Site specific weed management: mapping the tempo-spatial

development of purple nutsedge (Cyperus rotundus)

Hanan EizenbergTal Naamat, Ran Lati, Tal Miller and Roy Efron

Dept. of Weed Research, ARO, Newe Ya’ar Research Center, Ramat Yishay 30095, Israel

The 2nd International Conference on: Novel and Sustainable Weed Management in Arid and Semi-Arid

Agro-Ecosystems

Objective:The main objective of the entire study is to develop a multiple approach for the detection of purple nutsedge spatial development

To achieve our objective, the state of art in detection technologies for both above and sub-surface spatial development were used

Our approach thus includes: • Spectral analysis of purple nutsedge

obtained from satellites or airborne• Leaf shape analysis obtained from image

data• photogrammetric (3D) analysis obtained

from image data and laser (LIDAR)• Modeling spatial sprouting of the weed

under different environmental conditions• Modeling the soil-subsurface development

and its reflectance on purple nutsedge spatial sprouting

Topic

Far range remote sensing

Close range remote sensing

Modeling spatial growth for the above and sub-

surface

Aerial RGB high resolution camera; 50x50mm pixel size.

Tal Miller et al.

RGB high resolution camera; 0.5x0.5 pixel size. 3D LIDARscenes and photogrammetric. Lati et al. RGB high resolution

camera; 1x1mm pixel size. Efron et al.

TechnologyRGB high resolution camera; 0.5x0.5 mm pixel size. Lati et

al. Minirhyzotron non destructive video camera.

Naamat et al.

Examples 1. Monitoring the sub-surface development of purple nutsedge using minirhyzotron video camera

13.2.08

9.3.08

23.3.08

Examples 2. Developing a robust method for purple nutsedge detection under varies light conditionsorHow to overcome the shading effect?

1-a 1-b 1-c

2-a 2-b 2-c

3-a 3-b 3-c

4-a 4-b 4-c

Fig. 1

Above soil surfacegrowth model

Sub surfacegrowth model

Purple nutsedge

spatial growth model

Biological model

Laser Scanner (description)

Laser Scanner

Color digital camera

Results from cotton field, Newe Yaar, Israel

Laser Scanner (principles)• Ranging - travel time based

measurement • Knowledge of the laser beam

direction determination of point in space

• Direct measurement of 3D coordinates of the objects in space

• Panoramic scan - complete 3D depiction of the surveyed scene can be achieved – generates 3D point clouds

Examples 3. Weed Mapping using web cam (RGB)

• Remote sensing using close range, high resolution (2 mega-pixel) RGB cameras

• Collecting geospatial information using dGPS (sub-meter accuracy)

• Logging information onto an onboard computer

ECPA - July 200817

GPS Antenna

Camera2

Field Of View

Camera1

Field Of View

Computer

MOV07482.MPG

14_right_cam.avi

Binary Image

Preliminary Classified Image

C. rotundus-

Single

C. rotundus-

Cluster

Broadleaved

Weeds

Cotton -

Single

Cotton -

Cluster

Mixture

Shade

Secondary Classified Image

dGPS Data

Weed Map

22

Weed Map•Number of weeds per image•Area of the weeds

Input of meteorological data (T°C; PAR)

Field history

Soil Sub-surface model

Above surface growth sub-

model

Biology Technology

Purple nutsedgegrowth model

Remote sensing for purple nutsedge

detection

Far rangeaerial sensors

Satellite; airborne

Close rangeTerrestrial sensorsRGB, NIR, LIDAR

Fixing data into GIS (using GPS)

Developing an algorithm for weed detection – range related

Creating a purple nutsedge spatial distribution map

Summary

Taking home massage

Integration of the biological and technological means for purple nutsedge detection support and complements each other, and may provide a new dimensions in its precision management.

message

Thanks•To my students, Tal N., Tal M., Ran, Roy.

•To my colleagues from the department of Weed Research Yosi, Radi, Dani, Tal L., Guy, Evgeni.

•To Victor, Yafit and Sagi.•To the Chief Scientist from the Ministry of Agriculture for funding the research.