Use of remote sensing and

GNSS in precision agriculture

Prof Graciela Metternicht

Chair, ICA Commission on Mapping from Satellite Imagery

Curtin University of Technology
Perth, Western Australia
Email: g.metternicht@curtin.edu.au

UN-Zambia-ESA Regional Workshop on the Applications of

GNSS in Sub-Saharan Africa - June 2006

Traditional management practices assumptions:

One soil type over the entire field
One management practice:
fertilisers
herbicide
seed volume

UNOOSA ESA Regional Workshop, Zambia, June 2006

Precision farming
Information and technology-based agricultural
management system to improve crop production
efficiency by adjusting farming inputs to specific
conditions within each area of a field.
PA as a cyclic optimisation process
Data Collection

Analysis & Evaluation

Management / Planning

Precise on field operations

UNOOSA ESA Regional Workshop, Zambia, June 2006

PA Enabling Technologies (1)

Remote Sensing
Variable Rate Technology
Traditional Land surveys
Wireless Sensors Networks
GIS

UNOOSA ESA Regional Workshop, Zambia, June 2006

PA Enabling Technologies (2)

Variable Rate Technologies. Main areas:

Tillage
Fertilizer or lime applications
Planting
Spraying
Irrigation
Harvesting and yield mapping

Traditional Land Surveys

Scouting (GPS-based, guided by remote sensing
derived products)
Soil testing (grid based, stratified)
UNOOSA ESA Regional Workshop, Zambia, June 2006

Precision farming: sites are exploited according to

their capability
Fields are divided into
management zones
Techniques
continuos
sensors to map
crop yield

intensive soil
sampling
Time consuming!!

Remote
sensing
UNOOSA ESA Regional Workshop, Zambia, June 2006

Information required for crop precision management

Seasonally stable conditions
crop yield, soil variability

Seasonally variable conditions

soil moisture
crop phenology
crop nutrient deficiencies
crop disease
Pests (weeds or insects)

determining the cause of soil/crop variability

UNOOSA ESA Regional Workshop, Zambia, June 2006

Considerations for a spatial mission Agriculture

9 satellite revisit frequency ?

9 spatial resolution ?
9 spectral bands ?
to fulfil farmers information needs, at an
affordable price
UNOOSA ESA Regional Workshop, Zambia, June 2006

Requirements for image-based precision farming:

Spatial resolution: 5 meters or better, as
management units of 10x10 m (0.01 ha) are
generally adopted.
Spectral resolution: most applications require
multi-spectral data. Number of bands and band
resolution are variables determining the type of
features that can be discriminated.
Timeliness: rapid image turnaround (for fertilisers
or herbicide applications).
Frequency of coverage: crucial times of crop
development.

UNOOSA ESA Regional Workshop, Zambia, June 2006

Satellite Remote Sensing Technology

UNOOSA ESA Regional Workshop, Zambia, June 2006

10

Opportunities for Remote Sensing of Agriculture

Planning

Monitoring

Mapping and Monitoring Crop and Soil Variability;

Efficient way of mapping and monitoring the effects of
any condition that affects plant health, yield, or quality
of a crop.
The Imagery can be applied to:

Monitor within/between field variability;

Map soil variations;
Investigate crop management practices;
Detect and map weed and pest infestations;
Optimise crop inputs;
Pasture growth rate.
UNOOSA ESA Regional Workshop, Zambia, June 2006

11

Developing value-added
products
Planning
Very High Resolution
Remote Sensing

High Resolution
Remote Sensing

Combining terrain information, existing soil maps and

Remote Sensing data:
Map soil
variability
Improve existing
soil maps

UNOOSA ESA Regional Workshop, Zambia, June 2006

13

Improving soil
information for PA
470000

472500

Sandy loam A-horizons (7-13cm) over sandy clay loam to light

LMU 1 -medium clay

475000

sand A-horizons (<3cm) over sandy clay loam to medium

LMU 3 - Clayey
clay

LMUs - Muresk Farm

Subsoils contain stones (7-17%), Restricted depths (10-20cm)

Neutral, Non saline soils
Medium levels of organic carbon in the topsoil (1.0-1.7%)
Medium to good capacity to store nutrient cations (ECEC 7.0-24.2)

6492500

6492500

Subsoils contain stones (8-17%),Restricted depths (10-30cm)

Neutral, Non saline soils
Medium levels of organic carbon in the topsoil (1.3-1.7%)
Good capacity to store nutrient cations (ECEC 14.2-24.7)

LMU 4 - Sandy loam A-horizons (5-12cm) over sandy clay loam to light

LMU 7

medium clay
Subsoils contain stones (6-16%), Some restricted depths at 30cm
Some subsoils slightly sodic (pHca 6.0-7.1)
Some subsoils slightly saline (EC1:5 7-20mS/m)
Medium levels of organic carbon in the topsoil (0.9-1.5%)
Medium to good capacity to store nutrient cations (ECEC 12.5-20.0)
- Clayey sand A-horizons (<3cm) over clayey sand
Subsoils contain stones (13-23%)
Some restricted depths at 20cm
Neutral, Non saline soils
Medium levels of organic carbon in the topsoil (0.8-1.4%)
Poor capacity to store nutrient cations (ECEC < 6.7)

6490000

6490000

LMU 2 - Clayey sand A-horizons (8-14cm) over sandy loam

Subsoils contain stones (7-20%)
Restricted depths (15-30cm)
Neutral, Non saline soils
Medium levels of organic carbon in the topsoil (1.1-1.6%)
Poor capacity to store nutrient cations (ECEC < 5.9)

LMU 10 - Sand with A-horizons depths(6-11cm)

Restricted depths (<20cm), Topsoils contain stones (5-7%)

Some topsoils slightly acidic (pHca 4.4-5.1), Non saline soils
Low levels of organic carbon in the topsoil (0.7-1.1%) indicating
poor nutrient storage and unstable structure.
Poor capacity to store nutrient cations (ECEC < 4.0)

LMU 9 - Loamy sand A-horizons (10-15cm) over clayey sand

Subsoils contain stones (12-22%), Restricted depths (20-30cm)

Neutral, Non saline soils
Low levels of organic carbon in the topsoil (0.8-1.1%) indicating
poor nutrient storage and unstable structure.
Poor capacity to store nutrient cations (ECEC < 3.9)

LMU 5 - Loamy sand A-horizons (10-15cm) over clayey sand

Subsoils contain stones (9-17%), Some restricted depths at 30cm
Neutral, Non saline soils
Low levels of organic carbon in the topsoil (0.8-1.4%) indicating
poor nutrient storage and unstable structure.
Poor capacity to store nutrient cations (ECEC < 4.0)

220 440

880

1,320

6487500

6487500

LMU 8 - Sand A-horizons (1-5cm) over loamy sand

LMU 6 - Loamy sand A-horizons (10-20cm) over clayey sand

Subsoils contain stones (3-8%)

Neutral soils, Non saline soils
Medium levels of organic carbon in the topsoil (1.0-1.4%)
Poor capacity to store nutrient cations (ECEC < 3.8)

1,760
A-horizons (10-15cm) over sand
LMU 11- Sand
Subsoils contain stones (1-7%)

Meters

Some subsoils slightly acidic (pHca 4.7-5.3)

Non saline
Low levels of organic carbon in the topsoil (0.6-1.0%) indicating
poor nutrient storage and unstable structure.
Poor capacity to store nutrient cations (ECEC < 2.3)

AGD 84 UTM ZONE 50S

470000

472500

Subsoils contain stones (6-10%)

Neutral, Non saline soils
Low levels of organic carbon in the topsoil (0.7-1.1%) indicating
poor nutrient storage and unstable structure.
Poor capacity to store nutrient cations (ECEC < 3.1)

475000

Assisting to determine the cause of variability in

crop production:
Problem areas on the ground can be identified and
located on the enhanced imagery, or multi-temporal
analysis;
Sites are visited (portable GPS) to identify the causes
of site variability => soil testing may be required.
Information can be integrated within a GIS, for
further temporal and spatial analysis.

X= 545
y= 6777
UNOOSA ESA Regional Workshop, Zambia, June 2006

15

Detect and map weed infestations

Mapping excessive weed pressure on paddocks. Our
investigations found weak correlations between DMSI
data and weed density in early season canola seedlings.
Good results were achieved in detecting weed infested
areas in fallow pasture paddocks;
Weed detection is a function of weeds amount and
shape.

(c
)

Remote Sensing supporting VRT: fertilizers

Knowing the causes and spatial extent of variability,
the images can support variable rate applications, to
apply fertilizers at higher rates on lower producing
areas;

UNOOSA ESA Regional Workshop, Zambia, June 2006

17

Looking at crop responses in paddock strip trials

UNOOSA ESA Regional Workshop, Zambia, June 2006

18

Developing value-added
products
Planning

Low Spatial Resolution

High temporal resolution

Remote Sensing for grazing management of

sheep: Pasture from Space program
FOO index

Sept
July
May

Feed on offer (FOO) is the amount of pasture in front of an animal at

any one time, measured in kilograms of dry matter per hectare (kg
DM/ha).
Its a balance between pasture growth and the removal of pasture by
grazing animals.
Pastures from Space uses satellite images (MODIS) and field data to
estimate pasture biomass with 97 % accuracy.
`
20
UNOOSA ESA Regional Workshop, Zambia, June 2006

Prediction potential locust invasion (1):

January 2004

January 2006

February 2004

February 2006

March 2004

March 2006

Relatively wet summer 2005/06 (compare with 2004) results in potentially

suitable habitat areas close adjacent to south-west agricultural area.
UNOOSA ESA Regional Workshop, Zambia, June 2006

21

Prediction potential locust invasion (2):

NDVI

Pasture Growth Relative to Last 40 Years

GIS-based
integration

Rainfall Relative to Historical Records

Total Rainfall (mm)

Prediction potential locust invasion (3):

Final Reported
Locust
Hatchings
for
2000
Final Locust
Risk
Map
2000
Final
Spray
Mapfor
for
2000
Locust Risk - numbers
in millions
0 to 5 Million
5 to 100 Million
100 to 250 Million
> 250 Million

UNOOSA ESA Regional Workshop, Zambia, June 2006

23

Developing value-added
products
Monitoring
Very High Resolution
Remote Sensing

High Resolution
Remote Sensing

Monitoring Plant Establishment (Cotton)

Base image acquired within a
week of seeding.

Change in PCD
Map

Second image
acquired 4 weeks
post seeding.

The Bare Soil image serves as

a base from which levels of subtle
vegetation gain can be detected.

Blue areas represent a 2%

gain in total vegetation ground
cover. Red areas represent 0%
gain, or no cotton emergence.

UNOOSA ESA Regional Workshop, Zambia, June 2006

25

Monitoring crop growth

Qualitative monitoring of crop growth, and
variations in crop conditions within a paddock (e.g.
density, canopy vigour or biomass);
Identify critical crop stages;
Monitoring within- & between-field
variability.

UNOOSA ESA Regional Workshop, Zambia, June 2006

26

Remote identification of variation within paddocks

using multi-temporal imagery (Skyplan):
Biomass Images (1994 to 2000)
Mean Biomass Image - 1994 - 1995, 1997 - 2000

1994 Lupins

1998 Lupins

1995 Wheat

1996 Pasture

1999 Wheat

2000 Canola

1997 Wheat

Retrospective, provides areas of variation within paddock

using historical data.
Data: Landsat TM
Enables farmers to isolate good from bad yielding areas.
Yield maps: based on biomass from imagery.
UNOOSA ESA Regional Workshop, Zambia, June 2006

27

Developing value-added
products
Monitoring

Low Spatial Resolution

High temporal resolution

Management Tool to Improve

Profitability for Livestock Producers
Remote sensing of pasture growth rate (PGR) for a
growing season by combining satellite data
(MODIS) with climate and soil data.
The PGR information can assist farmers with
management decisions such as:
grazing rotations, feed budgeting,
fertilizer application and other
"precision agriculture" techniques
Information is delivered to producers by email and
website (http://spatial.agric.wa.gov.au):
FOO monthly (within 5 days of satellite pass),
PGR weekly,
PGR forecasted 7 days forward and historical PGR
UNOOSA ESA Regional Workshop, Zambia, June 2006

29

Pasture from Space Program

PGR spatial distribution

PGR paddock averages

Download Data
UNOOSA ESA Regional Workshop, Zambia, June 2006

31

PGR (kgDM/ha.day)

Pasture Watch

TM

Farm Average
7d Forecast
Real time

Date (Day/Month)

Precision Agriculture and Satellite

based Positioning Systems
Global Navigation Satellite Systems
(GNSS)

Global Navigation Satellite Systems:

GNSS: generic term covering a number of
existing and planned constellations of
satellites together with supporting infrastructure
systems, used for determining positions across
the globe.
Current and proposed satellite navigation systems

GPS
GLONASS
Galileo
EGNOS
Beidou
DORIS and

their associated augmentation

systems.
UNOOSA ESA Regional Workshop, Zambia, June 2006

34

Precision Agriculture and GNSS

Agricultural Uses:

Coarse mapping functions, recording locations (e.g.

weed infestation, insects, etc)
Greatest accuracy is required if a satellite based
positioning systems is used for guidance during
planting and chemical applications (pesticides,
fertilizers).
Reliability is a critical factor for high dynamic
applications like air-spraying.

Farming activities using GNSS:

Soil sampling
Tillage
Drilling
Variable Rate Applications: Fertilizing,
Spraying
Harvesting
UNOOSA ESA Regional Workshop, Zambia, June 2006

GNSS usage in agriculture

Adoption of GNSS technology is concurrent
with grid sampling, yield monitoring, VRA
of fertilizers, seed or pesticide.
Common adoptions (US and Australia):

Grid soil sampling

Fertilizer VRA
Seed VRA
Pesticide VRA
Yield monitor

UNOOSA ESA Regional Workshop, Zambia, June 2006

36

Australia: perceived adoption barriers

High initial costs
Difficulties in assessing the benefit/cost ratio
Survey indicates GNSS-equipped tractors and other
terrestrial agricultural vehicles are being introduced
on Australian farms at a rate of 10% per annum.

Survey of Australian on-farm GNSS usage

Equipment category
High End

Mid range
Harvest monitors

Purpose
Use
Automated steering assist Planting irrigated
systems
cotton
Controlled traffic
farming
Sprayer guidance &
Visual guidance systems mapping
Broadacre cropping
Hand held receivers
Yield monitoring
UNOOSA ESA Regional Workshop, Zambia, June 2006

Accuracy required
2 cm

10-20 cm or submetre
2.5 - 10 m
37

Australian on-farm GNSS usage: terrestrial (1)

Automated steering assist or highend equipment:

Visual guidance or mid-range

equipment:

the tractor or specialized spray/fertiliser rig

is managed entirely by the driver but a
light bar on the bonnet, and often a
moving map display and audible tone assist
the driver to maintain the correct track.
Equipment available offers various levels of
accuracy

Yield monitoring and mapping or

harvest monitors.

the steering of the tractor is at least partly

controlled by satellite signals

Connected to GNSS equipment and

provided with suitable farm office computer
programs can create yield maps.

Low-end equipment

Hand held GPS devices in combination with

palm-sized computers for scouting,
recording specific paddock conditions.
UNOOSA ESA Regional Workshop, Zambia, June 2006

38

Australian on-farm GNSS usage: terrestrial (2)

Drilling

Soil tissue sampling

Varying seed rate application

GNSS enables exact location of

soil samples that are taken. Tests

based on soil characteristics or

environmental factors allows
are used to produce profile maps
farmer to optimise plant
which provide a clear analysis of
populations through regulating
varying soil types and nutrient
drilling rate and depth, helping
status over the recorded area,
maximise cropping potential on a aiding management and
specific basis.
optimising yield potential.

UNOOSA ESA Regional Workshop, Zambia, June 2006

39

Australian on-farm GNSS usage: terrestrial (3)

Machine guidance

Harvesting

High accuracy facilitates the use of Sensors on the combine can

manual and automatic steering
record yield harvesting. Yield
aids. Using satellite technology
for machine guidance helps
reduce skips and overlaps and
maximises operator efficiency.

information allows the farmer to

identify variations in his field. This
information can be used to
investigate reasons for yield
fluctuations and to implement
appropriate management plans.

UNOOSA ESA Regional Workshop, Zambia, June 2006

40

Australian on-farm GNSS usage: terrestrial (4)

VRA

Tillage

Variable rate fertilizer and lime

applications and the use of
nutrient status maps of a field
reduce input costs &
environmental impact. Through
automatic control of sprayers and
fertiliser applicators the amount
of pesticide or nutrient applied
can be varied on the move.

Tillage depth can be varied

according to soil profile or

compaction status. Assisted
guidance for cultivation work
helps minimise the skips and
overlaps and increases working
widths. Autonomous steering
systems allow more machine
operating hours per day, reducing
operator fatigue.

Australian on-farm GNSS usage: aerial &

others (1)
GNSS equipped aircrafts to:
Spray fertiliser and pesticide
Distribute seed onto agricultural properties.
Australian crops currently serviced aerially are cotton,
sorghum, sugar and wheat.

Other perceived uses in agriculture:

GNSS technology to track individual animals,
especially cattle in remote parts of Australia where
herds are measured in thousands.
Unique identification allows tracing back for disease;
precise count of animals and their movement;
In case of disease break, fast mapping of disease
spread based on animal movement.
UNOOSA ESA Regional Workshop, Zambia, June 2006

42

Integrated GIS-Remote Sensing-GPS

advantages:
Ultimate goal of detecting and managing field
variability is to save costs. In Australia, farmers
spend up to 25% of their gross income in
herbicides;
Reduction of pollution risks by applying fertilisers or
herbicides only where it is needed;
Provide field maps of weeds, nutrient deficiencies
that can guide farmers during the spray of their
paddocks;
Assist determining causes of field variability.

UNOOSA ESA Regional Workshop, Zambia, June 2006

43

Acknowledgements:
Department of Agriculture Western Australia
(www.agric.wa.gov.au)
Pasture from Space program
(www.pasturesfromspace.csiro.au/PGR.htm)
OmniSTAR Fugro.
Cooperative Research Centre for Spatial
Information: http://www.crcsi.com.au/
Projects:
www.spatial.curtin.edu.au/~graciela

UNOOSA ESA Regional Workshop, Zambia, June 2006

44