Transcript Geo-referencing FADN farms

Geo-referencing FADN Farms
Date: October 22/23rd 2015
Venue: EAAE Seminar, Edinburgh
1. Brian Moran; 2. Stuart Green; 3. Prof. Cathal
O'Donoghue (Teagasc);
Farm Accountancy Data Network (FADN)






Tool for evaluating the income of agricultural holdings and the impacts of the
Common Agricultural Policy.
Started in 1965 - It consists of an annual survey carried out by each Member
State. Legal Requirement.
The survey does not cover all the agricultural holdings in the Union but only
those which due to their size could be considered commercial.
The aim is to provide representative data along three dimensions: region,
economic size and type of farming.
The annual sample covers approximately 80.000 holdings. They represent a
population of about 5.000.000 farms in the EU, which covers approximately 90%
of the total utilised agricultural area (UAA) and account for about 90% of the
total agricultural production.
The FADN is the only source of microeconomic data that is harmonised,
i.e. the bookkeeping principles are the same in all countries.
Irish FADN – Who’s Involved?

DAFM – Department of Agriculture,
DAFM
Food and the Marine
 Oversight

Farming Population
CSO - Central Statistics Office
 Population Data

Teagasc – Agriculture and Food
Development Authority
 Liaison Agency
Teagasc
FADN
in
Ireland
CSO
Teagasc – Liaison Agency





Then Ireland entered the EEC in 1973 – obligation to deliver data to FADN
Whilst DAFM had responsibility they requested that AFT would collect analyse
and deliver data to FADN
Unit established in AFT in 1972
Has run consecutively for over 40 years as the Teagasc National Farm Survey
Main Objectives
 Provide Statutory Irish Data to EU Commission (FADN)
 Determine Output, Costs, Incomes By Farm System(6), Size(7) & Regions
 Provide Database of Micro Data on Irish Agriculture for Research, Policy
analysis & Stakeholders********
 Measure Variation in Technical and Financial performance for Farm
Management/Advisory Purposes

Currently operated in the Agricultural Economics and Farm Surveys Department
Points to Note

The farmers who participate are chosen at random
 About 10% of the sample changes each year

Farmer participation is voluntary

The data collectors are employees of Teagasc

Farms are visited 2-3 times per year

An accounts book is completed – customised MS Excel Program is used

Submitted for validation and processing by internal staff

Farm Management Reports are generated

Populations from CSO

Produce the NFS Final Report

Finalise FADN data
Data Collected
1. On average there's about 1000 variables per farm.
2. Economic and financial data, sales and purchases, production costs, assets,
liabilities, production quotas and subsidies
3. Physical and structural data, crop areas, livestock numbers, labour and recently
location.
4. All at the farm level – Cost allocation is not done on an enterprise level
5. Data collection is the responsibility of a liaison agency in each Member State
FADN – Current State of Play

Due to its unique position there are often request to collect new pieces of data

FADN unit have recently completed a process of reviewing and revising the farm
return, in use for the first time for 2014 data which will be submitted in
2015/2016.

Very tedious and laborious process to undertake changing the farm return, took
2-3 years to complete.

The end version is a very much ‘watered down’ from the original proposals.

Different data collection methods in each member states.
New Developments & Expansion of the FADN Dataset
accepta
very
difficult
ble
difficult
Geo-reference
Irrigation system
Agricultural training of the manager
Share of OGA labour in total work
Asset Valuations at historical value.
Recording quantities of N, P and K in fertilisers.
Crops: Information on the area of GMO- and energy crops
Livestock - Recording the destination of sales
Average weight of animals slaughtered - External File
OGA directly related to the holding
Distinguish sources of financing (EU, co-financed, national)
Distinguish between basic units (heads, ha, ton/farm/other)
XML becomes the only accepted format for data transmission:











FADN 2014-New Farm Return




Includes a geo – coordinate for each holding
The agricultural holding is located where main part or all agricultural production
takes place. It can be an agricultural building (i.e. largest administrative
building/construction used to house livestock or other buildings or constructions
used for agricultural production e.g. a greenhouse) or another identified part of
the holding such as the most important parcel of the holding.
In case there is no agricultural building to which a location of the holding could
be attributed, the most important parcel will be chosen as the reference point.
The same is valid for the agricultural holdings having the land area in different
regions.
The holder’s residence can be considered as the reference place only when it
lies within 5 kilometres (in a straight line) of the place where the main part or all
the holding’s agricultural production takes place.
Confidentiality and rounding possibilities



The usual FADN rules as regards confidentiality will apply. In the Commission,
farm location data will only be used for grouping farm data according to
geographic criteria other than those in the farm return.
(i) precise coordinates are not required. It will be necessary to provide the
location only to the nearest 5 minutes
(ii) There are also country specific rules, for UK, Sweden & Finland.
How the data is to be collected
 Reference to administrative records such as cadastre databases (LPIS):
 Conversion of the address to the latitude and longitude coordinates by
appropriate software;
 Using maps to derive the geo-coordinates;
 Using a GPS device providing exact coordinates of the location of the holding
Our Experience in Spatially enabling the National Farm Survey

The aim was to try and associate a geographic X-Y point (in Irish National Grid
coordinates) for each participant in the NFS for 2007 in order to attribute new
environmental, geographical or meteorological data to each farm.

The only geographic information collected is the address of the correspondent

The history of Irish toponymy is a complicated history of local place-names
surviving against imposition of standards by different authorities.

The official allocation or recognition of place names (vested in An Coimisiún
Logainmneacha) is based upon the historical development of administrative
units
Example Addresses, all “correct”

Teagasc Research Centre, Malahide Rd, Kinsealy, Co. Dublin
Examples of common alternate address forms:

Teagasc, Kinsealy, Malahide, Co. Dublin

Teagasc, Kinsaley, Malahide, Co. Dublin

Teagasc, Malahide, Co. Dublin

Teagasc, Mullach Ide, Baile Atha Cliath
All of these addresses are “official” and correct. On top of these official variations
there are accidental misspellings, colloquial alternative spellings and reversals:

Tegasc, Kinsealy, Malahaide, Co. Dublin

Teagasc, Malahide, Kinsealy, Co Dublin
GEO-Directory


This product is created based on the OSI cadastral database of building
locations against the An Post database of delivery addresses
The Geodirectory is supplied as database with tables and fields allocating
every address to a building and every building to a geographic 6-figure
position in Irish National Grid (ING) coordinates
Methodology
There are three parts to the problem of spatially enabling the national
farm survey for allocation of environmental attributes:
1. Matching addresses in the NFS to possible addresses in the GDD
2. Allocating a geographic point that represents the matched GDD
addresses that deals with the one-to-many matching possibilities
and retains an element of confidentiality in the data.
3. Ascribing a representative sample of the environmental attribute to
the point
The first was achieved within MS ACCESS and the second two using
ArcGIS.
Address in Ireland does not uniquely identify a rural address
 As the majority of NFS addresses match to multiple building points
we have to decide how turn this into one point for the correspondent.

Because of the inherent precision in the environmental datasets
wherein the digital soil map has a 75m limit to precision, the
Teagasc Indicative habitat map has a minimum mapping unit of 1ha
(and the climate models have a 1km cell size – there is no need for
precision > 100m.
 Another point to bear in mind is that the GDD point is allocated to
the farmhouse not the farm, this is relevant for large farms or
fragmented farms.
An example of the
geo-coding process.
Each
cluster
of
points of the same
colour have been
matched to a single
NFS address. One
point of a given
colour means that
the NFS matched to
a single address in
the GDD.
The same
matched clusters
but with the
geographic centre
marked as a black
star and 1 standard
deviation of the
cluster from that
centre marked
with a grey circle.
Rural Economy Research
Centre
17
Project Outcome

This initial project successfully assigned a geographic coordinate label to farms in the
2007 national farm survey database. The process was then extended to cover all farms
that participated in NFS since 1995.

With a coordinate, spatially referenced environmental, economic and geographic data can
now be assigned to each farm thus expanding the usefulness of the NFS dataset.

The backdating of the process means its is now possible to retrospectively include
historical data into analysis using the NFS
Matching Method & Confidentiality

Researcher matches their spatial data to the Geo_directory building IDs

They send that data to the NFS team

We hold a reference file that contains the FADN Farm details and the building ID

We match the file and then remove the building Id and return the farm data and
the spatial variables.

Therefore we don’t release the location of the holding
New Developments & Expansion of the Irish NFS Dataset

Inclusion of new environmental to enable us calculate the carbon footprint of
milk and beef – LCA analysis.
 Turn out dates
 Slurry Storage and Spreading data
 Monthly data – Milk, Animals Numbers, Feed Usage

Data Collection – Extend the scope and reach of the survey by linking up with
external databases

Other new variables
 Cereals – Moisture content & Basis of sale – eg if crop forward sold
 Dairying – Forward sale
Opportunities
 To enrich the data without burdening the farmer
 Farm level Data collection is very time consuming
 To Date - Link has enabled use of
 weather data
 soils information
 distance data
 Habitat maps
 Potential now through FADN to extend this to all EU member states
 Link with LIPIS
Challenges

Maintaining confidentiality

FADN sample is not picked to be spatially representative

Multiple Parcels
What we can do and can’t do


Can
Introduce spatially referenced data
 Soils
 Weather
 Altitude
 Local Market Information
 Local Crop Growth Curves
 Land Parcel
 Water Quality



Can’t
However spatial resolution of data is not necessarily at farm or parcel scale
Regression Grass growth = f(Agronomic Variables)
 Farm R2 – 28%
 District R2 – 66%
Where are we going with All this?







Spatial Context within Analysis
Productivity Analysis – better handle on farm efficiency conditioning on
Spatial Policy – impact of move to agronomic based Areas of Natural
Constraints
Innovation pathways – spatial correlation
Like environmental risks with Agri-environmental decisions
Transport Costs
Farm Level Bio-Economic Models

Localised Advice
Localised Advice



Events happen locally
 Fodder Crisis
Agronomic Conditions
Localised Financial Management
 Low take-up of decision support tools and financial planning
 Build upon real-time administrative, meteorological and satellite data
 Link Farm Spatially related production and cost models as a function of real
time spatially based data
 To simulate top middle and bottom third financial benchmarks
 Better to have predicted financial data than no data
 Developing interpretative tools for farmers to understand what they are
being given
 Training to assist understanding