Previous PageTable Of Contents

NIR, remote sensing, and cereal crop management for yield and quality NIR

G.D. Batten1,2,3, A.B. Blakeney1,4, S. Ciavarella5, T. McVicar1,6, B. Dunn1,7 and L.G. Lewin1.

1 Cooperative Research Centre for Sustainable Rice Production, NSW DPI, Yanco Agricultural Institute, PMB Yanco NSW 2703.
2
School of Agricultural & Veterinary Sciences, Charles Sturt University, Wagga Wagga NSW 2678
3
Present address: Sea Spec Pty Ltd, PO Box 487 Woolgoolga NSW 2456
4
Cereal Solutions, PO Box 201 North Ryde, NSW 1670.
5
Irrigation Research & Extension Committee, C/- NSW DPI, NSW DPI, Yanco Agricultural Institute, PMB Yanco NSW 2703.
6
CSIRO, Division of Land & Water, PO Box 1600 Canberra ACT 2601
7
NSW DPI, Yanco Agricultural Institute, PMB Yanco NSW 2703.

Introduction

To meet the quality demands of consumers and remain viable, cereal producers are now able to remotely monitor crops from aircraft and satellites. This technology can, potentially, aid crop fertilizer management. This paper reports that data from infrared images of crops can be utilized to determine the growth and nitrogen content of rice crops at the panicle initiation stage of crop development. This report should encourage further development of NIR-spectroscopy for a range of crops and other land-uses. There is a need to explore the potential of the technology to aid crop fertilizer management to achieve high yields and the desired grain quality. Regular monitoring of crops is suggested to assist management up to and including the harvesting operation. The information, which can be generated, has enormous potential to influence grain-marketing strategies.

Successful cropping involves knowing what the consumers demands; assembling hard data which may be historical, from the current crop year, or from previous seasons and using that information to formulate a management plan usually - with minimal inputs to achieve a sustainable and economic return.

For the rice industry in southern NSW the production and yield per ha plot (Figure1.) reveals the impact of water availability and seasonal conditions. In the last 3 seasons production has been limited by the lack of water to grow rice and also by cold which reduces yields relative to the general increased in yield increase gains achieved.

Crop yields and quality are strongly influenced by the nitrogen supply to the crop (Lacy et al 2004, Batten and Marr 2000.) and a crop nitrogen analysis service, based on near infrared spectroscopy (NIRS) has been used since 1987 to assist growers determine the appropriate rates of nitrogen fertilizer to apply to achieve optimum economic yields, optimise grain quality (protein and colour) and reduce waste of fertilizer which may enter the general environment.

Figure 1. Annual production and average yields for rice in southern NSW 1925 to 2005.

The reliability of any crop assessment is largely determined by the ability of the manager to obtain a representative sample for analysis. Information in maps provided

with samples submitted to the SunRice NIR Tissue Testing Service has revealed that many samples are not representative of the highly variable crops from which they were collected. In some cases the samples were collected using a biased sampling pattern; in other cases the samples did not represent the whole crop.

In recent seasons rice growers have been able to purchase images of crops. Some images were taken using CCD cameras mounted in light aircraft (Terrabyte Pty Ltd, Wagga Wagga, NSW) and offered NDVI (see Eqn 1) at a resolution of 0.5 to 1.0 m. More recently the NDVI images were obtained from the SPOT satellite but the resolution was only 30 m. (Batten 2004, Lewin et al 2005.).

Normalised difference vegetation index (NDVI) = (IR – R) / (IR + R)

(Where IR is the reflectance from the canopy in the waveband range 780 -840 and R is the reflectance from the canopy in the waveband 660 – 680 nm.)

Spectra covering the visible and near infrared regions of the electromagnetic spectrum were available from the Hyperion sensor on Earth Orbiting Satellite EO1 and these were used to develop calibrations against the nitrogen concentration and fresh weight in rice crops growing in the Coleambally Area of NSW in 2003.

The initial calibrations (Table 1) indicate that the full spectrum promises to allow very useful calibrations for fresh weight, nitrogen concentration and nitrogen accumulated in whole shoots (Ciavarella, S., Blakeney, A.B., Batten, G.D. 2005).

These calibrations needed to be confirmed by similar testing in several more seasons. If these are satisfactory then a system can be devised to assess rice and other crops. The following aspects of remote sensing of crops need to be considered -

  • Timely capturing and processing of data to aid crop management
  • Resolution from satellites – smaller pixels will provide better images
  • Spectral resolution (? nm)
  • Atmospheric interferences, including clouds, water, dust, pollution, pollen, and weather at key growth stages
  • The presence of weeds, pests, diseases,
  • Land degradation – low P, high salt
  • Crop variety and irrigation layout changes

 

NDVI (R2)

Full spectrum
(R2 SECV)

N(%) in rice shoots

0.03

0.88 (0.16%)

Fresh weight of shoots (kg/ha)

0.54

0.84 (33 kg / ha)

Nitrogen in shoots (kg/ha)

0.42

0.89 (13 kg / ha)

Table 1. Comparison of calibrations of NDVI and PLS of full spectra (minus water bands) and rice crop parameters.

History suggests we can expect progressive improvement in the prediction of crop N uptake by rice using satellite-born NIR Sensors. The SunRice NIR Tissue Test calibrations improved with time. For example in 1987 the r2 and SEP for nitrogen were 0.93 and 0.15 compared to 0.996 and 0.06% in 2002 (Ciavarella, Blakeney and Batten., unpublished data.)

NIR spectroscopists must work with remote sensing scientists, agronomists, plant breeders, and chemometricians to achieve the gains indicated by this study.

Acknowledgements

We are grateful to Ms Jacqui Boschenok from the FOSS Pacific Pty Ltd North Ryde, for assistance in transferring the Hyperion spectral data into WINISI and to the RIRDC and the Rice CRC for funding the project.

References

Batten, G.D. & Marr, K.M (2000) Protein, minerals and rice paste viscosity. In “Cereals 2000” Eds J.F. Panozzo, M. Ratcliffe, M. Wootton, & C.W. Wrigley. (Royal Australian Chemical Institute, Cereal Chemistry Division, North Melbourne, VIC).

Batten, G.D. (2004) Near infrared spectroscopy: A key to more food, better food and a safer environment. The 2004 Tomas Hirschfeld Lecture, PITTCON Chicago, 7-12 March 2004. NIR News 15 (2), 4 - 8.

Ciavarella, S., Blakeney, A.B., Batten, G.D. (2005) IREC Farmers Newsletter (in press)

Lacy, J. et al. (2004) Ricecheck Recommendations 2004. NSW DPI/RIRDC. pp20.

Lewin, L.L, Ciavarella, S., Blakeney, A.B., Batten, G.D. (2005) System maintenance of the NIR tissue test for maNage rice, other rice research projects and grower tissue samples. Final report to the RIRDC on Project DAN222A. June 2005 pp25

Previous PageTop Of Page