Operational use of Earth Observation data in agriculture

1
Operational use of Earth Observation data in
agriculture
Gábor Csornai REMOTE SENSING CENTRE INSTITUTE
OF GEODESY, CARTOGRAPHY AND REMOTE SENSING
(FÖMI) BUDAPEST
2
Content

• The Hungarian Agricultural Remote Sensing Program
  (1980 -)
• The Operational Crop Monitoring and Production
  Forecast Program (CROPMON 1997-2003?)
• Other RS applications on the CROPMON basis
• EU memberstate requirements LPIS and RS control
• Special operational endeavour ragweed monitoring
  and control in Hungary

3
Hungarian Agricultural Remote Sensing Program
4
The Operational Crop Monitoring and Production
Forecast Program (CROPMON 1997-2003 ? )
5
Early steps in the Hungarian Agricultural Remote
Sensing Program (1980-)

• FÖMI RSC investment 300 man-year RD
• Internationally presented results (EU, USA)
• The evolution of the research results

• 1983 1-3 farms Hajdúság ( Hajdú-Bihar
  county)
• 1984 Hajdú-Bihar county
• 1986-1987 Multitemporal evaluation (crop
  development assessment)
• 1990 Crop maps for 3 counties area estimation
• 1993-1996 NCTD-MARD-FÖMI project RD
  preparation to operational monitoring more
  international projects
• 1997-2003 The Operational Crop Monitoring and
  Production Forecast Program (CROPMON)

6
Major areas of RS information support to the
Ministry of Agriculture and Rural Development
(MoARD) by FÖMI
MoARD Rural development
MoARD Strategic decisions and requirements of
EU-membership
MoARD Daily operative tasks
National subsidies
Market intervention
Special cases
EU-conform control of subsidies
Administrative management
Sustainable growth
Agro-environmental issues
Included in subsidies
EU-funded subsidies (area based, structural funds)

• Reliable crop production forecasts
• Area and yield of major crops(CROPMON)

• Cadastre, national base dataset provision
• Land knowledge expertise
• National map archive, system of digital maps
• Up-to-date institutional system

• CORINE
• EU-conform information system on land use and
  land cover
• Support for strategic planning
• Evaluation of digital map databases

• Flood and waterlog monitoring and forecasting
  system
• Drought monitoring
• Ragweed monitoring
• Gypsy moth damage assessment

• Support of AEP LPIS, blockmaps printing,
  controls with remote sensing
• LPIS thematic layers for rural development LFA,
  ESA

• Integrated Administrative and Control System
  (IACS, LPIS)
• Control of subsidies with Remote Sensing (CwRS)
• National vineyard register in GIS (VINGIS)
• National orchard register support with remote
  sensing

7
The objectives of CROPMON(1997-2003...)
To provide reliable, accurate, timely information
for the Ministry of Agriculture and Regional
Development (MARD)

• 8 major crops in Hungary winter wheat, winter
  barley, spring barley, maize, silage maize, sugar
  beet, sunflower, alfalfa
• area assessment, the estimation of final yield
  before the harvest (optional crop development
  assessment, draught alarm)
• Reported Data for 8 crops
• Crop Area
• for the counties
• for Hungary (plus crop map)
• Crop Yield Forecast
• for the counties (plus detailed yield map)
• for Hungary
• Reports (from 15 June to 18 Sept final 1 Oct)
• 4-(6) reports/year by a predefined calendar
• Drought damage forecast reports

8
Earth Observation field data in crop
production data outData Flow in CROPMON
Precalibration, historical data
Crop maps and area figures
Reference data
Production Reports
Development assessment
CROPMON INFORMATION EXTRACTION SYSTEM
Yield forecast
High resolution satellite data
Drought Alert Reports
Low resolution satellite data
9
Reference data processing

• Sources
• Cultivation maps of some farms
• Field data collection

10
The central data for crop monitoring the NOAA
AVHRR series

• Features
• more satellites (NOAA-12, 14, 15, 16, 17, 18.)
• several coverages/day (10-20/24 hours)
• not too large pixel (120 ha)
• multispectral value
• stabile data flow

RAW
PREPROCESSED
21 July, 1998 (124 RGB)
21 July, 1998 (211 RGB)
11
Main assets in CROPMON operation
1. Data security NOAA AVHRR receiver (FÖMI RSC)
Landsat, IRS, SPOT (ERS, RADARSAT), VEGETATION
2. Established methodology (FÖMI RSC
innovative methods) 3. Pre-calibration
measurements
12
The FÖMI RSC crop yield forecast model
good

• Basics
• combination of spatial with spectral/temporal
  information (high res. AVHRR)
• unmixing NOAA AVHRR images using crop maps gt
  crop specific temporal profiles
• Features
• generic works for more crops
• year independent (validated for a range of
  seasons 1991-99)
• area independent
• reliable, accurate, timely

bad
bad
Corresponding subset of a NOAA AVHRR colour
composite (211 RGB)
Part of a crop map with 1.1 km grid
overlay, corresponding to the NOAA AVHRR pixel
size
13
CROPMON additional applications on this basis
7 years of operational RS Crop Monitoring and
Production Forecast (CROPMON- Hu, 1997-2003)
Land Parcel Identification System (LPIS)
Control with Remote Sensing (national 2000-2003)
Control with Remote Sensing (EU 2004-)
Flood and waterlog monitoring
Drought monitoring
VINGIS
ESA-FÖMI Prodex-ENVISAT KF project and PECS
agreement
Monitoring of damages caused by gypsy moth
Ragweed monitoring
14
(No Transcript)
15
Other RS applications on the CROPMON basis
16
Flood and waterlog monitoring by remote sensing
in Hungary 1999-2001
The waterlog causes major loss to agriculture
yearly. FÖMI RSC developed the remote sensing
based model to map and monitor waterlog. Sample
of a typical waterlog map from 1999
Flood monitoring and prevention is important in
Hungary as 95 of its surface water flows in from
abroad. Shots from the monitoring of year 2001
flood over the Hungarian-Ukrainian border on
river Tisza.
17
August 2000, 2001 NOAA AVHRR
Drought monitoring and loss assessment, as a
component of a CROPMON
August 2000, 2001 SPOT VEGETATION
18
Examples from the results of ESA-FÖMI
Prodex-ENVISAT RD project
Waterlog map obtained from IRS MOS (26 February
1998), IRS WiFS (1 April 2001) and SPOT
VEGETATION data (2 April 2001). Degrees of wet
soils and wet vegetation could be identified and
differentiated more effectively using the lower
spatial but higher spectral resolution SPOT
VEGETATION images
The elements of real-time flood monitoring
carried in March 2001, partly in the framework of
CROPMON (high res.) and Prodex (low and medium
res. NOAA AVHRR, IRS WiFS)
Difference map of MERIS MGVI (2003) and mean of
NDVI (1998-2000) derived from IRS-1D WiFS images
for May-June 2003. The regions hit by drought in
2003 can be clearly identified. red seriously
affected by drought, orange moderately affected
by drought, yellow weakly affected by drought,
light yellow not significantly affected by
drought, green not affected by drought
19
Remote sensing supportto the vineyard and
orchard census for the Central Statistical Office
Basic maps for the census
For supporting the implementation of the
regulation CXLIII. 2000. about the census of the
vineyard and orchard in Hungary.
Sample from the supported maps showing vineyard
and orchard areas derived from HR time series
20
Special case study in Hungary monitoring of
damages caused by gypsy moth in 2004
Gypsy moth has damaged 70 000 hectares of forest
Satellite image time serries
Map of damaged forests
26 May
18 June
4 July
20 July
21
Special case study in Hungary monitoring of
damages caused by gypsy moth in 2004
22
Requirements in an EU memberstateI. LPIS and
its management
23
HR data in the LPIS-Hu creation
Topographic map
2000 / (2003) year ortho-photo


2000-2003 HR image time series (Landsat, IRS,
SPOT) for confirmation of construction lineaments
plus the land use stability
Integration of aerial and satellite data
24
II. Remote sensing control of the agricultural
subsidies
25
RS control program for the national subsidies
2000-2003

• 1997-98
• Synthesis of EU regulations, national adaptation
• technical and organisational preparation of
  system design proposals

• 2000
• Operational remote sensing control for 7 sample

• 2001
• Operational remote sensing control for
• 4 sample

• 2002
• CwRS of nat. subs. (sample from all counties)
• 6 sample

• 1999
• analysis of EU practices
• NPAA VIII programme
• system design on CROPMON bases
• experimental control
• (3 counties)

• 2003
• National program
• DG JRC-FÖMI program
• 6 sample

2004 8662 claims (4,2 of total 208 000 claims)
have been controlled by remote sensing on 550 000
hectares and 39 500 parcels.

• Technical features
• Strong application of HR satellite data time
  series
• VHR use pilot (2002) and operational (2003- )
• Developed CAPI
• The application of CTS diagnostic codes and
  tolerances at parcel level
• Advanced error-documentation (maps)
• GIS handling
• Follow on field inspection by the independent
  MARD system

26
Control with Remote Sensing (CwRS) of EU
area-based subsidies scheme (2004- on)

• Satellite images

• Alphanumerical claim data in database

• Parcel drawing in GIS

• Block maps with farmer drawings

High resolution (10-25m ) time series
CwRS Central database

• LPIS-Hu (MePAR)

6. CAPI Computer Aided Photo Interpretation

• Decision
• at claim / block level

• Control documents

Very high resolution (0,5-1m)

• Control results database

7. GAEC control with DEM and satellite img.
Agricultural and Rural Development Agency (ARDA)
Institute of Geodesy, Cartography and Remote
Sensing (FÖMI)
DG AGRI, DG JRC
27
Special operation ragweed monitoring and control
in Hungary
Ambrosia
28
Ragweed impact on the health of the citizen

• 1. Story - first advent around 1920
• - gradual spread
• - approx 8-10 of the Hungarian population is
  allergic to ragweed pollen
• Some 500-700 000 hectare is covered by ragweed.
• 80 of that area can be pinpointed by HR remote
  sensing (arable land)
• Some 28 -40 M / year for medication and medical
  visits, tests
• Public movement to push it back resulted in legal
  actions amendment (Act. XXXVIII, 2005) of the
  Law for plant protection (Act. XXXV, 2000).
• Breakthrough if detected and documented
  (mapped), it can be removed, after the
  announcement of the declaration of the local
  government or the responsible service!

29
Program for control ragweed from 2005 on

• Ragweed contamination assessment for 2004
• Through media, web, maps on RS assessment for
  2004 were available for the people (100k spots,
  200k hectare)
• Webpage (MARD, FÖMI) from 16 June on 2004
  status,
• from 15 July 2005 status
• Realtime measures in 2005 (from June on)
• A) Surveillance of ragweed spots (remote sensing
  FÖMI RSC, public, services)
• B) In situ validation, documentation (agri
  experts (150) of the Land Offices Network LON,
  MARD)
• C) Legal announcement by the Plant and Soil
  Protection Authority (PSPA, MARD)
• D) Cut off ragweed (PSPA)
• E) Recovery of justified costs from land users
  (PSPA, TAX Office)

30
Data and documents flow among the institutions
ARDA Agricultural and Rural Development Agency
FÖMI
Public info
FÖMIwebpage
Ragweed maps
Decision by PSPA
In situ records by LON
In situ records by LON
PSPA Plant and Soil Protection Authority
CENTRAL RAGWEED SERVER
Ragweed maps
Decision by PSPA
LON Land Offices district 116, county 20
In situ records by LON
Inevitable for efficient implementation of
the legal process
31
E-info on ragweed cover (required by law)
http//www.fomi.hu/ Projektjeink/ Parlagfu
felmérés
32
Ragweed monitoring a remote sensing challenge
Ragweed life cycle
33
The 2005 campaign utilized some 80 HR IRS
AWiFS images
IRS P6 AWiFS
automated image processing
Landsat TM5
34
Report category 1 non cultivated arable
spots (ragweed or other allergic weeds)
E-transmission of the maps to districts
5 693 spots 13 505 hectar
35
Report category 2 ragweed on cereals stubble
13 556 spots 45 176 hectar
36
In situ recording and documentation
Spot subdivision by cadastre / landuser
Palmtop GIS software GPS
37
The spots are measured and recorded in the
palmtop by the LON officers
GPS measurement
Spot derived by remote sensing (HR)
In situ E-record
38
Summary and conclusionfor Ragweed Control

• Remote sensing 20 000 spots, 60 000 hectare
• In situ records 10 000 spots, 19 000 hectare
• Fine 8900 cases, for 13 000 hectare (200
  /hectare)
• Conclusion
• A complex yet efficient control system can be
  driven by remote sensing
• Ragweed recognition is far harder than that of
  crops
• Regulations forced to integrate 4 high tech
  technologies
• The landusers cooperated well in the control
  (eliminated much more in 2005)
• Next year 2,5-4 times bigger area can be cleaned
  from ragweed at the same resources investment

beyond 1,3 (0,8) hectare
39
The ragweed recognition is as easy as listening
to grasshoppers that are behind a couple of
working airpressure hammers
40
Conclusion - The Hungarian Agricultural
Remote Sensing Program stands on a 25 years
background.- The RS production forecast is the
starting point for more additional
applications.- FÖMI Remote Sensing Centre still
loves challenges, though operational problems
(e.g. ragweed control)