General enquiries on this form should be made to:



|General enquiries on this form should be made to: |

|Defra, Science Directorate, Management Support and Finance Team, |

|Telephone No. 020 7238 1612 |

|E-mail: petitions@defra..uk |

|SID 5 |Research Project Final Report |

• λ Note

In 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.

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

λ ACCESS TO INFORMATION

The 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 |ES0204 |

2. Project title

|Agricultural decision support tools for prediction and management of nutrient |

|input and loss: NIL impact |

|3. Contractor |Glasgow Caledonian University |

|organisation(s) |Cowcaddens Road |

| |Glasgow |

| |G4 0BA |

| |      |

| |      |

| |54. Total Defra project costs |£ 17.988 |

(agreed fixed price)

| |5. Project: start date |      |

| | end date |31 May 2006 |

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 Summary |

7. 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.

|The aim of this desk study was to collate and summarise information on those computer models and model-based decision support tools most widely|

|used in the prediction and management of nutrients input and loss, both in the UK and more widely. For a key set of these models, to be |

|determined by the project sponsors, information was collected on their developers, main function, component models, intended audience, actual |

|number and type of user, location of source code/model, inputs and outputs, and estimate of current success. The data collected from this |

|activity would then enable the generation of recommendations which could be agreed with industry representatives. |

| |

|While the broad aims of this work have been met, there have been several changes to the scope of the project which have had a direct impact on |

|the approach used and on the shape of the objectives, and ultimately on the shape of this report. These changes were: |

|A broadening of the focus of the model collection from those concerned with N,P and K losses to soil and water, to include those which deal |

|with gaseous emissions (Ammonia and Nitrous Oxide) and those which cover animal housing or manure management. |

|A greater number of models for which detailed information was required. The initial search for models generated far more than anyone had |

|expected (70 UK models/DSS) and it proved impossible and impractical to reduce the set to a number that would have allowed a very detailed |

|analysis. |

|A broadening of focus for the vision of integrated models. The initial focus on nutrient inputs was also seen to be too narrow by the experts |

|at the workshop where discussions made it apparent that modelling of many more areas is needed to answer the types of questions posed by policy|

|users. |

| |

|Model information collation exercise |

|Data on models was gathered using keyword search of web-based data sources and a snowball approach. This was followed up by personal telephone,|

|email or postal contact with model owners to add to and correct data from the initial search. The material was collated in an Access database |

|using a simple structure consisting of two tables. The database contains 180 records including 75 UK models. The database can be searched by |

|model name, nutrient and keyword and permits the viewing and printing of several standard reports. |

| |

|Recommendation generation exercise |

|Twelve organisations were invited to send representatives to a workshop to discuss the feasibility of meeting policy and land management |

|requirements with existing models and platform delivery systems and the potential for the integration of models. Thirteen people attended the |

|workshop, held in Birmingham on 16th May and a large number of issues were discussed and a set of recommendations agreed. The output of this |

|workshop was circulated and comments taken. The final set of recommendations drawn up from this and other activities within the project are as|

|follows: |

| |

|Recommendations for further scientific research. |

|In order to provide missing information to allow scientific institutes to answer the questions identified by Defra units the following areas |

|should be investigated. |

|The links between nutrient loss from land and impacts on fresh water and coastal ecology. |

|Ecological recovery following reductions in nutrient pollution of fresh and coastal waters (long term study). |

|The processes that occur at the boundaries between well-modelled areas i.e. between soil/groundwater and surface waters and between surface |

|waters and coastal waters. |

|The integration of erosion and soil movement models with catchment area nutrient models |

|The feasibility of upscaling models from field to catchment level |

|The impact of organic farm practices on nutrient loss |

|The potential for incorporating urban models on water run off from hard surfaces with farm level nutrient loss models. |

|The establishment of consensus on baselines and targets for water quality |

|The combined impact of farm practice and nutrient inputs on nutrient loss at farm level |

|The updating of models for delivery to land manager to include changed farm practices and new crops and varieties |

|The linking of socio-economic factors into nutrient models at the catchment and national level. |

| |

|Recommendations for data gathering, data sharing |

|To improve the availability of data necessary for the long-term provision of models and model based decision support tool the following should |

|be provided by Defra, or through Defra programmes. |

|A single source of quality data about farm practice that contains individual farm data and includes those variables of specific relevance to |

|nutrient modelling. |

|A single low cost source of quality, model appropriate weather data (climate and daily) for land management model delivery systems and a tool |

|set which allows models to extract data at an appropriate timestep. |

|A mechanism for linking forecast weather data from commercial sources into land management model delivery systems. |

|A system for collating information on the impact of nutrient reduction activities and ecological recovery over time |

| |

|Recommendations on model integration and delivery |

|To facilitate a gradual move towards the integration of models and model delivery systems the following additional actions are recommended: |

|To continue to support technical delivery systems for both policy and land management as they stand but work towards a single vision for |

|delivery which includes: |

|1) The establishment of a mechanism for obtaining consensus, nationally (and possibly on a European scale) on standards for model development |

|at all levels. These standards should cover all aspects of model development that would allow models of different types to work together in an |

|integrated system e.g. parameter naming conventions, units of measurement, development of scenarios etc. |

|2) The establishment of a mechanism for obtaining consensus on standards for model and decision support system delivery for land managers. |

|Consensus needs to be obtained on appropriate recommendations (e.g. for nutrient inputs), appropriate display conventions (e.g. presentation of|

|uncertainty), and other delivery specific areas common to a wide range of models and sectors. |

|To start the process of the standardisation and integration of models in a single well defined area (e.g. N recommendations for land managers) |

|but ensure that this is agreed as a starting point by the wider community. |

|To support the education and support of land managers in the effective use of models and model based decision support systems. |

|To continue discussions with crop management software developers to identify mutually acceptable approaches to data sharing. |

|To initiate wider and more focused dialogue between model developers and policy users on policy requirements and model design. |

| |

|Recommendations on use of non-agricultural and non-UK models |

|Initiate cross-disciplinary workshops to define approaches to answering catchment scale questions. While modellers within a given area appear |

|to be highly cognisant of the existence and value of models from other countries, it would be useful to bring modellers who normally operate |

|within distinct disciplines (e.g. forestry, upland management, fresh water ecology, coastal water ecology, farm management, etc.) together to |

|discuss integrated approaches to answering Defra questions. |

|Initiate discussions with the team managing UKWIR/EA Project CL/06/C: Effects of Climate Change on River Water Quality; Phase 4 – Water |

|Quality Modelling to ensure a consistency of approach to model integration and delivery. |

| |Project Report to Defra |

8. 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 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:

λ the scientific objectives as set out in the contract;

λ the extent to which the objectives set out in the contract have been met;

λ details of methods used and the results obtained, including statistical analysis (if appropriate);

λ a discussion of the results and their reliability;

λ the main implications of the findings;

λ possible future work; and

λ any action resulting from the research (e.g. IP, Knowledge Transfer).

1 Introduction

Modern intensive farming practice requires that nutrients are applied to agricultural land to ensure a competitive and profitable supply of food products. However, not all of these nutrients are retained in the soil or taken up by plants. A lack of knowledge about the sources and levels of nutrients in the soil can lead to over-application and consequent run-off with serious implications for the environment. In 2002 the Policy Commission on Farming and Food reported that 70% of nitrates and 40% of phosphates in English waters had their origin on agricultural land [1]. Agriculture also accounts for nearly 90% of the UK’s national emissions of ammonia, originating from livestock housing, the storage, treatment and application of all types of animal manures and the use of inorganic fertilisers. When ammonia eventually returns to the ground it can lead to the eutrophication and acidification of land and water, causing damage to sensitive ecosystems.

The activities of agriculture in relation to nutrient pollution from all sources are increasingly under scrutiny. The European Water Framework directive requires all inland and coastal waters within defined river basin districts to reach at least good status by 2015 [2]. Limits to national emissions of NH3 have been determined as part of the Gothenburg protocol to reduce atmospheric pollution [3]. In addition the EU Integrated Pollution, Prevention and Control Directive (IPPC) will require large pig and poultry units to begin reducing their emissions of NH3 by 2007.

Tools which allow policy makers and land managers to monitor and manage nutrient pollution are therefore becoming increasingly important. Models (conceptual, analytical and numerical), help to identify, explain and tackle problems and can be valuable in presenting the potential problem to land-managers and policy-makers [4]. Decision support tools based upon models can extend this support by presenting key information in a decision appropriate fashion and by using computer power to search though many thousands of options for viable management solutions.

A wide range of models, and several computer based DSS, have been developed in the UK in recent years which have the potential to support the management of nutrients within agriculture and to support those involved in policy creation and implementation. While some of these are well known within the circles within which they were funded and developed, knowledge of the full range of other models which might potentially support the needs of the policy and land manager has to date been limited.

1.1 Scope of the project

The aim of this desk study, as stated in the original CSG7, was to collate and summarise information on those computer models and model-based decision support tools most widely used in the prediction and management of nutrients input and loss, both in the UK and more widely. For this key set, to be determined by the project sponsors, information would be collected on their developers, main function, component models, intended audience, actual number and type of user, location of source code/model, inputs and outputs, and estimate of current success.

The data collected from this activity would then permit the generation of a set of recommendations for future research and development, indicating where there could be opportunities for the integration or expansion of models or where non-agricultural models may be of potential value within the UK agricultural sector. The results of this investigation were to be validated with a panel of representatives from key research agencies.

While the broad aims of this work have been met, there have been several changes to the scope of the project which have had a direct impact on the approach used and on the shape of the objectives, and ultimately on the shape of this report. These changes were:

• A broadening of the focus of the model collection from those concerned with N,P and K losses to soil and water, to include those which deal with gaseous emissions (Ammonia and Nitrous Oxide) and those which cover animal housing or manure management.

• A greater number of models for which detailed information was required. The initial search for models generated far more than anyone had expected (70 UK models/DSS) and it proved impossible and impractical to reduce the set to a number that would have allowed a very detailed analysis.

• A broadening of focus for the vision of integrated models. The initial focus on nutrient inputs was also seen to be too narrow by the experts at the workshop where discussions made it apparent that modelling of many more areas (see section 4.1) is needed to answer the types of questions posed by policy makers.

2 Approach adopted

There were two main tasks to be completed within the context of this project:

1) Model information collection

2) Recommendation generation

The approach used to complete each is described below.

2.1 Model information collection

The aim of this task was to identify and catalogue key nutrient models and model based DSS. Other model collation exercises have resulted in searchable databases: for example the Register of Ecological Models maintained by the University of Kassel and the German National research centre for Environment and Health [5] and the River Basin Manager's Toolbox [6] developed as part of the Benchmark Models for the Water Framework Directive project within the 5th Framework Programme. These collections, whilst extremely useful, lack some of the information that it was deemed important to collect and a new database was proposed. This task was broken down into a number of smaller objectives, the methods used to meet each of which are described below.

a) Identify all relevant UK based and key international models and model-based DSS;

A series of keyword based searches of internet information sources were undertaken. These included general searches using Google and Google Scholar and specific publication-based sources such as the ISI Web of Science, the sites of key journal publishers, other databases of models and the publications and final reports areas of the Defra, EA, SEERAD and UKWIR websites. The ‘snowball’ approach to data collection was adopted where references and key-words from one source were used to provide the direction for the next search activity.

In addition, those individuals who were contacted for additional information on their models provided names of other models and modellers in their own or related areas.

b) Agree sub-set of models to focus project effort on

This activity was dropped after the initial search identified seventy models and it became apparent that there wasn’t a useful way to reduce the list without potentially losing valuable information. The data collection component of the project was therefore expanded but at the expense of some of the detail.

c) Identify name, author, main function, component models, intended audience, actual number and type of user, location of source code/model, inputs and outputs for each;

Publications, web-pages, entries in other databases, and personal contact provided the input for these values in the database. As the data sources were diverse and not always under the ownership of the model developer the entries were sent to each developer for checking and correction/expansion.

d) Where possible identify success of UK models, in terms of number and type of user and number of hectares of land influenced;

e) Identify where possible the stated success criteria or reasons for failure;

Where possible the data on success of models was gleaned from the literature and from contact with the model developers. The initial plan included detailed discussions with the model authors about their systems but these were not possible given the large number of models involved. A partial solution to the contact problem was provided by circulating a first draft of the database contents to authors and requesting their co-operation in completing the information set and in checking its validity.

f) Store all information in a searchable Access™ database

A basic database to house the information was designed and created using Microsoft Access. A simple interface to facilitate basic searches was created to allow use by those not familiar with the Access software. The interface and search tools were designed for use by the direct clients of this research in the pursuit of the aims of this project and were not subject to any detailed requirements analysis.

2.2 Recommendation generation

Having collected data on the relevant models the next stage of the project was to make recommendations based on the information gathered. The objectives and the methods used to meet them are described below:

a) Identify those areas where there is potential to bring together disparate models within an integrating DSS structure;

b) Draw attention to those non-agricultural models that might be useful;

c) Suggest areas for further development and expansion of models;

d) Obtain agreement on main recommendations from UK scientific research organisation representatives.

The initial plan for these objectives was to take the detailed results from the limited set of models and to examine ways in which they might be brought together within an integrating DSS structure. Initial ideas for the integration would be discussed with the individuals from the organisations represented by the model developers. A final workshop would then be held in which the key recommendations would be agreed and detailed.

In the absence of a short-list of models this approach had to be reconsidered and reconstructed using a requirements analysis approach. Here an initial statement of requirements is produced and used as the basis for the generation of potential design solutions: the solutions to include ideas about the collation of models under a DSS structure, areas for further expansion and development and recommendations on appropriate ways for collation to proceed.

2.2.1. Identification of requirements

The idea of integrating disparate models within an integrating DSS structure was predicated on the belief that a group of models would be able to meet the requirements of users better than single models operating alone. In order to identify areas where integration might be useful it was therefore necessary to identify the requirements that were to be met. In decision support terms, requirements for DSS can be thought of as the questions that the user asks of a support system tool [7,8]. The first task was therefore to collate the questions that the users wanted to ask of such a system.

The policy users questions were provided by members of the Water Quality, Environmental Protection and Nutrient Management units. As direct consultation with the other group, land management users, was not factored into the original desk study, the decision was taken to obtain a view of their requirements from the sponsors. Defra can be considered a secondary stakeholder in any system that supports the management of nutrients in farm practice. While a direct consultation with primary users would be more valuable, this approach was considered reasonable for the purposes of providing an initial starting point for discussion. It also provided a useful insight into the requirements the funding stakeholders have for systems developed for land management use. As some of the requirements presented by policy were quite complex they were ‘unpacked’ by the author so that each question represented an aspect of nutrient management that might feasibly be modelled. These ‘unpacked’ questions can be seen in Appendix A.

2.2.2 Checking requirements and proposing solutions

A workshop format was used to bring key representatives from UK modelling institutions together to discuss the requirements as defined in the questions from Defra groups. Twelve organisations were invited to send one or more representatives to the event which was held on the 16th of May 2006 in a central and easily accessible location (Holiday Inn at Birmingham airport). Fifteen individuals agreed to attend and thirteen actually attended. A list of those who attended can be found in Appendix B.

The workshop was divided into a morning and afternoon session, with policy issues being discussed in the morning and those relating to land management in the afternoon. Participants were invited to attend one or both sessions. In the event all but one person attended both sessions. The workshop provided an overview of the results from the model gathering exercise. Break out groups were then asked to review the questions with a view to considering how many of them could already be answered, by which models, and where gaps existed. The whole group then discussed the issues arising from their individual discussions and these were recorded on flip-charts by the facilitator. People who were invited but did not participate were also asked to provide a response to the question sheets: three people did so.

The discussion points from the workshop and the emailed responses from the questions were sorted into themes and are summarised and discussed in the section below.

3 Results

3.1 Model database

The database is constructed in Microsoft Access™. It is a simple structure consisting of two tables, one containing the model details and the other the contact information for the person currently listed as responsible for the model. The database contains 160 records, an overview of which is provided below. The database is provided as an appendix to this report. A list of models and their contact organisations is provided in Appendix C.

3.1.1 Origin of models in the database

There are 75 UK models and 74 non-UK models represented in the database at present (there are some models of unknown origin). While it is expected that all the major UK models are contained within the database, it is likely that some less well known UK models (particularly in the areas of emissions and animal housing as these were later inclusions into the search) and a number of key-non UK models will have been excluded. Figure 1 shows the representation of non-UK models as it currently stands.

[pic]

Figure 1: Showing the breakdown of non-UK models in the database by country of origin

3.1.2 Summary of UK model characteristics

Table 1: Showing breakdown of known funding agencies for UK models by type

|Government authorities | | |Levy bodies | |

|Defra |35 | |HDC |4 |

|EA/NRA |13 | |HGCA |3 |

|SERAD |4 | |MDC |1 |

|SEPA |3 | | | |

|CEC |3 | |Water authorities | |

|DARDNI |2 | |Severn Trent Water |1 |

| | | |UKWIR |1 |

|Funding councils | | |Thames Water |1 |

|NERC |11 | | | |

|EPSRC |2 | |Others | |

|ESRC |1 | |Warwick HRI |3 |

|BBSRC |1 | |EN |1 |

| | | |Nirex |1 |

Nineteen organisations were identified as sources of funding for the UK models, this includes six Government bodies, four funding councils, three levy bodies, three water authority agencies and three other funders. These are listed with the models names in Appendix D. Not all models have identified funding sources. The vast majority of UK models not funded by Defra or an agricultural levy body were funded by the Environment Agency and its predecessor the National Rivers Authority. Twenty-three research organisations were involved in the development of the UK models, including thirteen university departments or units, these are listed in Appendix E.

3.1.3 Types of UK model

Table 2: Showing the scope of UK models in relation to nutrients modelled

|Nutrients modelled |No. of models |

|N (all types) |30 |

|P |12 |

|N (all types) &P |9 |

|N(all types), P & K |5 |

|Ammonia |2 |

|Ammonia & N2O |1 |

|P, Ammonia & N20 |1 |

|K |1 |

Table 2 indicates the scope of the UK models represented in the database with regard to the nutrient modelled. The most numerous deal with individual nutrients. The gaseous models may be under-represented due to their late inclusion in the data search.

3.1.4 Scale of UK model

Table 3: Showing the scale of the UK models in the database

|Field |Farm |Catchment |Region |National |European |Other/na |

Models can give results at more than one level.

Most of the models in the database have been designed to give results at either field or farm level (most of those that operate at field level also give results at farm level) or at catchment level.

3.1.5 Focus of UK model

Models tend to have a single area of strength or on which they are focused to provide answers to specific problems or questions. A breakdown of the primary focus of the UK models i.e. the areas of the nutrient cycles on which they specialise, are shown in table 4 below. Whilst relatively crude in the process of selection this categorisation gives some insight into the relative concentration of models in different areas. Some models have more than one area of focus.

Table 4: Focus of UK models

|Model focus |No. |

|Water diffusion (surface and coastal water) |37 |

|Soil movement/groundwater |27 |

|Plant uptake |14 |

|Emissions (animal and other) |11 |

|Animal deposits (manure) |6 |

Nutrient movement through the soil, surface and coastal waters have the greatest number of models associated with them. The low number of emissions models cannot be taken as a reflection of the relative number of these systems in comparison to the water models, as the focus was only entered into the data gathering process late and some systems may have been missed.

While some models have a broad focus which encompasses two or more of the areas listed above, no single system currently provides output across all areas of the nutrient cycle. A breakdown of the spread of modelling effort as represented in the database is given in table 5.

Table 5: Number of UK models with a multiple focus

|Model focus |No. |

|Water |43 |

|Soil and plant |21 |

|Soil and water |18 |

|Soil |15 |

|Soil, plant and water |13 |

|Plant |5 |

|Emissions |5 |

|Water, manure |3 |

|Soil, plant, emissions |2 |

|Soil, water, emissions |2 |

|Water, emissions |2 |

|Soil, plant, manure, emissions |1 |

|Soil, water, manure |1 |

|Plant, water |1 |

|Manure (spreading) |1 |

3.1.6 Areas of use

Analysis of UK models suggests that they have been designed and built for three main areas of use, policy, research and land management. Table 6 shows the number of models that were designed for use in each area and gives an indication of the nutrients that they cover.

Table 6: Models dealing with N,P & K for policy, research or land management

|Area of use |N |P |K |

|Policy |23 |15 |0 |

|Research |28 |13 |2 |

|Farm |13 |6 |4 |

In general the policy models have been developed for policy development i.e. testing ideas and looking at their impact in different future scenarios, and policy management i.e. exploring ways in which policy directives can be met, identifying areas for action and general monitoring. Research models are produced either to increase further understanding of some part of a nutrient cycle, or cycles or to deliver specific information to policy or to land management users. Land management models are developed to improve business performance and/or to support the meeting of legislative requirements. A list of the models and their use as policy, research or land management system is given in Appendix F

3.1.7 Development characteristics of UK models

Table 7: Showing known development languages for UK models

|Program Type |No |

|Fortran |29 |

|Spreadsheet |10 |

|C or C++ |5 |

|Fortran and C/C++ |4 |

|Web based |2 |

|Visual Basic + Excel |2 |

|Turbo Pascal |2 |

|Visual Basic |2 |

|ARC Macro Language |1 |

|Visual C++ |1 |

|ArcGis |1 |

|Sciconic |1 |

|Quick Basic |1 |

|Quattro Pro |1 |

|Delphi 5 |1 |

UK models have been developed in a wide variety of ways. The most common development approach represented in the database is programming using the Fortran language and the second most common approach is to use a spreadsheet.

3.1.8 Existing platforms for delivery

A platform is defined here as a group of features and functions that produces the actual or perceived integration of a number of models.

3.1.8.1 Policy platforms

This project has identified three UK systems which have been or are being developed to deliver integrated support to policy makers, these are:

3.1.8.1.1 INCA N&P

The INCA systems [9,10] were developed by Reading University to link a number of systems e.g. GIS interface, hydrological models, catchment process models and river pollution models to investigate the fate and distribution of water and pollutants in the aquatic and terrestrial environment. They model N & P nutrients and are used by Reading University to support the Environment Agency

3.1.8.1.2. MAGPIE

MAGPIE [11] was developed by ADAS as a framework in which to integrate national and catchment scale environment models, with a database of agricultural and environmental data. It currently integrates nitrate leaching models. It is used by ADAS to support the Environment Agency.

3.1.8.1.3 WIR Framework

The WIR Framework [12] is a new initiative, under the CL/06/C programme, which is being developed for use by the Water companies and the Environment Agency. Its aim is to generate an open framework for simulation models relating to impact of climate change and will include nitrates.

3.1.8.2 Land management platforms

Three platform systems designed to deliver integrated support to land managers were identified during this project, they are:

3.1.8.2.1 Agricultural Decision Support Platform (ADS)

The Agricultural DS (formerly Arable DS) [13] is an approach designed to integrate the presentation of decision support to farmers and consultants, and to allow the sharing of data resources. It supports both deep integration via the data environment or shallow integration at the user interface level and currently integrates winter wheat disease, wheat weed, oilseed rape pest and disease, grain management, nitrogen and manure DSS.

3.1.8.2.2 EMA

EMA [14] is a framework designed to bring together a suite of tools, information and assessment routines help the farming industry (all sectors) improve its environmental performance. It contains recommendations and calculators as well as an environmental auditing tool. Among a range of other areas the audit tool deals with the nutrient related areas of organic wastes, fertiliser applications; soil nutrients & pH; sewage sludge use; and lime use.

3.1.8.2.3 Morph

The Morph system [15] was designed as a platform to provide an integrated delivery and data sharing approach for horticultural models developed by Warwick HRI. The original system was designed to accommodate only HRI models but the platform is currently being redesigned and the new system will allow other models to be integrated and other interfaces to the models to be applied. It currently integrates 19 models which include bulb and top fruit pests, vegetable pests and diseases, vegetable spacing and the WELL_N nitrogen model [16].

3.1.9 Outputs from models

It is possible to use the database to provide an initial outline of the outputs from UK models. Appendix G provides such an outline in spreadsheet format. In this spreadsheet models are divided into Catchment/National level and Land Management level and the stated (and inferred) outputs from the models are listed. Models are sorted according to the minimum time step in which they run. The spreadsheet shows that while there is duplication between some outputs from models, generally where the models are derived from one another, most models have some aspects which differ from others and which answer slightly different questions. Most non-related models providing similar output also seem to do so at different scales or time-steps and therefore meet a different requirement. The areas where more than 2 models generate similar types of output are:

Catchment/national level models

• Total N (nitrate/ammonium) at any point in drainage system

• Total P at any point in drainage system

• Nitrate leaching

• N delivered to a water body from land

• P delivered to a water body from land

• Deposition of Ammonia (NH3)

• Loss from soil of P

• Contribution of land use to nitrate flux

• Contribution of agric sector to nitrate flux

• Contribution of land use to organic and inorganic-P flux

• Level of organic and inorganic-P in soil

• Level of organic and inorganic-P in groundwater

• Concentration and flux of No3 in soil solution

Land management models

• Distribution of non-adsorbed solute in soil layers

• Nitrate leaching from soil

• N uptake by plants

• Fertiliser recommendations

• Total N level in soil

• Nitrate content in the soil profile

• Ammonia loss from manure applications

• Nitrous oxide loss from manure applications

• N input to soil in plant debris

• Denitrification from ammonical fertilizers

The information in the spreadsheet should only be taken as indicative as it was based on information currently in the database, which while checked by many modellers it is still incomplete. The interpretation of model outputs as represented in the spreadsheet has also not been checked by modellers as this activity was initiated very late in the project. It also does not include models for which no information was available or for purely hydrological models.

3.2 Key issues in the integration and delivery of models

The views expressed in this section are those of the scientists who took part in the Birmingham workshop supported by the views of others who contributed by email. In some places these views are supplemented by those of the author. The discussions that generated these views were, in most cases, prompted by observations from the model gathering exercise and the questions raised by Defra.

3.2.1 The nature of models

While models are very powerful tools and have great potential to be used in the control of nutrient pollution their actual nature, and their limitations, are often only partially understood by those removed from their development. Discussions at the workshop within this project and the author’s observations from twelve years of working directly with modellers and model-based DSS suggest that an understanding of these issues is critical to any discussion of their future use. Two key issues are described here.

3.2.1.1 Models are incomplete descriptions of the world

Models are by their name and nature incomplete descriptions of the real world. The real world contains interactions between elements that are rich and highly interrelated, models on the other hand contain cut down and restricted approximations of those interactions. Some models pay attention to more elements than others and some describe interactions in more detail than others. Each model has been designed to reveal a useful pattern in one or more processes for a specific purpose. Each model has been developed using the tool set which appeared to its creators to be most appropriate for the job in hand. Models are not developed according to any single guiding principle but have been produced to meet the specific needs of those who fund their development and or to fit the specific research interests of the developer. Some areas have received a lot of attention, e.g. the nitrogen cycle, and there are a correspondingly larger number of models that deal with this nutrient than with P or K (e.g. NIRAMS [17], NEAP_N [18], NITS [19], SUNDIAL [20], WELL_N [16] etc). Other areas have received a lot less attention e.g. the relationship between nutrient levels and freshwater ecology, and there are few models that describe this area (none listed in the UK database). Models are also often created from data gathered in a particular location and may therefore only be at their most accurate for that or similar areas. Modellers too may be influenced in their approach by the environment in which they work and this may also limit the range of their model’s applicability.

The focus on specific areas e.g. the nitrogen cycle, is inevitable. Funding resources are finite and attention frequently has to be prioritised on the most pressing issues of the moment. It is also true that it is not always obligatory to understand a system completely to be able to produce useful information to inform decision making. The level of detail needed for any given task is entirely dependent on the demands of that task. Sometimes all that is needed is enough information to be able to identify the best questions, which can then be answered by other more finely focused tools.

Task focus means that each model has an area in which it functions most effectively and a number of areas in which the output is less robust. For example the DNDC model [21] is primarily concerned with Nitrous Oxide. It has the capacity to give answers about other gases but it wouldn’t be selected as a key tool for these as the output would not be as useful as models designed specifically for those gases. Most of the UK models contain sub-routines which deal with areas that are not their primary focus.

Any attempt to bring models together to answer questions has therefore to ensure that the differences, in focus, detail and scope between them, can be accommodated. It also has to recognise that the knowledge we currently have about the nutrient cycle and the role of agricultural processes within it is far from complete.

3.2.1.2 Nutrient models are based on assumptions that may, and have, changed over time and that may differ between modellers

Models do not operate in the presence of a complete dataset about activities in the real world, and often have to make an informed guess about the value of some of the variables they include. They are based on assumptions. These assumptions are well known and understood by the developer but may cause problems when the model is used as part of a decision support tool. There are several reasons for this:

1) Assumptions may not remain valid over time. For example the workshop discussion suggested that some land management models contain assumptions that a full dose of pesticide will be used. This was common practice at the time of the model’s development but is no longer the case.

2) Assumptions may not be common between developers. Models which generate answers to similar questions may have different assumed values for some parameters. Neither of these assumptions will be wrong, as such, but represent a specific and valid scientific viewpoint. They will however cause confusion by generating slightly different answers to the same question.

The use of assumptions is well understood within the modelling community. Reported understanding, and the authors personal experience from a range of DSS development project however is that this facet of models or DSS is little understood by either policy or land management users. There is also a concern that it would be difficult if not impossible to make all the assumptions within current models transparent to the user.

3.2.2 Scale and the precision/complexity of the modelling instrument.

Two levels of precision are dominant in UK nutrition modelling. There are very detailed models that focus on one specific area of the nutrient management problem (most direct application in land management DSS e.g. MANNER [22], PSALM [23]) and then there are more general models that look at a much wider view of the problem area (most direct application in policy level DSS e.g. Export Coefficient Model [24], INCA [9,10]).

The split at the science level is produced by expediency rather than a real difference. At field and farm level it is possible to get very detailed about the inputs, local weather, actual practice etc., but there isn’t enough data or computing power to scale this up. The wider view models are therefore not built up of many smaller detailed ones but instead focus on the most important information from those detailed areas. They trade precision at the micro level for answers at the macro level. They model the interactions between many of the micro areas but because they only focus on key aspects they can’t be used to look in detail at those interactions. For example a catchment scale model may show that there is a higher than desired level of nutrient in the water supply and that it comes from farming practice. It is unlikely to be able to show that the problem comes from a small sample of farm units in that catchment, or provide any indication of why they are contributing a disproportionate amount to the pollution levels. A field/farm scale model on the other hand may be able to cast light on specific farm issues e.g. the location of a stream next to a large paved area – run off into stream, but it is not currently possible to scale this approach up to generate catchment or national level data. Some information, which would be useful at the catchment level e.g. the use of buffer zones, is currently only available at the field/farm level. It was suggested that it might be possible to model specific catchments from the field up to test the impact of detailed modelling on catchment level modelling.

Many of the questions posed by the Defra units suggest that there is a desire for systems that allow the ‘zooming’ in of focus from a catchment or national scale model and the discussions suggest that it is possible to work towards this end but that it will require access to much more detailed data across a much wider area (see sections 3.2.5 & 3.2.2).

The wider view models can however be used to allow problems to be scoped and, as one of the workshop attendees suggested, they can be used to “allow policy users to identify the right detailed questions to ask”. These models can also be used by the science users to get a better overall understanding of the issues within the subject to allow them to understand the answers they get and to explain the issues to those requesting information from them.

There is another view which suggests that it may not be feasible to use models developed to predict outcomes at one level to predict outcomes at a higher level as there are general equations that exist for phenomena at each level of detail. An example from the automotive industry is the quantum electrodynamic theory which makes accurate predictions about the properties of new materials. This theory cannot however be successfully used to make predictions about the performance of a new car as other factors are involved.

3.2.3 Who should the user of a model be?

There is a concern amongst modellers that a failure to understand the two issues of incomplete knowledge and the use of assumptions will lead any non-specialist user to misunderstand the level and type of uncertainty surrounding the results of any model run. They fear that failure to understand these issues will lead to poor decision-making. A modellers reputation, and to some extent future funding, may be closely linked to the perceived success of their efforts (models) to describe and predict things in the real world. This leads to a dilemma. Users want tools that help them to complete a task and models can be used to do this. User would often like to be able to make use of these modelling tools themselves without having a third party to run them for them with the corresponding loss of time and interruption to the flow of the decision or policy formulation task. Scientists on the other hand are reluctant to give models to others to use where there is a possibility of the output being misinterpreted or over stated, as their reputations may depend on them.

There is another reason to make the uncertainty and the assumptions behind models visible to users. Land managers in particular are suspicious of ‘black box’ answers. Some individuals may want a system to tell them the answer but most want to know why and how the answer has been generated. Land management decision-making currently involves a constant balancing of uncertainties, weather being one of the greatest. Systems for use by land-managers and policy users have to allow the user to integrate an understanding of their outputs and its uncertainty with the other elements of the decision under consideration.

There was a feeling on the part of the modellers at the workshop that while it may be possible to give end-users a mental model of a simple system to enable them to use it safely and effectively it becomes increasingly difficult as the focus of the system increases. Simple systems containing one or two models, were considered feasible for use by most land-managers and policy users.

3.2.4 Issues in linking models

As models have been developed by many different people and organisations to suit many different purposes there are many differences that need to be reconciled if they are to be brought together in any integrated fashion.

Differences in assumptions: As stated in section 1.1.1.2 all models are based on assumptions and these may differ even between models of the same kind.

Differences in modelling approach: Many approaches to modelling have been used e.g. mechanistic, stochastic, statistical, export co-efficient and hybrid types. These approaches themselves make different assumptions about the nature of the behaviour they describe.

Differences in data names/measures: There is no such thing as a naming convention or an agreed set of variable terms within the nutrient modelling community. E.g. soil definitions, there are currently different definitions for workability for arable leaching and grassland models, and RB209 uses another one again. This means that passing values between models or using a common database, or even presenting a common view to the user would require translation routines or a degree of rewriting.

Differences in time-step/granularity of data requirement: Models are likely to differ in the granularity of the data they require, a common data source would have to accommodate all levels to support integration. Time-step is discussed further in section 3.2.5.2

Sector differences: Catchment level models tend to look across sectors whereas field and farm models are usually based within a sector. Sectors have very different priorities and need answers to different questions. Arable, horticultural and livestock sectors have different goals, priorities and requirements and their questions of nutrient models will necessarily differ.

3.2.5 The importance of a consistent quality data set

The quality of output and the level of granularity of a model design is to a large extent dependent on the availability of the data to feed it. In nutrient modelling this will include both farm (and other general and point source data) and weather data. Workshop attendees felt that it was necessary to understand the landscape of current practice before it was possible to predict the impact of legislation.

3.2.5.1 Farm activity data

Farm scale models may be able to access very detailed information about fine scale activities on an individual farm, particularly if that information has already been entered into a farm management system for use in other activities. Even then it is likely that additional information will have to be entered by the user to meet the specific information requirements of the model. Catchment scale models do not have access to anything like the same level of detail and cannot expect the end-user to be able to enter much in the way of other variables at run time. Neither policy nor land-management user will wish to spend much time physically entering data before being able to use a model or model-based DSS. For systems at either end of the spectrum to be successful then relies on the availability of a detailed data set of farm practice information with specific relevance to nutrient management.

There are two potential sources of farm practice information, Crop Management Systems in use on individual farm units and data gathered by Defra or other governmental bodies. The existing and potential links between nutrient models and these sources are discussed in detail in section 3.2.6.1 below.

3.2.5.2 Weather data

The vast majority of models (nutrient management and other farm practice areas) need weather data both to develop them and to run them once developed. The data may be historical (previous years weather), current (the actual data from the current year) or forecast (a projection of expected weather data for a given period ahead). There is currently no single viable source for any of these three types of data and there are issues relating to time-step, location, quality, and cost relating to all three of them.

Time step: The time-step of a model, or the gap between each iteration in its running, may vary from a minute to a year. At each time-step the model will update the values for the variables it is using in its calculation. Bringing models together requires that they operate at the same time step. Time-step also has implications for the linking of field and farm scale models, to give more accurate catchment information, as it would require a great deal of computer processing power and time to scale up from these using their current time-steps.

Location: The behaviour of many processes, including the nutrient cycle, will vary depending on the particular circumstances at any given location. Nutrient loss to water for example may occur more rapidly in one location than another based on the characteristics of geography or on the level of rainfall in that location. Field and farm models would operate more accurately if they could access weather information at that level of granularity (some at field scale), but this is often not possible as few farms operate their own weather stations and there are no wide scale networks of weather stations at the local level. The most commonly used weather data sources tend to offer regional rather than local level data.

Quality: All current weather data appears to be subject to error, caused by faulty equipment, failures in calibration or data entry problems. Historical data may also suffer from these problems if the data has not been checked and corrected. The quality of the models generating weather forecasts may also vary.

Cost: Weather data is one of the most expensive components of farm and field level models. While agreements exist which allow weather data to be used at a low price for research purposes the costs frequently become prohibitive when model developers attempt to deliver their systems to land-managers.

While these issues were recognised and an effort was made to address them in the Foresight project InterMet (HGCA-2200), the absence of an affordable source of clean, local data at relevant time-steps is still a major problem for field and farm scale models. Catchment and national models are affected far less by these issues as they are often run by research institutes for policy clients and can therefore use the cheaper research data-sets, and as they operate on a larger scale access, to local data and finer time-steps is much less of a problem.

3.2.6 Sources of farm activity data

3.2.6.1 Farm Management Systems

A large percentage of the basic information relating to farm practice at the scale that field and farm scale models require is stored in commercial Farm Management Systems (FMS) such as MultiCrop( (Farmade), CropWalker( (Muddy Boots) and Sheep Focus( or Cattle Manager ((FarmPlan). Without integration between the models and the data in the Farm Management Systems the user is faced with a high overhead of data entry whenever a model is used. The overhead of data entry has been frequently cited as a key reason for the non-adoption of a number of model based systems eg [25].

Access to the data within an FMS needs a sizeable degree of co-operation from its commercial owners. Models either have to be integrated into the software by the company, which requires effort on their part, or the data within the FMS has to be accessed by the model software developer. The latter requires the data to be exported by the FMS into a common format or a willingness on the part of the commercial company to provide the model developer with a route into the data storage area of their software. Unfortunately the agricultural software market is very small and most companies operating within it cannot afford to spend time on activities which do not directly generate revenue. While some initiatives have shown promise e.g. delivery as a no-cost DLL or Dynamic Link Library to agricultural software companies [26], FMS producers are unlikely to co-operate with initiatives to integrate with their databases unless there is a high probability of return. UK FMS producers and FM packages include: Farmade, Farmplan Muddy Boots, Pear Technology, FarmWorks, Agrosoft. Agridata and Flockdata.

3.2.6.2 Data collection by central agencies

The key source of farm practice data for catchment and national level models is data-sets collected by Defra, the Environment Agency and other central agencies for their own purposes e.g. the June Census, Agricultural Survey and the British Survey of Fertiliser Practice. The problem with using these data sets is that they were not designed to gather information for nutrient modelling purposes: they are often based on small numbers, some of the information that would be useful is missing, the level of detail may also be inappropriate for some modelling activities. As small samples are used in some surveys scaling up to catchment level can produce nonsensical answers (e.g. if three pig farmers are in a survey and one injects slurry the scaled up data might suggest that 33% of pig farmers inject slurry). Averaging also doesn’t help to build models, for example the average fertiliser input according to statistics is as per recommendation but data is needed on those units where the application is above recommendation to develop strategies to reduce pollution. It was suggested that point source pollution datasets are currently quite poor.

The workshop participants indicated that improvements in access to farm practice data were extremely important to the development of more integrated modelling approaches. The use of the Whole Farm Approach initiative was seen as a way to supply data modelling without increasing the burden on individual data suppliers. One example of the potential of this approach was provided as it was suggested that in Denmark field scale data collected centrally is used to drive models and displays which all can access via the web.

3.2.7 Sources of weather data

The main source of weather data for the development of models is the Meteorological Office. Other sources are available from independent companies e.g. Viasala whose weather stations are generally along the road network, and weather data from individual or networked weather stations placed on farms.

3.2.8 Other issues

3.2.8.1 Socio-economics

Discussions during the workshop indicated that issues more closely related to socio-economics than to soil science, ecology or hydrology have to be integrated into both field/farm and catchment level models. This is particularly true where attempts are made to change farm practice to meet policy needs. Economic variables it was felt, may be more important than climate variables in predicting change. A change in one or more economic variables for example may make hill farming untenable which would have a major impact on nutrient levels in upland areas.

On the social side consumer attitudes towards the acceptable size, shape, visible quality and production techniques for vegetables over the past twenty years have had a direct impact on the way that they are produced, including the levels of nutrient and other inputs applied.

Risk perception is another example of a social variable that is key to behaviour change. It was suggested during the workshop that land management users are more likely to respond, and change behaviour, to an indication of potential loss of income (e.g. only getting 75% of potential yield) than they are to the promise of potential increase in income (e.g. increasing yield by 25%).

3.2.8.2 Organic farming

It was suggested that little is known about the impact of organic approaches on nutrient losses, however it is not possible, from the data in this exercise to comment further on the validity of the suggestion, apart from to note that no UK models for organic operations are represented in the database.

3.3 To what extent can existing models answer the questions posed by Defra?

While the format of the workshop did not permit the clarification of the extent to which any single question might be answered, it appeared that many of them could be answered at least partially. The discussions did provide some general points on the areas where answers are more likely to be found, the issues preventing answers and areas where more work is needed.

3.3.1 Policy level – results at national/catchment and field level

Questions about current levels and losses of nutrients and the gap between estimated levels and desired levels e.g. WFD standards

Twenty-eight of the UK models contained within the database are concerned with estimating the loss of nutrients to waters at catchment level, and a number of these provide information on nutrient level concentration or load. Four models deal with loss of P from soil and 2 with loss of nutrient to the atmosphere. A number of others look at the levels in soil or groundwater and one in gaseous levels across Britain (see Appendix G for details). Workshop discussions however suggested that there was a problem in answering questions about the differences between these levels and target levels. There was a view that there is little agreement on what targets should be and that in some cases the models are needed to generate the targets in the first place. It was suggested that only when appropriate and relevant targets are agreed will it be possible to produce answers.

It was also suggested that nutrient inputs were only one small part of the picture of loss of nutrients to water and that to provide a comprehensive view required the integration of many types of models e.g. hydrology, erosion, with nutrient movement models. Models from forestry, upland management and possibly some urban modelling tools could usefully be integrated to provide a more accurate picture. The impact of nutrients should also be considered in terms of the whole environment – soil, water, air and the economy of the sector and related sectors. This would require a much wider range of models to be brought together. While some of the models in the database take a range of farm and economic factors into account (e.g. MEASURES [27]) there are none which link the full possible range of relevant variables.

There was a view that enough was known about gaseous emissions to be able to provide a framework to deal with questions at the catchment level, but not to answer the questions without some additional development work. Examples of current UK models which offer support in this area are FRAME [28], LADD [29], MEASURES [27], NARSES [30] and SCAIL [31].

Relative contribution from agricultural practice to nutrient losses

‘What if’ analysis on changing levels of one or more agricultural practices?

There are a number of models which are capable of providing answers to the first of these questions (e.g. ADAS Evenflow [32], INCA [9,10], NIPPER [33], NIRAMS [17], PSYCHIC [34], NEAP-N [35], PLUS [36] and N-Catch [37]) but the extent to which any model can currently do so is limited. It was suggested that the relationship between changing one farm practice and the impact on nutrient losses to water is very incompletely understood. While some aspects of point source pollution have been dealt with there are other areas such as run-off from farm steadings or the impact of major weather events that haven’t.

Discussion and communication from modellers suggests that more needs to be understood about the behaviour of P (and to a similar extent K) as it is transferred from soil to waterways and that this has to be linked to models of erosion and soil movement to get an accurate picture.

Changes to farm practice also have to be considered over time and the discussions suggested that we don’t currently understand enough about how nutrient levels in soil, water or atmosphere, change over time, or the impact of farm rotations on the pace of change, to provide useful answers to long term questions. Discussions also suggested that a baseline of nutrient levels was required for some of these calculations and that none existed.

Workshop discussions suggested that if there was a move towards linking the field and farm scale models together to generate more detailed answers at a catchment scale then more work would be needed to understand processes at the boundaries of model areas. For example while nutrient movement in groundwater and nutrient movement in fresh water are both areas which have been reasonably well mapped, movement between ground water and fresh water has received less attention.

It is possible at the moment to use current models to predict total pollutant loss, and loss of total N or total P but it is not possible to break it down into its different types e.g. the different types of phosphorous loss.

There is little integration at the whole system level e.g. in relation to pollution swapping. No model system can currently predict well what might happen to losses if farmers change from one livestock to another as the impact on cereal production, etc. is not factored in.

Gaseous models are also not integrated enough to be able to use in a coherent way for what if’s on policy. They won’t be able to tell the user what impact cutting back on one will have on levels of other gases. It is also not possible to sub-divide them e.g. give a value for the emission factor for wheat and then ask what would happen if we changed from winter to spring wheat.

It was suggested that the design of many of the catchment models, while appropriate for the questions they were designed to answer, are not suitable for questions that require simulation. For simulation some degree of mechanistic modelling is required and that needs data at farm and field level. The workshop felt that the lack of a detailed set of farm practice data was a key factor preventing the development of models that are capable of what if analysis (see sections 3.2.2 & 3.2.5).

There has been no discussion on the way that individual farm practices might be scaled up. Each model has its own approach but there is currently no consensus.

It was suggested that models may need to develop an optimisation function to be able to offer support for finding policy measure that manage behaviour. For example a tax on pesticides may just change what is grown rather than how much pesticide is applied (e.g. if the crop has a higher value to compensate for the tax level), changing a crop rotation pattern may have a greater impact on pesticide usage. An optimisation tool would run the model many times to present a list of possible outcomes, and may result in fewer possibilities being missed. Other views were that scenario production and expert discussion would be as effective in supporting this activity.

Impact of nutrients on ecology

‘What if’ analysis on nutrient levels and ecology

Discussion suggested that while the pollutant models at catchment level are good at indicating levels of pollutant loss from soil to air and water they are not good at representing speciation of pollutants (e.g. separating phosphorous into particle absorbed phosphorous and soluble bio-available phosphorous) which is key to understanding the impact on ecology.

Workshop discussions suggest that there is currently insufficient data describing the interactions between nutrient inputs and freshwater ecology to produce useful models to support these questions. The quantitative relationships between N and freshwater ecology are not known, so no predictive models exist. There are some basic models linking P and chlorophyll levels but not agreed models for P and macrophytes. The situation is better for rivers than for lakes.

Ongoing research on nutrient levels in Loch Leven [38] for example suggests that the picture is much more complex than originally thought, particularly in the area of recovery for water. A halving of phosphorous inputs did not have an immediate impact on the nutrient levels or on the ecology of the loch. This was because other factors, such as nutrient recycling from the sediment and the effect of increased fish predation on grazing zooplankton, had delayed the expected recovery. In this particular instance recovery has taken fifteen years. This example may be equally applicable to other areas of ecology.

Because the impact of nutrients on ecology over time is poorly understood it is currently not possible to say with any confidence what will happen to the ecology of a given system if the inputs are changed. There is limited understanding of what causes a rate of change and what the rate of change will be. This makes it particularly difficult to manage the expectations of those investing money, or those who have to change their practices, as they look for a visible indication of the impact of their efforts.

No representative of coastal ecology was available to attend the meeting and therefore discussion of this area was limited.

Cost-effectiveness of any given initiative/policy or any combination of initiatives or policies?

This is another query that is affected by the lack of quality farm practice data and understanding about the way that levels may change over time. While risk is explicitly mentioned as an output by several models, MEASURES is the only model which appears to examine explicitly cost effectiveness of initiatives at a policy level.

3.3.2 Land management

What are the crop fertilisation needs (dependent on soil type, crop, current soil nutrient levels etc.)?

‘What if’ I change this rate?

Discussions suggested that at first glance there are plenty of models available to answer questions about plant nutrition needs. Unfortunately workshop discussions suggest that many of these are now out of date and that farmers don’t believe the answers. The assumptions underlying the models need to be adjusted (eg rates of pesticides applied) and new crops and varieties incorporated. New trials would need to be carried out to generate data to incorporate these new crops and varieties. Glasshouse crops are not currently covered by these models and they may be important point sources.

In both vegetable and arable production the recommendations coming out of the older models are seen in some cases as too high and in others as too low, as they are very dependent on changes in market forces and variety characteristics. More information on the interactive effect of nutrients on growth and leaching is needed.

Workshop participants pointed out that it was important to understand what the land manager does with advice before offering it, for example application at 10% above advised levels is not uncommon.

What is the farm gate nutrient balance? (to emphasise the systems approach that should be taken)

Discussion of this question once again raised the point that the problem is much wider than nutrient inputs and that it should include all farm management activities, soil erosion, crop residues, weather, nutrient loading, etc.

While some models cover the nitrogen story, phosphate and potassium levels are less straightforward as chemical processes in the soil have as much of an impact on levels as the quantity applied or taken up by plants.

The general view from the workshop was that it was not possible, using existing models, to provide an answer to this question for all nutrients.

What is the best manure management plan to maximise crop benefits/storage/compliance?

Questions about the nutrient value of manure/storage

Questions about manure values under different feed conditions etc

It was felt that it was possible to answer most of these questions using models that already exist (e.g. MANNER [22], PLANET [39]). Some aspects of storage management are not however covered by these systems e.g. the impact of storing manure under different conditions/levels of cover for varying lengths of time.

What is the optimum application rate for lime for this type of crop at this time?

It was felt that this was a non-question as all land managers assume that soil ph will be adequate, and use top up applications of lime every 5 years or so.

3.4 What areas are definitely missing/where should work be focused?

The workshop participants agreed that the following activities would allow scientific institutes to provide answers to the questions posed by Defra and make it easier to integrate models:

• The provision of a quality data source for farm practice which contains actual farm data and includes those variables of specific relevance to nutrient modelling.

• The establishment of consensus on baselines and targets for water quality

• Longer term projects to aid understanding of the impact of actions and recovery over time

• A system for collating information on long-term impacts of actions and recovery over time

• More research on the links between nutrient loss from land and impacts on fresh water and coastal ecology

• More work on the processes at the boundaries of current modelling sectors e.g. nutrients and ecology, soil and water, fresh water and coastal water

• Updating of crop nutrition models to include changed farm practices and new crops and varieties

• Taking a broader look at nutrient management at farm/field level to include farm practice as well as nutrient inputs

• Creating consensus, nationally and possibly on a European scale on the standards for parameters, naming conventions, units of measurement etc to facilitate model integration.

• Upscaling of models from field to catchment level

3.5 Delivery

This section deals with the issue of delivering models to those who wish to use them i.e. those answering policy questions and those dealing with land management issues. It is based on the assumption that answering the more complex questions will require an integration of models at least at the level of the user interface.

3.5.1 Single models

The adoption of a single model to answer specific questions did not seem to the workshop participants to be a viable option. It was felt that it would be very difficult to identify a single model that provided everything that was needed by either land-manager or policy users. This is supported by the list of outputs in the spreadsheet in Appendix G. As discussed earlier in this document models differ from each other on a number of dimensions and a model which produces good results under one set of conditions, for one sector, may not deliver well under another. Discussions suggest that the scientific community favours a multiple model approach, to maintain progress through competition between scientists and to allow the selection of specific models to suit a given purpose. It was suggested that this approach, coupled with agreements on standards, would support a gradual convergence between models and would allow for the integration of new science. Tools which allow for the delivery of a single view on the output of multiple models should however be encouraged.

3.5.2 Single or multiple platform delivery

The results of this desk study and discussions with the industry representatives suggest that while integration of models within platforms is a good thing, it is not possible to identify a single existing platform which would meet the needs of all users within policy and land-management and across sectors. The reasons for maintaining a number of distinct platforms, at least in the near future are:

Differences in requirements

Policy and land management users have very different needs and ask different questions of models. Individual sectors at the field/farm scale also have different priorities and questions.

Differences in model types

The differences between the data requirements of different models may make it difficult to integrate some models within the same platform.

Differences in platform aims

Current platform systems have been designed to meet different aims/requirements (e.g. focus on quality in horticulture, focus on yield in arable), an agreement between these aims/requirements would be necessary before a single platform could meet the needs of the client/user for disparate systems.

Levels of investment

Considerable public funds have already been invested in delivering models using different approaches.

Although a single platform approach appeared to participants to be infeasible at the current time, movement towards integration of models is both possible and desirable. The vision is one where agreement between modellers and funding agencies on basic issues relating to data and communication allows models to be linked together as and when required to meet specific task needs, and agreement on basic display issues reduces the burden on end-users and increases usability.

3.5.3 The single vision alternative

No single technology is likely to ‘stay the course’ and it is very difficult if not impossible to predict where things are going to go in the next ten years. Picking a single track and focusing on it is therefore most likely to lead in a sub-optimal solution in at least part of the development. Flexibility is the key. The ability to respond to change quickly and move with the technology and the social demands placed on it is essential. The more flexible a model is and the more platforms or systems it can be used in the greater its chance of being adopted. An examination of any software tool developed 10 years ago, however innovative its design, will reveal the use of many communication and display conventions that are now no longer in use.

Workshop participants agreed that modellers and funding agencies need to build a common understanding of what is needed at this moment in time, but with the expectation that the needs will change.

Having a common vision is more important at this stage that a common platform. Agreeing standards that allow for the easy exchange and use of data between models and between data sources is the most useful approach to the solution of this problem. Standards allow for the integration of models as and when needed, to suit different purposes. They allow broader DSS to emerge and may lead in the long term to single delivery solutions.

3.5.4 Pace of change

The workshop participant’s experience in developing integrated systems suggests that the process is not an easy one, largely because of the differences between the models and the world view of the modellers. It would therefore be most cost-effective to start the standardisation and consensus process in one area where there are already clearly defined synergies and then roll out the standardisation to others, allowing new models to be built and old ones to be updated.

3.5.5 Problem of different answers to the same question

A key driver behind the desire for a common platform or selection of single models, for land management questions, is the desire to provide a common response. At this level it is undesirable to have a number of models all generating different recommendations in a given situation e.g. for fertiliser recommendations. The nature of models (as discussed in previous sections) however makes this almost inevitable.

If it is not possible to select a single model or platform to provide a consistent answer then the only way to deal with this issue is to get agreement between those delivering this information to the land management sector. Agreement on the models used and the way that the differences between those model recommendations might be combined or displayed to the user is essential. There aren’t many systems in active use that deal with a given nutrient question and this may not be as large a task as it might initially appear. For example Appendix G indicates that five systems deal with fertiliser recommendations at land management level (WELL_N [16], SUNDIAL [20], N-Gauge [40], PLANET [39] and N-Fert [41]) and of these only three are in active use as DSS.

On the policy side, agreement on data sources, calendars and scenarios would reduce the degree of difference between answers but would not eliminate it.

4 Recommendations

The key recommendation from this work is the development and agreement of a vision of what is required from nutrient models, and the form of its delivery, from policy development and policy implementation perspectives. A vision will provide the direction and focus for any future integration of models. It will also allow the clear specification of requirements against which the existing model set and any future developments can be checked.

As a result of the data collection exercise and discussions with industry representatives the following specific recommendations are made:

4.1 Recommendations for further scientific research.

In order to provide missing information to allow scientific institutes to answer the questions identified by Defra units the following areas should be investigated.

• The links between nutrient loss from land and impacts on fresh water and coastal ecology.

• Ecological recovery following reductions in nutrient pollution (long term study).

• The processes that occur at the boundaries between well-modelled areas i.e. between soil/groundwater, soil/atmosphere, water/atmosphere and between surface waters and coastal waters.

• The integration of erosion and soil movement models with catchment area nutrient models

• The feasibility of upscaling models from field to catchment level

• The impact of organic farm practices on nutrient loss

• The potential for incorporating urban models e.g. on water run off from hard surfaces with farm level nutrient loss models.

• The establishment of consensus on baselines and targets for water quality

• The combined impact of farm practice and nutrient inputs on nutrient loss at farm level

• The updating of models for delivery to land manager to include changed farm practices and new crops and varieties

• The linking of socio-economic factors into nutrient models at the catchment and national level.

4.2 Recommendations for data gathering, data sharing

To improve the availability of data necessary for the long-term provision of models and model based decision support tool the following should be provided by Defra, or through Defra programmes.

• A single source of quality data about farm practice that contains individual farm data and includes those variables of specific relevance to nutrient modelling.

• A single low cost source of quality, model appropriate weather data (climate and daily) for land management model delivery systems and a tool set which allows models to extract data at an appropriate timestep.

• A mechanism for linking forecast weather data from commonly used commercial sources into land management model delivery systems.

• A system for collating information on the impact of nutrient reduction activities and ecological recovery over time

4.3 Recommendations on model integration and delivery

To facilitate a gradual move towards the integration of models and model delivery systems the following additional actions are recommended:

• To continue to support technical delivery systems for both policy and land management as they stand but work towards a single vision for delivery which includes:

- The establishment of a mechanism for obtaining consensus, nationally (and possibly on a European scale) on standards for model development at all levels. These standards should cover all aspects of model development that would allow models of different types to work together in an integrated system e.g. parameter naming conventions, units of measurement, development of scenarios etc.

- The establishment of a mechanism for obtaining consensus on standards for model and decision support system delivery for land managers. Consensus needs to be obtained on appropriate recommendations (e.g. for nutrient inputs), appropriate display conventions (e.g. presentation of uncertainty), and other delivery specific areas common to a wide range of models and sectors.

• To start the process of the standardisation and integration of models in a single well defined area (e.g. N recommendations for land managers) but ensure that this is agreed as a starting point by the wider community.

• To support the education and support of land managers in the effective use of models and model based decision support systems.

• To continue discussions with crop management software developers to identify mutually acceptable approaches to data sharing.

• To initiate wider and more focused dialogue between model developers and policy users on policy requirements and model design.

4.4 Recommendations on use of non-agricultural and non-UK models

• Initiate cross-disciplinary workshops to define approaches to answering catchment scale questions. While modellers within a given area appear to be highly cognisant of the existence and value of models from other countries, it would be useful to bring modellers who normally operate within distinct disciplines (e.g. forestry, upland management, fresh water ecology, coastal water ecology, farm management, etc.) together to discuss integrated approaches to answering Defra questions.

• Initiate discussions with the team managing UKWIR/EA Project CL/06/C: Effects of Climate Change on River Water Quality; Phase 4 – Water Quality Modelling to ensure a consistency of approach to model integration and delivery.

4.5 Further work on database and analysis of models

Two directions for further work on the model analysis are suggested. The first would seek to clarify the picture, or vision, of what a comprehensive nutrient management tool might look like (what areas of the nutrient cycles it would model, at what level of granularity, at what time-step etc) and begin the task of specifying the requirements for policy and land management users. The second would refine and expand the database tool to make it possible to check for matches between the models it contains and the developing vision and requirements. In particular, the model outputs should be carefully categorised to facilitate searching. Accessibility of the information within the database could be greatly improved by making it accessible via the web. Web-enabling would also allow, if desirable, model developers to update the records for their own models.

4.6 Summary

This project has fully met most of its objectives but has had only partial success in some. It has identified the key UK and international models and model-based DSS and collected and stored information on them within a searchable Access database. It has identified areas where additional scientific research is required to develop and expand models to provide answers to some of the questions raised by Defra and it has obtained agreement on these areas from a range of UK scientific research organisation representatives. What it has not been able to provide is clarity on which parts of the nutrient management picture are fully covered by existing models and where there are gaps (although the spreadsheet in Appendix G does provide an initial outline of current coverage). It has also not been able to show where existing UK models may usefully be combined to meet specific requirements. The author believes that it will not be possible to do either of these things until there is a clear vision of the shape of the desired nutrient management system (or systems) or what the requirements for information of policy or land management users are. Integration of current systems is likely to prove difficult and costly for the reasons outlined in the sections above, establishment of a vision will allow integration to be accomplished in the long term and permit more focused targetting of resources in the short-term.

| |References to published material |

9. 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.

|[1] Policy Commission on the Future of Farming Farming and Food (2002) Farming and Food: a sustainable future. Crown Copyright publication. |

| |

|[2] |

|[3] |

|[4] Elliott M. & de Jonge V.N.(2002) The management of nutrients and potential eutrophication in estuaries and other restricted water bodies |

|Hydrobiologia 475/476: 513–524 |

|[5] |

|[6] |

|[7] Arinze, B. (1992). A user enquiry model for DSS requirements analysis: a framework and case study. International Journal of Man-Machine |

|Studies, 37, 241-264. |

|[8] Parker, C.G. (2001a) An approach to requirements analysis for decision support systems. International Journal of Human-Computer Studies, |

|55, 423-434. |

|[9] |

|[10] Parker, C.G. (2004) Decision Support Tools: Barriers to Uptake and Use. Aspects of Applied Biology, 72. Advances in applied biology: |

|providing opportunities for consumers and producers in the 21st Century. AAB |

|[11] |

|[12] |

-----------------------

[pic]

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download