Real-time predictions of pesticide run-off risk which: multi-scale visualisations of water quality risks and costs

Lead Research Organisation: University of Leeds
Department Name: Sch of Geography

Abstract

This Research Translation Project develops a proof of concept to tests the value of real-time predictions of agro-chemical run-off risk at two scales of decision making: field scale for on farm decisions about agro-chemical applications risk and catchment scale for water company groundwater abstraction decisions.

Agro-chemicals (fertilisers, pesticides, herbicides, etc) are less effective if they are washed away soon after they are applied. They can also negatively affect ground water quality and the environment. The farmer may have to re-apply the agro-chemical and water companies may have treat groundwater to meet drinking water quality standards, and in some cases change water abstraction locations. For both farmers and water companies additional costs are incurred.

This project develops proofs of concept for 2 web-mapping tools to model the risk associated with agro-chemical applications: a catchment-scale tool to support water company decision making and a field-scale tool to support farmer decision making. Both tools combine live, real-time data from the Met Office on rainfall type and probability with landscape models of underlying soil, landform, drainage, land use etc. in order to model agro-chemical runoff risk. User-groups will feedback their experiences about the operational use and functionality of the tools to provide information for the modelling and programming teams to adjust the background engine and front-end functionality.

The project outputs will include the specification of for national decision tools, targeted at farmers and water companies, to quantify the risks associated with a full set of common agro-chemical applications designed be accessed using desktop PCs and smartphones.

Key Words: Agro-chemical run-off, water quality, environmental risk
Stakeholders: Defra, farmers, water companies, AHDB, SARIC members

Planned Impact

One of the challenges in translation research grants is to provide the link beyond proof of concept to full commercial implementation of the product under development. Our engagement with the newly formed Agrimetics provides this route.

The inclusion of Agrimetrics provides a route to ensure that the tools we develop potentially have a long-term commercial home. Collectively, the RTP consortium provides the key expertise and knowledge components for tool development, proof of concept and for effective knowledge exchange, supporting water companies to determine drinking water quality and abstractions risks, via this demonstrator project and for its subsequent development. However, the structure, organisation and remit of Agrimetrics provides pathways for a successful translation and rollout of the proof of concept tools developed by this RTP. Critically, the project links with Agrimetrics will ensure that the outcomes of this RTP are scalable, that they are realised into full tools and that they have a long-term legacy (it is envisioned that Agrimetrics will host the server for the full roll out of the generic tools).

Agrimetrics is a company funded by the UK governments Agitech strategy as the first of its Innovation Centres. Its mission is to capitalise on the proliferation of agricultural data and rapid advances in agri-food modelling capabilities to put the UK at the heart of a new agri-food revolution based on data and analytics.

By the time this grant is initiated Agrimetrics will have built the world's first commercially focussed linked data platform for the agrifood sector. Linked data is a comparatively new way of organising data which is based on the philosophy of the world wide web. It emphasises the role of meta-data so that data in disparate locations and data-sets can be interrogated without acknowledging the boundaries between them. This will create a powerful resource which can subsequently be used by businesses to deliver actionable information to practitioners.

The proof of concept that is developed in this project will provide two very specific data streams at the field and catchment scale that are relevant to farms and water companies respectively. It will also test whether providing this information is value to users through the use of a dedicated prototype app. It is not envisaged that this prototype app will evolve directly into a commercial product. For this to be of use it will need to be integrated within existing software tools and advisory mechanisms to provide information on pesticide use in a fully integrated way that includes all agronomic decision making.

Agrimetrics will facilitate this by exploiting its extensive network to take the prototype to businesses that have a close connection to the industry including agronomists, levy boards and software publishers as a value proposition. In doing so it will show how it is able to provide the necessary data streams that will support the final product.
 
Description We have developed were stripped down versions of a soil-water simulation model, parameterised by historical rainfall data, and for which relationships between crop type and soil type were pre-computed. Now they have been implemented, encoded linked to a server providing live rainfall data and wrapped up in web-based interfaces. There are four models, 2 for each study area in Wales (Teifi) and East Anglia (Wissey) and these can be found at the following websites:
https://saric.shinyapps.io/tei_field/
https://saric.shinyapps.io/wis_field/
https://saric.shinyapps.io/tei_catch/
https://saric.shinyapps.io/wis_catch/

The Field scale model
In brief it provides point specific information of current and predicted future (5 day) runoff risks, again for each 1km2 grid cell. It seeks to support on-farm decisions about agro-chemical applications and to provide forecasts of potential field-scale runoff risk. Although a soil water balance model could be used to antecedent soil water conditions and the CN method (USDA, 2004) to assess potential field runoff in real-time, data and computational requirements are an important limitation. In addition, soil water balance models require a known starting condition and are prone to cumulative errors, particularly during periods of low rainfall. From an operation point of view, using a soil water balance model to estimate antecedent soil water conditions also requires the user (farmer) to collect and process rainfall data even during periods when runoff risk forecasts are not required. To overcome this, a meta-modelling approach was used to estimate antecedent soil conditions from soil type, long-term average soil water content for the day of year, recent recorded rainfall and short-term forecast rainfall.

The landscape scale model
The aim of the landscape scale tool was to generate a spatially-distributed delivery risk surface to inform drinking water abstraction decisions. The output predicts the spatial pattern of pesticide loadings to receiving watercourses where water abstraction might take place in the study area. The parsimonious approach combines layers of intrinsic landscape scale factors, runoff and pollutant transfer, national historical daily rainfall data from the CEH Gridded Estimates of Areal Rainfall dataset (Keller et al., 2015), with live data of current and antecedent rainfall, as summarised in the figure below.
Exploitation Route there is huge potential for this framework to be commercialised - it can be used to model runoff for ANY agricultural application
Sectors Agriculture, Food and Drink,Environment
URL https://saric.shinyapps.io
 
Description Impact for this project is in the future. The tools developed by this proof of concept translation project are being user tested. Therefore the project has the potential to have impact in the future if the parsimonious modelling framework is rolled out to develop a generic runoff risk tool that is made available to end users.
First Year Of Impact 2018
Sector Agriculture, Food and Drink
 
Description Meeting with Anglian Water 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact SARIC meeting Anglian Water 20th March, 2017
Tim Hess, Andy Smith, Lex Comber, Andy Turner
Rob Holland
Liz Boswell: on the phone
Martin Bowes: CamEO partnership - Cam & Ely Ouse Catchment Partnership interested in movement of soil, erosion, river health
Tom Edmondson -catchment advisor for Northants, Bedfordshire, Leicestershire and West Cambridgeshire, including the Nene, keep soil in the field

Summary

Agreed on the Upper Wissey catchment: it has a good mix of land uses, a MH problem, also sympathetic and known farmers / managers, and Tim Hess / Cranfield have other data on irrigation, infiltration, NALD data that may be available

List of Actions

1. Data
Action: AW to provide basic data needed by the project:
- Shapefile for areas  
- Abstraction points (eg Stoke Ferry)
- Sample points and Gauging points
Action: AW to find out whether this project can get access to the sampling data
- Regulatory data at abstraction locations
- Other catchment data from other sample points
o esp the long run of metaldehyde data that was alluded to
NOTE: **Adie Collins would like to leave the meeting on the 3rd with all the shapefiles that AW are able to pass over to us!**

2. Contacts
Action: AW to broker contact with groups
- Farms: South Pickenham estate is an innovative large land owner, Necton farms also
- Frontier and Agrii in Swaffham
- Norfolk Rivers Trust - planned projects on restoration
- WRE - has Wissey as a project area - interested quality quantity
o Amount of abstraction vs potential for run-off
Action: AW to broker contacts with
- Ed Bramham-Jones - independent agri advisory for CamEO and NRT - for fine scale data
- Bob Evans for field scale runoff data

3. Other
Action: AW to confirm status and availability of new high resolution weather data from new Adcon weather stations - Ed Bramham-Jones can advise on this and Neil Obbard (Agrii) can advise on access to their extensive network
Action: LexC / UoLeeds to find out about LiDAR data for the catchment
Action: TimH / Cranfield team to meet with Ed and Bob and Hannah at Stoke Ferry
Action: LexC to check on Adrian Collins preference for meeting
Action: RobH to arrange telecom for meeting on 3rd April if needed
Year(s) Of Engagement Activity 2017
 
Description Meeting with Dwy Cymru / Welsh Water 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Industry/Business
Results and Impact A meeting with Dwy Cymru / Welsh Water on 3rd December 2018 to discuss the project findings:
1. The project findings (report)
2. The user feedback (report)
3. Exploring next steps:
- more formal validations
- extending to other applications
- future App
4. Actions supporting next steps
- available data (e.g. data on acid herbicides for the Teifi)*
- is that anything similar to the national enhanced CSF monitoring pesticide for any catchments (including Teifi)?
- promotion of any testing / uptake
Year(s) Of Engagement Activity 2018
 
Description Meeting with Welsh Water 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Meeting with Welsh Water, 5th Jan 2017
WW attendees
Tara Froggat - Catchment delivery manager
Shaun Lewis - GIS
Matthew Kearon - Spatial Risk
Jess - Risk Management
Project attendees
Aidie Collins, Yusheng Zhang - Rothamsted
Lex Comber, Leeds
Tim Hess - Cranfield
Andy Smith - Bangor

1. Introductions
WW Risk team and Delivery team: remit to look at source waters; quality rather than quantity; consideration of the impacts on quality; aim to mitigate risks before extraction

Lex: Items for discussion: What kinds of outputs are WW interested in? What information do they get about pesticide usage? What inputs do WW have that the project might use (Data, Farm networks)?

2. Discussion about Data
- River network data: own? CEH?
- WW use a topographical boundary for ctahments, but looking for open data solutions, eg OS open data has geometry but not attributes
- WW seeking to identify what they need to know
- The WCs are caught in a data trap as a non-statutory body

3. Case study site: agreed Llechryd River Teifi
- identified as high risk agriculture (?)
- WW have chemcatcher data and 'weed wiper trial' but grassland herbicide trial, high profile via farm networks
- BUT still don't know how many farms

4. WW Wish list for project outputs
- WW want to look at risk in the field and want to understand where the risk is coming from
- Something WW could use out in the field to be able to do ground truth data
- Ground Truth to get confirmation of risk: does it marry up with WW perception of risk as catchment coordinators
- WW interested in Risk definition: At the moment WW are reactive but want to be a bit more proactive - early warnings, predictive systems, what could change, what could we see coming through the system
- Discussion about exceedance vs presence:
- WW: if pesticide then presence - if pesticide is present then viewed as risk. Big rivers have removal. But small ones do not. So do not have pesticide removal treatments on all of the abstraction location / plants (as they do in England which in general are on bigger rivers)
- WW does no biological monitoring at all.
- Agreed that project could describe relative risk (colour coding)
o Would allow Hotspots to be validated in the field. BUT also needs to know why / origins of risk
o Also interested in persistence of risk (so explanations of the risk assessment could be used to make these assessments)
- Because of data access issues for WCs project could include a scoping out the future roll out in terms of applications, areas AND data


5. More detailed discussion about Data
- Chemcatcher
- Point of abstraction data
- OS 5m height data
- Address Place premium - point data with assets
- UKwir - pesticide risk mapping - 3rd phase - did something similar field scale but not able to use the outputs because of the restriction
o Interventions in the field and how that relates to risk
o Failed in 3rd phase: too heavily focused on metaldehyde
o WW wanted LPIS data but couldn't have it so fell short
o No access to CLAD
Action: WC to provide a data wish
Action: WC to provide a list of data available to the project

6. Ground water vs surface water - do we deal with both?
- WW (Matt) described the Asulam / bracken example: the Asulam persisted in the GW for a long time, had to send tanker daily, downstream treatment works closed for 9 weeks. Buffers but what constitutes a water course? They followed the rules but boggy plateau - high water table - river easier to honour 10m buffer but not in this case
Year(s) Of Engagement Activity 2017
 
Description Towards real-time predictions of agro-chemical run-off / field losses risks and costs - Wales stakeholders workshop 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Professional Practitioners
Results and Impact This workshop was organised by collaborators of a Research Translation Project (RTP) funded through SARIC1 (Sustainable Agriculture Research and Innovation Club) by the Natural Environment Research Council (NERC). Its main purpose was to collect crucial information and feedback from relevant stakeholders involved in advice and decision-making about agro- chemicals applications and management and water quality in grassland areas in Wales about the pioneering proof-of-concept tools developed by the consortium.
The project is developing novel tools to deliver real-time, dynamic predictions about pesticide run-off / field losses risks, integrating real-time and historical weather data with catchment- scale and field-scale models, with view to support more informed agro-chemical management, both on farm and a catchment level. Initially, the project focuses on two case studies in contrasting agricultural areas: namely the application of acid herbicides on grasslands in Wales, and of metaldehyde on arable land in East Anglia.
In Wales, the project is supported by Dwr Cymru Welsh Water (DCWW), who have helped the consortium identify the Afon Teifi catchment as one where they experience pesticide run-off / field losses issues, especially from acid herbicides.
Twenty-two delegates from various relevant businesses and organisations to the agriculture, environment and water quality sectors responded to the organisers' call and were present on the day. The workshop was run for a half-day to minimise disruption in stakeholders' activities and was composed of two distinct parts: the first part was designed to introduce the context and provide information about this innovative project and the tools; the second part was more interactive and designed to capture critical information and feedback from participants on the project and proposed tools to ensure they are relevant and tailored to stakeholders' needs. For this reason, the workshop was attributed both BASIS and NRoSO CPD points.
Year(s) Of Engagement Activity 2017
 
Description Towards real-time predictions of field losses of agro-chemicals risks and costs - stakeholders workshop 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Professional Practitioners
Results and Impact Towards real-time predictions of field losses of agro-chemicals risks and costs - stakeholders workshop
- Summary -
The 'Towards real-time predictions of agro-chemical run-off risks and costs - stakeholders workshop' took place on Monday 27 March 2017 at the Sophi Taylor Conference Centre in Cambridge.
This workshop, organised by collaborators of a Research Translation Project (RTP) funded through SARIC1 (Sustainable Agriculture Research and Innovation Club) by the Natural Environment Research Council (NERC), introduced a pioneering project that aims to develop proof-of-concept tools that will deliver real-time predictions of pesticide field losses risks and support more informed and agro-chemical management to relevant stakeholders involved in the water quality and agriculture sectors.
The workshop had two distinct parts: the first session aimed to provide information about the context of the project and the project itself to attendees; the second session was designed to be more interactive and capture input from participants on the project and proposed tools.
After some refreshments and a brief introduction to the event, Patrick Goldsworthy, water quality consultant for The Voluntary Initiative, gave the audience more information about the issues around water quality and pesticides found in drinking water sources. Amongst substances having an impact relatively to surface waters and ground waters, metaldehyde presents the biggest issue to water companies as it cannot be treated and its removal is very costly. A map from the Environment Agency2 illustrates the extend of the problem in England drinking water protected area safeguard zones. Mr Goldsworthy then told delegates that rather to wait for the politicians to regulate, it was better to be proactive and raise awareness of these issues in the farming community and other related sectors through education, and improve practices through training. He highlighted that research had a role to play in advancing understanding and informing 'best practice'. Mr Goldsworthy explained the different ways pesticides reached drinking water. In recent years, the emphasis has been on runoff from farm yards more than field losses. Farmyard losses can often cause peaks (spills) as well as background levels as pesticides are washed off concrete; the risk applies to all pesticides all year round. A lot of preventive work and 'best practice' have been done to prevent contamination from yards. For some pesticides, a greater amount reach water sources by leaching through cracks, pores and fissures in the soil with rainwater to drains. Losses from the field are the principle source of pesticides reaching water especially in wet years; and drain flow is the dominant pathway, mainly affecting drinking water. If this contaminated flow was not taken by drains to surface water, it would risk reaching groundwater and emerging as springs elsewhere. Mr Goldsworthy highlighted several ways to mitigate the risks of pesticides leaving the field: soil management (more organics soils have been shown to act as 'filter'); buffer stipes; tramlines; crops; product; and weather forecast. Finally, Mr Goldsworthy brought attention of the audience to a campaign run by the Voluntary Initiative on behalf of DEFRA aiming at raising awareness and disseminating best practice for OSR herbicides3 use.
Next, Rob Holland presented a perspective from a water company: Anglian Water. In England and Wales, all water companies must supply water to defined drinking water standards; if any are exceeded, the Drinking Water Inspectorate (DWI) requires solutions to be implemented. Anglian Water oversees an area covering c. 9,000 km2 in East Anglia, where c. 50% of raw water comes from groundwater (i.e. 226 sources and >500 bore holes) and the other half from surface water (i.e. 9 reservoirs and 8 direct intakes). Mr Holland told delegates that pesticides represented the main issue for surface waters, while nitrates were the main issue in groundwaters. Mr Holland then presented the Wissey catchment area, chosen as one of the case study area for the SARIC project. Like most water companies, Anglian Water does weekly samplings at many point to assess the quality of water in the systems and if contaminants are detected above threshold levels permitted. Mr Holland showed a graph presenting the percentage of surface water samples taken that exceed the 0.1ug/l authorised limit and mentioned that most contaminants could be treated with the chemical treatment processes in place. He also mentioned that some pesticides that have been banned are still detected. Chemical treatment processes are ineffective at removing contamination from metaldehyde, which is detecting at relatively high level seasonally. Mr Holland told the audience about possible mitigation measures existing to reduce metaldehyde detection in water sources, the three main ones being product substitution; product rate; and product timing. Anglian Water has been running a scheme in sensitive areas where they pay farmers to use alternative products to metaldehyde close to water courses. They also organise training events for best practice when applying pellets. Mr Holland highlighted that the time delay in getting samples results back to detect contamination after a rainfall event was often an issue, as received after the peaks of contamination, and resulting in actions often being pointless. The predictive tools that will be developed by this project could help mitigate some of the timing. Mr Holland presented what Anglian Water is doing to encourage farmers to do to protect water courses. For instance, looking at the connectivity of field and yard to water courses; schemes to reduce drift; informing them about the new spray buffer zones; helping them to have a 'pollution emergency plan' so they know what to do and who to contact. Mr Holland gave a summary of his presentation and his final thoughts on water and agriculture sectors working together, ensuring provision of high quality and quantity water, whilst protecting actives to grow profitable crops.
Professor Alexis Comber from the University of Leeds gave the final talk of this first session, presenting this RTP project funded through SARIC to participants. Prof Comber, who is the lead Principal Investigator (PI) on the project briefly explained what SARIC4 (Sustainable Agriculture Research and Innovation Club) is and how the idea for this project was triggered at sand-pit event organised in March 2015, meeting and discussing similar issues with two water companies, Anglian Water and Dwr Cymru / Welsh Water [both industry members]. As a data scientist, he could see how developing and combining models at two different levels - catchment and field - could offer some solutions. A consortium was subsequently formed and the idea for a proof-of-concept project developed further, leading to funding being received from SARIC. Prof Comber introduced the team of academic experts part of the consortium: Prof. Collins (Rothamsted Research), expert in landscape scale agricultural and environmental models; Prof. Hess, Cranfield University (diffuse pollution and runoff); Prof. Smith (Bangor University), expert in plant-soil interactions; Prof. Tiffin (University of Reading), expert in agricultural economics; and Prof. Comber (University of Leeds), expert in data analytics and integration; the consortium also includes Agrimetrics and the Met Office, and works with end-user farmer groups and the water utilities - Anglian Water and Dwr Cymru / Welsh Water, both industry members of SARIC - in the chosen case study areas for this project. The aim of the project is to develop proof-of-concept tools that will test the value of real-time predictions of agro-chemical field losses risks and will support decisions at two scales: field scale for on farm decisions about agro-chemical applications risk, and catchment scale for water company groundwater abstraction decisions. In the first instance, the project will develop models of the real-time risk associated with two exemplar agro-chemical applications: metaldehyde in arable production in East Anglia; and acid herbicides in pasture-based livestock systems in North Wales Uplands. These models will integrate live Met Office data with spatial data on underlying soil; landform; drainage; land use; etc. to demonstrate the risks of field losses and provide more accurate information to support decisions made on farm and at abstraction points and better agro-chemicals management. Prof Comber concluded his presentation by telling delegates that the purpose of this workshop, beside introducing the project, was to give them the opportunity to give their feedback about it and to actively influence the future functionalities and design of these tools by providing valuable input during the next session. The consortium will also need volunteer farmers in the two specific case study areas (Wissey in East Anglia and Afon Teifi in North Wales) to test the tools once developed, to ensure they are useful and tailored to their needs when planning and managing agro-chemicals applications. Prof Comber made a call to people present to help with recruiting relevant farmers for this crucial next phase.
The first session concluded by some Q&A to all three speakers before a short refreshment break.
After the break, delegates reconvene for session 2 of the workshop and were split into two mix groups to take part in facilitated discussions and sand-pit exercises. Before the breakout discussions start, Dr David Haro, Cranfield University, gave the audience a short overview of the current practices and modelling approaches. He stressed that although they are a lot of models found in the scientific literature, most have not yet been developed into software or application that could be used by end-users such as farmers. He emphasised that this project aims to make the tools developed available to end users through web-based interface(s) and their input will be critical to ensuring the tools are most relevant.

Next, Dr Bovi introduced the breakout sessions and gave instructions to participants and facilitators to ensure everybody had their say and their input captured to inform the development of the tools by collaborators of this SARIC project. Delegates discussed three main topics: (1) Perception of field losses risks; (2) Applications planning and management decision-making process (current practice and information that would help); and (3) Delivery of information and design of tools. The information captured during these breakout sessions can be found in the appendix A attached to this report. A clear output from these discussions was that the tools developed should aim to provide recommendations reflecting the state of the crop (prophylactic / opportunistic or reactive application); the likeliness or state of pest infestation; and the time frame that the farmer was working in (e.g. 1 day; 1 week; 2 weeks; etc.). The recommendation delivered should advise when to apply products given these constraints and should offer some transparency about the factors taken into account in that recommendation.
The workshop concluded after lunch. Delegates came together one last time to hear Prof Comber summarise the workshop and hear about the next steps. The audience unanimously asked to be kept informed of progress of the project, and a final call to help recruiting relevant farmers in the two case study areas was made.
Year(s) Of Engagement Activity 2017