 Developments at the NUTS3 level:....................................................................................................... 38 Outputs at NUTS3 level: ....................................................................................................................... 38 12.4 Wider benefits to Defra ............................................................................................................ 38 Annexes ................................................................................................................................................ 39 Annex 1 Summary of TAPAS Work Plan ............................................................................................ 40 Overall aims ...................................................................................................................................... 40 This work will be accomplished by: .................................................................................................. 40 Review of current methodology ........................................................................................................... 40 Review of data sources..................................................................................................................... 40 Physical data ......................................................................................................................................... 40 Expected results of the action for which the grant is requested ..................................................... 41 Brief timetable of the action for which the grant is requested ....................................................... 41 Annex 2 Copy of Fact Sheet for Nitrogen Balance IRENA indicator (first page only) ........................ 42 Key message ......................................................................................................................................... 42 -3-


Annex 3a - Flowchart of OECD spreadsheet system ............................................................................ 43 Annex 3b Alternative flowchart of OECD spreadsheet system ......................................................... 44 Annex 3c Flowchart of amended version of OECD spreadsheet system........................................... 45 Annex 3c Flowchart of amended version of OECD spreadsheet system........................................... 45 Annex 4a Review of variability used in sensitivity analysis ............................................................... 46 Livestock manures ........................................................................................................................ 46 Total manure withdrawals ........................................................................................................... 47 Harvested crops and forage ......................................................................................................... 48 Biological Nitrogen fixation .......................................................................................................... 48 Seeds and planting material ......................................................................................................... 48 Atmospheric input ........................................................................................................................ 48 Denitrification ............................................................................................................................... 50 Summary........................................................................................................................................... 50 Annex 4b Results of sensitivity analysis ............................................................................................ 52 Annex 5 Other systems in related areas; nutrient management, NVZ, diffuse water pollution....... 61 UK systems/models ...................................................................................................................... 61 European Systems ............................................................................................................................ 63 Annex 6a Results, National level ....................................................................................................... 65 Annex 6a Results, National Level....................................................................................................... 66 Annex 6b Results, Country Level ....................................................................................................... 67 Table 1 Nitrogen, 2000 .................................................................................................................. 67 Table 2 - Nitrogen, 2007 ................................................................................................................... 68 Table 3 - Phosphorus, 2000 .............................................................................................................. 69 Table 4 - Phosphorus, 2007 .............................................................................................................. 70 Annex 7a - NUTS1 level results ............................................................................................................. 71 Table 1 - Nitrogen, 2000 ................................................................................................................... 71 Table 2 - Nitrogen, 2007 ................................................................................................................... 72 Table 3 - Phosphorus, 2000 .............................................................................................................. 73 Table 4 - Phosphorus, 2007 .............................................................................................................. 74 Annex 7b NUTS1 level results ........................................................................................................ 75 Annex 8 - NUTS3 level results, maps of NUTS3 areas for England ....................................................... 80 Annex 8 - NUTS3 level results change matrix .................................................................................... 84 Annex 9 Contribution to the project ................................................................................................. 85 Defra Staff......................................................................................................................................... 85 External experts................................................................................................................................ 85

-4-


1. Executive summary 1.1 Main aims of the work planThe work plan is broad in scope and aims to assess the overall approach, evaluate the effectiveness of the current spreadsheet system, check the quality of the data currently compiled for the UK estimates and improve the overall reliability of the balance sheets. The objectives can be summarised as being to Understand what information is required for producing national and regional level nutrient balances for the UK and to develop a framework for pulling this together. Assess the current availability of data; identify the main gaps and ways in which they might be filled. This will consider how far it is possible to use data already available to meet Eurostat requirements. Develop and test any new approaches.

1.2 Key project phasesThe key stages of the project were:Evaluation phase This describes the: overall GNB approach; current methodology (OECD); current spreadsheet system developed by OECD; quality of data sources; data currently provided by the UK within the OECD system. Recommendation phase This aims to: Identify any improvements to approach and methodology; Identify improvements to data sources; Identify improvements to the IT systems used to produce estimates. Development/Implementation phase This is intended to: Implement improvements to approach and methodology; Implement improvements to data sources for UK national level balance sheets; Develop IT systems for producing balances at NUTS1 and NUTS3 level; Develop methods of presenting results at NUTS1 and NUTS3.

1.3 ConclusionsThis project has identified a wide range of issues covering policy needs, methodology, data sources and presentation. The key conclusions from the study are outlined below.

1.3.1 Approach/methodology

-5-


The Gross Nutrient Balance (GNB) approach is a generally sound method of estimating overall environmental pressures from nutrient loadings to agricultural soils. The current methodology does not consider loss pathways, how the nutrient loadings are lost from the soil and whether, for example, they impact on air quality or water quality. The balances provide an estimate of total annual loadings but do not attempt to quantify the cumulative or long term impacts of these annual loadings. The current GNB approach does not take into account factors that can profoundly affect nutrient levels such as animal housing systems, feed regimes and methods of applying manures. As these are likely to be the areas where policies are intended to have an impact, the nutrient balances, unless further developed to incorporate such factors, should not be used to judge the efficacy of such measures; they could be described as being policy insensitive. There is considerable scope to link the nutrient balances with a range of work in related areas. This would allow a more efficient use of the source data that is common to each system and help ensure consistency. The main areas/systems that are linked to nutrient balances are: o greenhouse gas emissions inventory o national Nitrogen and Ammonia budgets o use of manures in biogas and anaerobic digestion o use of agricultural soils as a sink for sewage sludge

1.3.2 Data issues Balance sheets are a complex system with a number of diverse data sources. At a national level most of the data is from reliable and well-established data sources. Data sources need to be carefully checked: balance sheet systems and results should be subject to thorough review every few years. Coefficients represent an unusual form of official statistic, requiring careful management and documentation. The UK supports the proposal by Eurostat to set up a library of coefficients.

1.3.3Quality of specific data componentsA number of specific data quality issues have been identified in this project. The key issues are: Estimates for offtake from pasture are based on average pasture yields and an assumed rate of grazing. Pasture represents a dominant component of the overall offtake and improvements in these estimates would greatly improve the overall accuracy of the balance sheets. National average estimates of fertiliser use are currently used in the calculations at all spatial scales. Due to regional variation in agronomic practice, balance estimates would be improved if reliable regional application rates for each crop type could be estimated from the British Survey of Fertiliser Practice. The land to be used in the scope of the balance sheets must be correctly defined. If unfertilised land is included, the balance sheets will underestimate the total nutrient loadings. Reflecting this, the UK excluded land identified as rough grazing from the balance sheets.

-6-


Manure is assumed to be applied to the same parcel of land on which the livestock are grazed/reared. This assumption is robust at aggregated levels but may not be valid at finer spatial scales, particularly at a holding level.

1.3.4 Geographic scale The reliability of the data sources and therefore of the balance sheet estimates reduces, the finer the geographic scale. At national level the results are fairly robust. The coefficients will represent average values across all regions and errors introduced by random variation will to a large extent cancel out. At NUTS1 level, the estimates are less reliable but still provide a broad indication of nutrient pressures and allow comparison between regions. At NUTS3 level the GNB approach too simplistic to provide robust estimates of actual loadings. More process-based models populated by purposely collected data need to be used at finer scales (e.g. catchments). NUTS2 might represent the ideal scale at which to produce balance sheet estimates. However, standard statistical outputs are not currently produced at NUTS2 in the UK.

1.3.5 System issues Defra has developed a simplified methodology and system to produce balances at NUTS3 level. Variables correlated with input and off-take (e.g. total cattle) are used to disaggregate pro rata the appropriate component of the NUTS1 level balance (total N in manure from cattle). To produce NUTS3 balances using the full OECD method requires processing of large volumes of data. Reliability limitations of many data sources at this level raise questions over the value of using this approach. Ideally, results should be produced for geographic units relevant to policy needs. However, this requires:o a system that can process holding level data o all data sources to be fully geo-referenced Currently only livestock numbers and crop areas are available at a holding level.

1.4 Results - Project developments and outputs 1.4.1 National level balance sheet system and estimates Corrected the scope of the system by excluding rough grazing; Improved data on nutrient offtake from the disposal of straw; Identified and resolved a range of anomalies in the data in the UK system; Identified and corrected errors in cell links and calculations; Improved design for IT system, reducing volume of data entry and improving consistency of data.

UK balance sheet for 2000 to 2007.

-7-


1.4.2 NUTS1 level balance sheet system and estimatesDevelopments at the NUTS 1 level: All the improvements to national level system apply to NUTS1 system. Outputs at the NUTS 1 level: Development of a system for producing balance sheets for all NUTS1 regions based on the OECD system Balance sheets for UK NUTS 1 regions for 2000 to 2007 (not previously available)

1.4.3 NUTS3 level balance sheet system and estimatesDevelopments at the NUTS3 level: Design of a spreadsheet system to derive estimates at NUTS3 level based on disaggregating NUTS1 results Development of capacity to produce GIS outputs and year on year change matrix of balance totals Outputs at NUTS3 level: NUTS3 level maps of N balances for England for 2006 and 2007 Year on year change matrix for England, 2006 to 2007

-8-


2. Introduction to the TAPAS work plan 2.1 Aims of the work planThe work plan is broad in scope and aims to assess the overall approach, evaluate the effectiveness of the current spreadsheet system, check the quality of the data currently compiled for the UK estimates and improve the overall reliability of the balance sheets. Understand what information is required for producing national and regional level nutrient balances for the UK and to develop a framework for pulling this together. Assess the current availability of data, identify the main gaps and ways in which they might be filled. This will consider how far it is possible to use data already available to meet Eurostat requirements. Develop and test any new approaches. A full summary of the TAPAS work plan is in Annex 1.

2.2 MethodsThis project is broad in scope and covers a range of topics and disciplines. It required a detailed understanding of the following key issues: OECD approach to GNB Policy requirements for nutrient balances data science behind nutrient management the environmental issues related to nutrient management farm practices that are relevant to nutrient loadings range of data sources and quality issues To develop the necessary understanding of these issues, expert advice and views were sought from a range of relevant experts within and outside Defra: Defra policy leads on soils, nutrient management, water quality, air quality Experts from NDPBs, Centre for Ecology and Hydrology (CEH), Institute of Grassland and Environmental Research (IGER) Experts from commercial consultancies ADAS1 Statisticians/data owners responsible for physical data on livestock numbers, crop areas and yields, fertiliser usage Significant time was spent becoming familiar with the current spreadsheet system developed and adopted by OECD and Eurostat.

2.3 Key project phasesThe key stages of the project were:Evaluation phase overall GNB approach1

www.adas.co.uk

-9-


current methodology (OECD) current spreadsheet system developed by OECD quality of data sources data currently provided by the UK within the OECD system

Recommendation phase Identify any improvements to approach and methodology Identify improvements to data sources Identify improvements to the IT systems used to produce estimates Development/Implementation phase Implement improvements to approach and methodology Implement improvements to data sources for UK national level balance sheets Develop IT systems for producing balances at NUTS1 and NUTS3 level Develop methods of presenting results at NUTS1 and NUTS3

3. Introduction/background to nutrient balancesThe methodology that forms the basis for this project is the standard approach developed by OECD and adopted by Eurostat. Details of the system are contained in the OECD/Eurostat Handbook2 The OECD system calculates nutrient loadings for nitrogen and phosphorus to agricultural soils. It is a fairly complex system based on wide range of data sources. Nutrient inputs and off-takes are estimated by applying coefficients to physical data. The physical data covers livestock numbers, crop areas, crop yields and fertiliser use. The relevant coefficients have been developed by empirical research by experts (e.g. ADAS) within a large programme of research projects. The nutrient balances represent a generic analytical tool relevant to air quality, nutrient management, water quality, and greenhouse gas contribution to climate change. The overall balance (kg of N per ha) gives a useful headline figure of potential environmental pressure. This can be used as a high level indicator, allowing trends over time to be monitored and comparisons to be made between countries. The strengths and weaknesses of the nutrient balances are considered in detail in the following section 4. Evaluation of Gross Nutrient Balance approach. Nutrient balances are of direct relevance to a number of European directives including the Air Quality Directive 3, the Water Framework Directive4 and the Habitats Directive5. Nutrient balances are therefore an important part of the evidence base to a number of European and international organisations including Eurostat, DG AGRI and DG Environment and the European Environment Agency.OECD/Eurostat Gross Nitrogen Balances Handbook, December 2003 Directive 2008/50/EC of the European Parliament and of the Council of 21 May 2008 on ambient air quality and cleaner air for Europe 4 2000/60/EC of the European Parliament and of the Council establishing a framework for the Community action in the field of water policy 5 Council Directive 92/43/EEC of 21 May 1992 on the conservation of natural habitats and of wild fauna and flora3 2

-10-


One of the agri-environment indicators developed under the IRENA program, 18.1, is on gross nitrogen balances. The indicator fact sheet identifies links with a number of other IRENA indicators: Input indicator links: IRENA 08 - Mineral fertiliser consumption IRENA 13 - Cropping/livestock patterns IRENA 14 - Farm management practices Output indicator links: IRENA 18sub - Ammonia emissions IRENA 19 - Emissions of methane and nitrous oxide IRENA 30 - Nitrates/pesticides in water IRENA 34.2 - Share of agriculture in nitrate contamination The first page of the fact sheet is in Annex 2.

4. Evaluation of Gross Nutrient Balance approach 4.1 Understanding the policy needs for Gross Nutrient BalancesThe nutrient loadings on agricultural soils represent a serious potential environmental threat. There are a number of possible pathways by which the nutrients can be lost from the soil. Each of these loss pathways can lead to a different environmental problem. From consultation, it was clear there is strong interest from nutrient management, air quality, greenhouse gas emissions and water quality policy areas in Defra. Indeed, each of these policy areas has funded significant research in areas closely related to nutrient management in recent years.

A standard unit of measurement across all policy areasA key benefit of the nutrient balance approach is that it provides a single common unit to measure environmental pressure that is relevant to all policy areas. This benefit should not be underestimated. Where policy responsibility is split between different areas there is always a risk of tensions between different policies and an improvement in one area may come at the cost of another area. By ensuring a common unit of measurement is used across air and water quality it should be possible to develop coherent policies based on a net improvement in the full range of impacts.

Environmentally relevant geographic unitsOne of the key issues that emerged from consultation is the need for environmentally relevant geographic units for calculating nutrient balances. The NUTS nomenclature is based on administrative boundaries that have limited relevance to agriculture or the environment; consequently these geographic units are not the most useful for agri-environment policies. The most appropriate geographic unit will vary depending on the specific policy area.

For river water quality, river catchments and Nitrate Vulnerable Zones offer the most relevant units. For air quality, ammonia emissions are of greatest concern for their potential impact on sensitive habitats, causing eutrophication as a result of nitrogen deposition. The UK has -11-


developed a classification system known as National Character Areas6 that would provide suitable geographic units

Agriculturally relevant unitsThere is also much value in using agriculturally relevant units. Farm level balances will always be challenging to estimate because of the large number of data items required. Geographic units are less directly relevant because the nutrient balance for a given area reflects the type of farming activity taking place rather than any inherent spatial property of that area. Therefore, calculating estimates by farm sector would be more policy-relevant in terms of targeting measures at improving farming systems.

Meeting specific policy needs and timescalesWhen considering policy needs for data, it is important to balance the amount of resource required to produce data with its value. Data must be fit for purpose by meeting a number of quality criteria including reliability and timeliness. Given the significant resources required to produce a full set of nutrient balances at a fine spatial scale there may be scope to tailor the specific nutrient balances outputs to meet specific data needs at a particular point in time. For example a full set of results at a fine spatial scale could be produced every few years to provide a baseline or for monitoring at key milestones such as the deadline for a Framework Directive. For less policy critical intervening years a simplified approach could be adopted or results could be produced only at a more aggregated geographic level. To be used as part of the evidence base for developing and monitoring agri-environment policy the nutrient balances should be sensitive to the very policies whose impacts they are measuring. For example, if a policy is introduced to change housing systems to reduce the loss of ammonia, it must be possible to monitor the impacts of that policy in order to gauge its effectiveness. It is a key weakness of the current nutrient balances approach that most of the farm practices that have significant impacts are not reflected in either the data or coefficients used in the balance calculation. Indeed, there is a dearth of statistical information about these, making it impossible to incorporate them using current data sources.

Presentation and interpretationFor policy customers to make the best use of nutrient balances, consideration needs to be given to issues of presentation and interpretation. At a national level, the overall balance figure provides an indicator which can be presented in a chart as a time series to show state and trend, both long- and short-term. NUTS 1 balances can be shown as a table, a series of charts or using maps. There is limited value in mapping at Member State scale given the relatively low number of data points, although at EU scale such maps would have value. At NUTS3 level a map is almost essential to present the results. With over 100 data points in the UK it is not possible easily to show overall results or trends using charts or tables, whereas a map can clearly show key trends and important geographic patterns.

6

http://www.naturalengland.org.uk/ourwork/landscape/englands/character/areas/default.aspx

-12-


Whilst the main policy interest may be in the overall balance, individual components of the balance sheet may also be of value, particularly in specific policy areas. For example the nutrient inputs from cattle manure would be directly relevant to policies aimed at reducing diffuse water pollution from dairy farms. Another valuable role for the nutrient balance system is in scenario modelling. Key data items can be adjusted to answer what if questions, for example, what would be the impact of a 50% decline in dairy farming in the South West. More details on presenting the results are given in section 9. Output and interpretation of results

4.2 Evaluating the OECD Gross Nutrient Balance approachThe nutrient balances represent a generic analytical tool relevant to air quality, nutrient management, water quality, and greenhouse gas contribution to climate change. Were they an accurate measure of the true levels, the overall balance (kg of N per ha) would give a useful estimate of overall environmental pressure. This could be used as a high level indicator, allowing trends over time to be monitored and comparisons to be made between countries. There are several other systems and models that have been developed for nutrient management. These generally focus on answering more specific policy questions. Compared to these systems, the GNB approach has both strengths and weaknesses.

4.2.1 Strengths Robust and reliable approachThe overall GNB approach is in principle conceptually and methodologically sound. It is largely underpinned by reliable and established data such as livestock numbers and crop areas which are collected according to standards laid out by Eurostat.

A generic and flexible measureThe overall balance provides a high level pressure-type measure that can be relevant to a wide range of policy contexts.

Open and transparentAlthough a number of calculations are performed on a large volume of data, it is straightforward to see what data is used and how the calculations are done. There are likely to be clear relationships between trends in the overall balances and the most influential source data for example an increase in surplus might be traced back to a corresponding increase in livestock numbers. Correspondingly, changes that are locally significant may be obscured by more dominant data.

Impartiality and integrityDue to a combination of other factors described in this section, the GNBs have the advantage of being produced with impartiality and integrity, key qualities for official statistics. They are not susceptible to interference from, for example, political expediency. Assuming a consistent approach can be assured, there is also limited scope for subjectivity in their interpretation.

-13-


Easy to interpretThe overall balance is a simple, single value expressed as a loading per unit area, in kilogrammes per hectare. In terms of level of measurement, it is a ratio measurement on a continuous (rather than discrete) scale and has a non-arbitrary zero. Differences and ratios between arbitrary pairs are meaningful. This gives the balance some very useful properties when comparing one unit with another or a change over time for a given unit. For example an increase from 20 kg/ha to 40 kg/ha can be meaningfully compared with an increase from 50 kg/ha to 70 kg/ha (same absolute increase) or from 50 kg/ha to 100 kg/ha (same percentage increase). In keeping with a desirable criterion for any indicator, the GNB is simple to interpret in terms of whether things may be improving or getting worse.

Allows benchmarking between units/countriesBy taking a standard and consistent approach, meaningful comparisons can be made between units (regions) and between countries.

Time series data availableData sources for the balances are well established so historic data is generally readily available on a consistent basis for the UK. This allows a consistent time series to be produced to allow analysis of trends over time.

4.2.2 Weaknesses Large volume of data and processingCompared with most official statistics the system requires a large volume of data to be brought together from a diverse range of sources. Processing is mathematically simple but takes place on a large scale so care is needed to be accurate.

Data sourcesReliability of the data varies between the different sources. For some data sources, national level estimates are reliable but become unreliable at finer spatial scales (crop production for minor crops). For other data sources there are breaks in the time series (cattle numbers). Details are given in the section 5.2 of this report.

CoefficientsThe input and offtake coefficients are quite different to most sources of official statistics. They are derived from empirical research and their provenance can be hard to determine and document. Improvements in the estimates of coefficients generally require applying retrospective revisions to historic data to produce a consistent time series. The proposal by Eurostat to develop and maintain a library for the coefficients for all Member States would be advantageous in this respect. .

Needs clear co-ordination and single providerThe complexity and range of data sources create scope for inconsistencies if the balance sheets are produced by different institutions. To avoid this Defra should take the lead in the UK and act as a single authority in developing and maintain the system and its outputs.

-14-


Large error marginsThe physical data and coefficients have error margins that introduce fairly large confidence intervals around the estimates. This is particularly an issue at a fine spatial scale because the precision of data is reduced for small area estimates, and the coefficients are estimated at national scale and often will not reflect local conditions. As a result, care is needed in policy interpretation and decision making.

Lack of flexibilityIn some areas the current system can be considered too prescriptive. Where coefficients are applied to physical data there should be flexibility to use whichever categories can be supported by the most reliable data. Livestock categories, for example, do not need to be identical for every country. As long as there is consistency to how the coefficients are applied to matched physical data the balance estimates will be consistent and comparable.

Lack of focusThe generic and flexible nature of the balances can also be a limitation. It provides less relevant and specific data for particular policy purposes. This is reflected in the fact that policy areas have sponsored more elaborate and focussed tools in the form of complex models and databases. These tools provide answers to much more specific policy questions and can include a range of other factors such as financial costs to farmers.

No estimate of loss pathwaysNitrogen surpluses can be lost to air or water and can impact on air quality, water quality, and damage sensitive habitats from nitrogen deposition and acidification. However, the GNB approach does not provide any estimation of these loss pathways or eventual impacts. The assessment of loss pathways is a separate issue both in data requirements and conceptually. Estimating losses requires data on a wide range of issues on farm practices and soil management as well as environmental data such as weather, topography and hydrology. There is currently little data available on some of these factors in the UK. The main farm practices and other factors not taken into account by the current approach are: Livestock breeds Feed regimes Housing systems Slurry storage systems Manure and fertiliser spreading - methods, timings in relation to weather

No assessment/interpretation of cumulative effectsThe GNBs provide an annual estimate of total nutrient loadings but one of the key environmental issues is the long term impact on nutrient levels and the cumulative effects of a surplus over many years.

5. Develop and appraise options at National level 5.1 Current OECD system

-15-


The current system has been created by OECD and is based on a series of linked worksheets in MS Excel. It is logically structured and being based in Excel, it is relatively easy to use and is clear and transparent. It requires a significant amount of data to be entered and the format does not lend itself to automated data transfer. The system, like any complex calculation based on spreadsheets, is hard to verify and quality assure. In summary, the current system features: One spreadsheet for national level balances for N One spreadsheet for national level balances for P Produces a time series Uses same format for N and P Produces a full itemised balance sheet This spreadsheet system is summarised in a flow chart in Annex 1. The open and transparent format of the Excel system does increase the risk that cells will be accidentally modified, so formulae may be overwritten or links between worksheets can be changed. There is scope to improve the usability by improvements in style and presentation for example using colour. Cells could be colour coded to indicate whether data is keyed directly or calculated via formulae. Worksheets could be colour coded to show where they link together. Formulae cells could be protected to defend against accidental modification. Details of how the current system can be developed and alternative approaches are described in 8. IT systems for calculating balances.

5.2. Data issues 5.2.1. Physical dataThe system relies on bringing together a large volume of physical data, summarised below: Livestock numbers by category (head) Crops areas by category (ha) Crop production estimates (tonnes) Fertiliser use (tonnes) Land use (ha) Disposal of sewage sludge on farmland (tonnes) Many of these data items are collected from annual surveys run, or managed, by Defra; the latest quality assured data have been used in the calculation of time series of balances reported here. Crop area data Reliable estimates of crop and livestock data come from the Farm Structure Survey and represent an annual snapshot from June each year. Data is available at a holding level. Data is also available from a survey in December but this offers no additional benefits because it only provides another snapshot at a different point in time and is based on a smaller sample size. Crop production data Figures are based on area, yield and production estimates submitted to Eurostat each year. For the main crops, these figures are collected from a sample survey and are reliable at NUTS1 level. For minor crops, estimates are only reliable at a national level. Generally speaking, estimates for minor crops at a

-16-


finer spatial scale will be self-correcting due to the dominance of production in key areas. The overall yield for a crop will be mainly determined by the yields achieved in the regions where production is highest. Regions where production is low may have inputs and/or yields that deviate from the average but the low levels of production mean that this has little impact on the overall balance estimates. Livestock data Reliable estimates of crop and livestock data come from the Farm Structure Survey and represent an annual snapshot from June each year. Data is available at a holding level. As with crop areas, data is also available from a survey in December but this offers no additional benefits. Population profile data has been available since 2005 for cattle, taken from the Cattle Tracing Scheme. An example is given in figure 1. Policy interest is in the total loading from nutrients over the course of a year, including whether there are any peak loadings, for example at a given point in the growing cycle for crops. Population profile data would, in theory, provide useful additional data on variations throughout the year. However, the current methodology is not sophisticated enough to take such variations over the year into account. In effect it uses average estimates for every parameter and assumes the loading is distributed evenly throughout the whole year. Given the limitations of the current approach, profile data will give only slightly improved estimates of an annual total where production cycles are annual and changes in livestock numbers follow a regular pattern. The potential benefits of this type of data are greater in the event of external shocks such as disease outbreaks.

Figure 1 Population profile data for cattle taken from the Cattle Tracing Scheme Matching data categories

-17-


The categories used within the nutrient balances system are not always consistent with those for which physical data are available. This is particularly an issue for livestock data. Where there is not a direct match, the best possible match must be made, thereby introducing a degree of approximation. This is illustrated in figures 2 and 3 which show the categories for pigs in the FSS and nutrient balance sheets. A possible solution to this problem would be for the nutrient balance sheet categories to be based on the categories in the FSS. In the longer term, one possible approach would be for FSS data to be collected using the nutrient balances categories, reflecting the growing importance of environmentally relevant data and the diminishing importance of market management data needs. This would of course lead to a loss of comparability with earlier data and a break in the time series.

Total pigs Breeding pigs Female breeding herd Sows in pig Gilts in pig Other sows Boars being used for service Gilts 50kg and over not yet in pig but intended for breeding Barren sows for fattening Other pigs (liveweight) 110kg and over 80kg and under 110kg 20kg and under 50k under 20kg Figure 2 June Survey of Agriculture categories for pigs Total pigs Piglets Pigs 50kg Breeding pigs >50kg Boars Sows Other pigs Figure 3 Nutrient Balance categories for pigs Offtake from pasture/fodder production data Estimates of offtake from pasture are currently very crude. These apply an average pasture yield (of 8t/ha) to both permanent and temporary pasture to estimate a production figure. This yield is assumed constant for all years, although this is very unlikely to be realistic in view of annual variation in weather. This is then adjusted by a utilization factor to reflect the fact that not all pasture is grazed due to "wastage" and where grass is removed for conservation (hay or silage). Wastage is caused by treading, -18-


poaching and fouling of grass during grazing. Based on expert opinion an overall figure of 70% has been used for UK estimates for both temporary and permanent pasture. Recent expert opinion suggests more accurate estimates of utilization rates are nearer 75% for permanent grass and 85% for temporary grass but these need to be assessed before they are adopted. One approach to estimating fodder production is to use a feed balance. This is based on a model which assumes:(a) Meat/milk production (b) fodder + (c) feed If we were to have reliable estimates of total meat and milk production and animal feed in the form of compounds and hay and silage it would be possible to estimate the fodder component. The yield and grazing proportion are both constant for all years and an average value is used for all regions and countries. There is likely to be annual variation plus significant regional variation in yields which is not currently accounted for because a national average yield is used. Recommendations for improving the quality of the offtake from pasture are given in section 10.1 Improved estimates Fertiliser use data Reliable estimates of fertiliser usage are available at a country level from the British Survey of Fertiliser Practice7. Estimates are also available at a NUTS1 level but these are derived by applying average application rates for specific crops to NUTS1 data on crop areas. This approach provides reliable estimates at fairly coarse spatial scales but not at fine spatial scales. Further work will be done outside the scope of this project to improve estimates at a finer spatial scale. Gaseous emissions during housing Losses of N will take place during livestock housing and storage of manures through volatilisation and denitrification. Some of this loss may ultimately return to agricultural soils through N deposition. As the GNB approach is a soil balance, any gaseous losses during housing are not accounted for. The approach requires that atmospheric deposition is only from non-agricultural sources. If gaseous emissions during housing/storage could be estimated, this particular loss pathway could be included in the balance sheet. It would then need to be accounted for elsewhere in the balance sheet by including agricultural sources of atmospheric deposition. In effect this can be seen as a double-entry on the nutrient budget which makes no overall difference but provides a useful estimate of a specific loss pathway. Nutrient input from manures Nutrient inputs from manures will reach the soils via two routes. The direct route is when manure is voided by animals in situ whilst grazing. The second route is where manure and slurry is collected in housing systems and subsequently spread on the land. The OECD method assumes that all nutrients from livestock are input onto the land where those animals are reared. This assumption is sound for voided manure. Where livestock are housed, there may be movement of the manure off farm or from one farm to another. Statistical estimates for movement off farm are available only where the manure is used for combustion to produce power. No estimates are available for movement between farms but there is good anecdotal evidence that this does not take place over large distances or on a large scale7

http://www.defra.gov.uk/FARM/environment/land-manage/nutrient/fert/bsfp.htm

-19-


because of the low monetary value of manure. This assumption is felt to be sound, particularly at a gross spatial level. At fine spatial scales (e.g. holding level or 10km grid) however, this assumption may not be tenable and may introduce quite significant errors into the calculations. Accurately defining the scope of the balance sheets As the OECD handbook identifies, the balance sheet result should be related to the area of agricultural land which is potentially fertilised, to avoid a bias in the result for countries with large extensive and not utilized areas. In theory defining the scope of the balance sheet should be straightforward but in practice this is not so straightforward due to the limitations of the available data. What is required is a standard criterion for defining fertilised land, which is to be in scope for the balance sheet. Ideally, this criterion could be applied in a consistent way across all countries. Secondly, having defined the land in scope, it is necessary to identify those livestock that are reared on that land. Various approaches to this were considered in this study, the most realistic options identified were: All agricultural land Exclude land in Less Favoured Areas (LFA) Exclude rough grazing LFA was not considered to be an appropriate criterion. The designation is largely based on policy-related issues and lacks consistency across member States. In the UK, rough grazing is defined as unimproved grazing that is not pasture, meadow or lowland grass. This provides a very useful and relevant way of defining the scope of the balance sheets. We investigated the feasibility of identifying livestock grazed on rough grazing land. There are key data limitations that prevent this being done. Data is collected in the June Survey of Agriculture to estimate the number of livestock and cropping and land use. However, there is no direct link made between areas of land and where the livestock are grazed. It is therefore not possible to identify the number and type of livestock reared on the rough grazing land. Even if such data were available from the June Survey, this would only relate to the position on the snapshot date in June. Livestock movements during the year, which are known to take place on a significant scale, would not be recorded or reflected in the estimates. These data limitations have an impact on the balance sheet estimates. Overall, the nutrient balances will slightly overestimate the total nutrient loadings. At a national level this will not be significant. At regional level, only those regions with significant areas of rough grazing will have balances that are overestimated. The overestimate will generally only apply to the manure from sheep component of the balance sheet. To quantify the scale of errors for England balances: Rough grazing accounts for about 12% of total grassland Manure from sheep accounts for 13% of total N from manure from livestock Manure from sheep accounts for 5% of total N inputs Another scope issue is how to deal with uncropped land, either left fallow or set-aside. Given that neither land use will be fertilised, including the land area would give a misleading estimate of the balance per hectare as, in effect, the nutrient loading would be spread over a larger total area. Fallow and set-aside land should there be excluded from the balance sheets. A given parcel of land may move in and out of scope of the balance sheets from year to year but this is not a problem. This approach is consistent with the real world outcomes linked with nutrient surpluses, given that putting more land into fallow would not reduce the nutrient loadings on the fertilised land and would not reduce the risk of environmental damage.

-20-


-21-


5.2.2. CoefficientsCoefficients are used throughout the system to convert physical parameters such as number of livestock into nutrient levels. The main coefficients are applied to derive estimates of the following components of the balance sheets: Level of N input from livestock Level of N input from mineral fertilisers Atmospheric deposition of N N input from seeds N fixation from legumes etc. N offtake from crop production The input and offtake coefficients have a number of unusual properties that make them quite different from most sources of official statistics. This project considered a number of these properties and its implications for managing the data. Data quality The coefficients are derived from empirical research and their provenance can be hard to determine and document. It is difficult to assess the quality of the coefficients or to estimate ranges and error margins. Coefficients used in the calculation of the balances reported here are derived from various sources including OECD/Eurostat default values current at 2008, results of recent scientific research (much of it commissioned by Defra and published in reports accessible from http://randd.defra.gov.uk/, and the scientific literature. The provenance of the coefficients used in this project has been documented and the values and metadata will be made available to Eurostat to be included in the proposed EU library. Revisions It is important to make a clear distinction between where a coefficient is revised to provide a better estimate and where it is updated to reflect a genuine change. Where a better estimate becomes available, the improved estimate will need to be applied retrospectively to historic data to produce a consistent time series. Matching coefficients with physical data It is essential to derive the coefficients to be consistent with the available physical data to which they will be applied. The coefficients for manure from livestock must be matched so as to be consistent with livestock categories from Farm Structure Survey data. The coefficients are derived to take into account systematic variation between different categories (e.g. age, size, weight) within a species. As a result there is not always a direct match between the coefficient categories and the categories for the livestock numbers. An approximation is made in the matching which introduces errors into the calculations. Livestock data is based on a snapshot estimate for a specific point in time. Exact livestock numbers over a course of a year will vary as a result of production cycles and empty sheds. The coefficients have been derived to take this into account. If livestock data is collected on a different basis (e.g. population profile over the whole year) the coefficients would need to be adjusted to account for this. Coefficients must also be matched correctly with the physical data in terms of moisture content. Crop production estimates will be based on a specific moisture content and this must be taken into account to ensure the coefficient is applied on an equivalent basis.

-22-


Variability in the coefficients The coefficients represent average values, around which there is known to be significant variation. This variation will be both geographic and over time and is driven by a range of factors. Some of the variation is systematic (e.g. pasture quality) and others more random (e.g. feed regimes). If any of these variations are built into the coefficient values, it is important to be transparent about which factors are included. In practice, it is likely to prove difficult to take the random factors into account. The more systematic factors could be taken into account by the use of regional level coefficients but this would entail substantial additional experimental studies and data collection. Where there may be variation over time, it may be difficult to make clear whether the change in the value of the coefficient over time reflects a genuine change (e.g. driven by improved farming practices) or whether it is a better estimate from new information. One approach to improve the transparency would be to use the same coefficient each year but apply an adjustment to take into account variation as a result of a correlated factor. The nutrient input from a dairy cow will vary with the milk yield. Rather than change the manure input coefficient every year to account for variation in milk yield, it would be more transparent to apply an adjustment factor to the coefficient. Other adjustment factors could also be applied in a similar way. The final coefficient would be different from the original coefficient but it would be apparent what adjustments had been made to it. Harmonisation Bearing in mind the issues outlined above, the proposal by Eurostat to develop and maintain a library for the coefficients for all Member States has considerable merit. To allow accurate and meaningful comparisons across different countries, it is essential to ensure a consistent approach to how the coefficients are developed and used.

5.3 Sensitivity analysisSensitivity analysis was carried out to determine relative contribution of data components and assess the potential envelope of the estimated nutrient balances given the likely range of variation in key coefficients. Much of the variation in coefficients, especially those relating to manure and fodder, relates to farming practices. Consequently, the results of the sensitivity analysis are helpful in considering the fitness-for-purpose of the balances and the spatial scale at which they are sufficiently robustly estimated to reliably inform policy. This analysis will also help inform priority for future developments, including improving quality of data items and coefficients. Information on ranges for coefficients was found in the scientific literature and in reports of research commissioned by Defra on aspects of the nutrient cycle. Information on the sources and range used is recorded in the spreadsheets supplied. The review of the approach (see annex below) clearly illustrates the many uncertainties and the variability present in many of the coefficients and data items. Many of the coefficients of nutrient content of animal excreta strongly depend on the nutrient content of their feed. The nutrient content of the resulting manure is then very dependent on the animal housing, collection and storage. The nutrient delivered to land varies according to the method of delivery (broadcast or injection) and the weather. All these factors lead us to the view that the OECD/Eurostat approach is too simplistic to provide robust estimates of actual loadings especially at small spatial scales. To perform the calculation, either a simple average must be used, or it will be necessary to find some way to form a weighted average were there to be any reliable information as to the relative numbers of animals fed various diets, housed in different conditions, manure storage, and other key items with substantial variability. This would add another layer of complexity and it is questionable as to whether the data would be available or robust. Clearly, the finer the spatial scale, the more important it would be to account for the variability in feed, husbandry, etc as these are known

-23-


to vary in different regions according to local conditions and practices. It is an open question as to whether the OECD/Eurostat approach should be used only as an index at the largest spatial scales with more process-based models populated by purposely collected data used at finer scales (e.g. catchments). Indeed, we should seriously consider whether the OECD/Eurostat approach is still scientifically defensible in the light of the very considerable effort and investment in model-based approaches both in the UK and across the EU. An excellent review of these has been sponsored by Defra: Nil Impact - (CSF models project) - ES02048. This report (and associated database of models) draws together information on 180 models (of which 75 are UK). The report makes a number of pertinent recommendations that are relevant to the approach that Defra should take to the coordination of data and models in this area, and hence to the approach that should be taken in a European context also. The European Commission has funded a number of large collaborative projects in nutrient science under Framework 6; results of these should also inform EU policy on nutrient balance statistics. The evidence sources used by Defra policy to support their work are more often drawn from the recent scientific research sponsored by the Department (amounting to tens of millions of UK pounds) than from the OECD/Eurostat data, which has not to date been robust, timely, or at a geographical scale relevant to decision making. It is doubtful, given the issues mentioned above in respect of variability and availability of data and the OECD methodology itself, whether a fine-scale version of the GNB would be regarded as fit-for-purpose by either the scientific community or by policy. Results from the Sensitivity Analysis are presented in Annex 4 and clearly demonstrate the substantial contribution in England to the balance from animal manures and forage/pasture. These key elements are also the least well-known and have potential for the greatest variability according to farm practice and environment. The spatial variation in key environmental variables such as weather, soil and topography (and hydrology) across England will lead to the regional pattern being different from the national average, with very substantial differences at small spatial scales such as NUTS3. The variability in farm practice that determines the actual N contributions from these key components will also vary spatially and according to the management of the farm enterprise in that sector of the livestock industry mediated by regulation, economics and farmer behaviour (such as willingness to adopt new better practices). As there are few, if any, data on these practices it is currently impossible to incorporate them into a more robust balance calculation. These are also the key issues where policy seeks to modify the impact of agriculture on the environment but, in the current framework, it is not possible to gain any insight into the success of policies in this area from the GNB statistics.

6. Develop and appraise options at NUTS1 level 6.1 Data IssuesThere are few issues for the UK specific to producing estimates at NUTS1 level. Data available at a national level are generally also available at NUTS1 regions. However, NUTS1 regions are at too coarse a spatial scale to attempt to match them with other geographic units relevant to environmental or agricultural issues such as river catchments or farm sectors.

6.2 System IssuesThe key issue in determining the best system is a trade-off between reliability and ease of production. The standard spreadsheet system developed by OECD produces a time series at national8

http://randd.defra.gov.uk/Default.aspx?&SearchText=es0204

-24-


level. To produce balance sheets for each NUTS1 region (11 in the UK) requires either a separate spreadsheet for each region, or to produce results for each region but for only one year. This adaptation can be made relatively quickly and easily. There are various methods available to simplify the approach to calculating the nutrient balances. Such methods involve making assumptions or using approximations. For example it would be possible to use aggregated livestock or crop categories to reduce the amount of calculations. Another approximation would be to estimate the balances for small units by disaggregating the balances calculated for larger units. However, at NUTS1 level it is hard to justify using any form of simplified approach as this would inevitably lead to a reduction in the quality of the estimates whilst only delivering a very modest reduction in resources required. There are a number of issues to consider relevant to the use of IT systems to produce estimates at NUTS1, covered in more detail in section 8.3 NUTS1 level

6.3 Example of a simplified approachAs part of a project on nutrient balances for water policy, ADAS developed a simplified approach at NUTS 1 level for UK9 . This approach was based on deriving a net balance per unit (ha, animal) at a national level. A national level soil balance and farm gate balance is needed to derive these pro rata values. The pro rata coefficients are available only for aggregated livestock categories (e.g. total sheep, total cattle). These unit values can then be applied to livestock and crop area totals at any scale to estimate the total nutrient balance. The method allows fine-scale estimates to be produced relatively quickly. Overall, however, it is felt that the benefits of a simplified system are relatively modest and outweighed by the loss of reliability and the additional need for a farm gate balance. Another weakness of this approach is that the unit values would need to be updated fairly regularly, if not annually, and that the unit values cannot be calculated without a national level balance.

7. Develop and appraise options at NUTS3 level 7.1 Data issuesBeing based on administrative boundaries, NUTS3 regions are not the most useful or relevant agricultural/environmental unit for which to produce nutrient balances. However, as a standard administrative unit, much of the underlying data is readily available. By contrast, to produce estimates for any non-standard units would require full geo-referencing for all the data that feeds into the balance sheet calculations; this is not currently available. Eurostat has expressed a strong interest in the feasibility of producing NUTS3 level results. This is understandable given that the NUTS nomenclature provides a consistent geographical framework across all member states. NUTS3 can be seen as a case study for a range of other geographic units at a similar spatial scale. It is likely that any issues identified at NUTS3 level will be applicable for other units at a similar spatial scale. It is worth noting that NUTS2 may represent a more practical level for the UK, providing a better compromise between spatial scale and reliability/availability of data. However, standard outputs are not currently produced at NUTS2 in the UK.

Enter keyword WQ0106 at http://randd.defra.gov.uk/Default.aspx?Location=None&Module=FilterSearchNewLook&Completed=0

9

-25-


To be of greatest value to policy customers, balances should be calculated at the same spatial scale as the observed variation, such as 1km grid scale. The key issue for NUTS3 level results is whether the physical data and coefficients are reliable enough to produce estimates that are fit for purpose? The sources of data and their availability at NUTS3 is summarised in table 1. Date item Source Crop areas FSS Crop yields Area, yield and production Livestock numbers FSS Fertiliser use Fertiliser Practice Survey Fodder consumption Estimated Table 1 Data sources and availability at NUTS3 NUTS3 Yes No Yes No No

The most important data not available is: Crop yields Fertiliser use Fodder consumption Given the increasing pressure to drive down the costs and reduce respondent burden of collecting data it seems unlikely that reliable estimates for these components will be available in the foreseeable future. Some form of estimation of these data items at NUTS3 level will be needed, perhaps by disaggregating data collected at a larger spatial scale.

7.2 System issuesA spreadsheet system has been developed to produce NUTS3 level balance sheets using a simplified system. Without the simplified approach it is doubtful whether a system could be developed in Excel. See section 8.4 NUTS3 level results for further details.

7.3 Proposal for a simplified methodThe simplified approach is based on NUTS1 level results. For each NUTS1 region, the main balance sheet items are apportioned out among the NUTS3 units within that region. This apportioning is based on a highly correlated variable for each balance sheet item. For example, the total N from cattle is based on the total number of cattle within each NUTS3 region. This approach is based on an assumption that the breakdown of cattle by category will be the same for each NUTS3 unit within the NUTS1 region. There is a reasonable level of homogeneity within each NUTS1 region in the UK, so this assumption is reasonably sound and should not introduce large error margins. A similar method of disaggregation could be applied directly to national level balance sheet components rather than NUTS1 level data. Whilst this would avoid the need to calculate NUTS1 level results, this method introduces significantly greater error margins. There is much less homogeneity at a national level than at NUTS1 level so for example, the breakdown of cattle categories may be very different between, say, the dairy dominated South West and the beef systems in the uplands of the North. The correlation between total cattle and manure produced from those cattle will therefore be weaker. The charts below show the relationship between the results using NUTS1 regions as the base data (horizontal axis) and UK level results as the base (vertical axis). They show a good overall correlation but there are significant variations, including for those regions with high total loadings.

-26-


N loading (kg/ha)180 160 140 120 100 80 60 40 20 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 N loading (kg/ha)

N balance (kg)60,000

50,000

40,000

30,000N balance (kg)

20,000

10,000

-

-

10,000

20,000

30,000

40,000

50,000

60,000

70,000

-27-


The spreadsheet system developed to produce estimates with this approach is described in detail in section 8.4 NUTS3 level results

8. IT systems for calculating balances 8.1 Requirements of IT systemThere are a number of properties that any IT system should ideally possess to ensure the system is easy to use and produces reliable results. Maintenance and development Data input Automated links to source data where possible Flexible and able to receive data from other statistical systems in their standard output format to reduce pre-processing requirements Allow easy input of alternative values for scenario testing Its design and structure should be as simple as possible The calculations being performed should be clear and transparent Easy to use and maintain Make efficient use of data through links and shared data sources to minimise replication of data entry

Data output Flexible design to produce balances at different geographic levels Generic structure to allow use by any country

8.2 National level dataAt a national level there is a relatively modest volume of data to deal with. Survey estimates of physical data are all available at national level. Current OECD spreadsheet represents a good compromise between ease of use and power. If results are only required at a national level this approach probably represents as good a design as any and is fit for purpose.

8.3 NUTS1 level dataAlthough the OECD spreadsheet system works well at national level it becomes relatively cumbersome to produce estimates at finer spatial scale. There are three dimensions (region, year and balance sheet item) that cannot be easily accommodated in a system built up from two dimensional tables. There are different options to include the third dimension, shown below. Option 1 is to create a separate version of the spreadsheet system for each region, of which there are 11 in the UK. Option 2 is to create a separate version for each year and use a column for each region, where currently each column is for a different year. Option 3 is to stack all the tables for each region in one worksheet in the current format.Option 1 Format presenting each NUTS1 region

-28-


NUTS region 1 2003 N inputs N offtake N balance N per ha

2004

2005

2006

2007

Option 2 Format presenting all regions for a given year

Year = 2007Region1 Region2 Region3 Region4 Region5 Region6 Region7 Region8 Region9 Region10 Region11

N inputs N offtake N balance N per haOption 3 Format presenting all NUTS1 regions in one worksheet

2003 NUTS region 1 N inputs N offtake N balance N per ha NUTS region 2 N inputs N offtake N balance N per ha NUTS region 3

2004

2005

2006

2007

Options 1 and 2 provide a more manageable structure, whereas Option 3 keeps the full NUTS1 system in one spreadsheet and so is more compact. Option 1 is better for comparing results within a region but across several years. Option 2 is better for comparing results for a given year across different regions. Recommendation: Options 1 or 2 seem to provide the best balance between ease of use and presentation. The UK has set up a system based on option 1.

8.4 NUTS3 level resultsWith over 100 NUTS 3 regions in the UK, the volume of data is too large for it to be practicable to replicate the full OECD methodology at this scale in spreadsheets. In addition, the spreadsheet system is too cumbersome and would require either 100+ spreadsheets (option1), spreadsheets with over 100 columns of data (option 2) or worksheets with up to several thousand rows (option 3). Of these, the only approach that might be workable is option 2. However, due to data issues discussed in section 5.2. Data issues, it is not recommended that the full OECD system is used at NUTS3. Section 7.3 Proposal for a simplified method describes a simplified approach to producing NUTS3 estimates. This approach makes efficient use of the available data and reduces the amount of calculations required. As a result the processing can be achieved in a relatively simple spreadsheet -29-


system. This system uses 5 worksheets to produce NUTS 3 level balance sheets and a summary table for a specific year. 1. Worksheet 1 contains physical data on crop areas and livestock numbers at NUTS3 level. 2. Worksheet 2 calculates the proportion of the NUTS1 total for each parameter that each NUTS3 region accounts for e.g. Darlington accounts for 18% of the total cattle in NUTS1 region North East. 3. Worksheet 3 contains the full itemised balance sheets for each NUTS1 region. 4. Worksheet 4 applies the proportions from the relevant parameters in worksheet 2 to estimate each balance sheet item for each NUTS3 region e.g. Darlington accounts for 18% of the total cattle in North East so it will also account for 18% of the manure from cattle. 5. Worksheet 5 provides a summary of the balances and loadings per hectare for all NUTS3 regions. Worksheet 1 NUTS3 level data on crops and livestock Region Wheat BarleyNORTH EAST HARTLEPOOL AND STOCKTON-ON-TEES SOUTH TEESSIDE DARLINGTON DURHAM CC NORTHUMBERLAND Etc.

Permanent grass

Total crops

Total area

Worksheet 2 NUTS3 proportions of NUTS1 totals Region Wheat BarleyNORTH EAST HARTLEPOOL AND STOCKTON-ON-TEES SOUTH TEESSIDE DARLINGTON DURHAM CC NORTHUMBERLAND Etc.

100% % % % % %

100% % % % % %

Permanent grass 100% % % % % %

Total crops 100% % % % % %

Total area 100% % % % % %

Worksheet 3 Balance sheet items for each NUTS1 regionN Inputs Fertilisers Manure from cattle Manure from sheep Manure from pigs N Outputs Cereals oil crops Balance

NUTS1_1 NUTS1_2 NUTS1_3 NUTS1_4 NUTS1_5 NUTS1_6 NUTS1_7 NUTS1_8

-30-


Worksheet 4 Balance sheet items for each NUTS3 regionN Inputs Fertilisers Manure from cattle Manure from sheep Manure from pigs N Outputs Cereals oil crops Balance

NUTS3_1 NUTS3_2 NUTS3_3 NUTS3_4 NUTS3_93

Worksheet 5 Summary of balances and loading per hectare for NUTS3 regionsN Inputs NUTS3_1 NUTS3_2 NUTS3_3 NUTS3_4 NUTS3_93 N Outputs Balance Loading per ha

8.5 Producing estimates at any geographic scaleIf a system is to be created to produce results at any chosen geographic scale then a spreadsheetbased system is not recommended. The key data requirement is to store the full range of physical data at a fine spatial scale with full geo-referencing. These data items could then be aggregated into appropriate spatial units, requiring a flexible system. In practice this is likely to require holding-level data for most variables. As a result a database system such as MS Access should be used. It would be relatively straightforward to develop an Access-based system. This would be less prone to errors in specific cells but would require skills in Access to understand. It would require users to build up significant expertise in the system to use it competently. The final balance sheet could be produced in the same format as the OECD system. The intermediate outputs used to perform the calculations would not be the same however, so consistency would be lost with the OECD approach. This could be overcome by producing a series of queries that replicated the OECD approach but doing so would lose some of the benefits of a more flexible and powerful database system.

9. Output and interpretation of results 9.1 Use of results at different spatial scalesThe nutrient balance sheets consist of a significant number of components, shown in figure 5 below. NITROGEN INPUTS Fertilisers Inorganic Nitrogenous Fertilisers Total Organic Fertilisers (excluding livestock manure) Net Input of Manure -31-


Livestock Manure Production Cattle Pigs Sheep and Goats Poultry Other Livestock Withdrawals Change in Manure Stocks Manure Imports Other Nitrogen Inputs Atmospheric Deposition Biological Nitrogen Fixation Seeds and Planting Material NITROGEN OUTPUTS Total Harvested Crops Cereals Oil crops Pulses and Beans Industrial Crops Other Crops Total Forage Harvested Fodder Crops Pasture BALANCE ( Inputs minus Outputs) Figure 5 balance sheet items The practical use of the balance sheets will depend on the spatial scale and the corresponding number of units for which results are available. Where there are only few units (e.g. national level) the full details of the balance sheet can be assessed. By contrast, where there are a large number of units (NUTS3), it becomes impossible to comprehend the full set of balance sheet items and more aggregated results must be used. The areas of interest and relative benefits at each scale are summarised below.

National level A single headline balance in kg per hectare Overall balance for a range of years to give a time series Allow comparisons between countries Relative importance of different balance items e.g. manures v- fertilisers Maps are uninformative for a single country Maps useful for comparing across EU

NUTS1

-32-


An overall headline balance for each region Overall balance for a range of years to give a time series Allow comparisons between regions Comparison of different balance items between regions e.g. cattle in South west v- pigs and poultry in Eastern region Maps add value for comparing across EU Maps add little value for a single country

NUTS3 Too many units to consider full balance sheets Focus on overall balance (kg/ha) as headline figure Look at trends in overall balances over time and between regions Maps become preferred form of presentation Data can also be summarised as frequency distribution Year on year change can be summarised in change matrix

9.2 Use of GIS & mappingThere is clearly an important spatial dimension to nutrient balances. The use of GIS offers a powerful method of analysis and presentation, particularly at finer spatial scales such as NUTS3. There are various options available for mapping the results: Total loadings per ha Ranked loadings Variation from a mean loading or critical loading Change over time between a reference year and year of interest (e.g. change between 2000 and 2007) The choice of geographic unit is critical in providing relevance to policy use and interpretation of nutrient balance data. Maps represent a particularly powerful method of presenting the nutrient balance results. They can also present results in a format that helps mitigate the problem of poor reliability of the estimates at fine spatial scales such as NUTS3. The maps can use fairly broad size bands for the nutrient loadings. These bands could also be given descriptive labels rather than the exact threshold values (see table). This approach should help prevent over-interpretation by users of the data. Nutrient loading (kg/ha) Descriptive Category

nutrient balance

Documents

system issues

gross nutrient balances

current system

data issues

spreadsheet system

current oecd system

tapas work plan

physical data