Precision Agriculture - New Production Technologies for an Old Crop

 

Haneklaus, S., Bloem, E. and Schnug, E.

 

Institute of Plant Nutrition and Soil Science, Federal Agricultural Research Center Braunschweig-Voelkenrode, Bundesallee 50, D-38116 Braunschweig, Germany, e-mail: silvia.haneklaus@fal.de

 

Abstract

Oilseed rape cropping requires inputs of different origins and most of them are maintained according to certain soil parameters. One major problem is that the variability of soil fertility features together with the uniform treatment of fields causes inefficient factor utilization and unnecessary environmental burdens. Thus neither the potential yield, nor the sustainability of oilseed rape cropping may be optimally achieved. The conceptual idea behind precision agriculture is that of managing resources locally in order to address inputs correctly to soil features. In this contribution precision agriculture technologies with special view to different sampling strategies and practical variable rate fertilizer input of the environmentally relevant elements, nitrogen and sulfur, are discussed.

 

Keywords:  Directed sampling, Global Positing System (GPS), self-surveying, semi-variogram, surf-eyes, Variable Application Rate (VAR)

 

Introduction

 

Man has probably been aware of the problem of spatial soil variability ever since cropping began (10.000 bc) but it took nearly till the end of this millennium before precision agriculture technologies were available to address it in terms of variable application rate technology. Four years ago precision agriculture (PA) technologies were practically applied exclusively to geocoded soil sampling and on-line yield measurements, while meanwhile variable application rate (VAR) technology for fertilization was implemented widely on production fields (Haneklaus et al., 1998; Simchen and Schnug, 1998).

The use of nutrients in agriculture causes environmental problems which need to be solved whilst still maintaining the profitability of agriculture. In oilseed rape cropping regular surpluses in the nitrogen (N) balance (Schumann et al., 1997) and increased N2O emissions (Kohrs, 1999) make it a highly important nutrient for VARs in order to reduce N losses and minimize N surpluses (Haneklaus and Schnug, 1998). Severe S deficiency is a major problem in the mineral nutrition of oilseed rape which reduces N utilization drastically (Schnug et al., 1993). It was the aim of this contribution to discuss strategies for geocoded soil sampling and to demonstrate a site-specific approach for a VAR of N and S.

 

Materials and Methods

 

Geocoded soil samples were taken in 0.3 m layers up to a depth of 0.9 m in a 30*30 meter grid in Mariensee (E 9° 28’; N 52° 33’) and Neuenkirchen (E 10° 28’; N 52° 02’ ). In Kassow (E 12° 06’; N 53° 10’) soil samples were taken along a transect in 0.3 m layers up to a depth of 1.5 m (Figure 1). Soil samples were air-dried and sieved at 2mm. Available N (Nmin) was determined according to Anon (1991) and available sulfate (SO4) according to Bloem (1998) before each of in total two N applications. S was not applied. Geo-referenced yield data were recorded on-line employing the flowmeter sensor system (Murphy et al., 1995). Allocation of and navigation to sampling positions were carried out by a differentially corrected Global Positioning System (DGPS) with an accuracy of ±1 to 2 meter.

 

Figure 1. Topography and sampling locations along a transect in Kassow.

 

 

Results and Discussion

 

Soil sampling strategies: grid sampling

The practicability of determining the N and S demand on the basis of the spatial variability of the available Nmin. and SO4 contents in the soil was tested on two sites in northern Germany, Mariensee and Kassow. The results of the semi-variogram analysis reveal ranges of 121 meter for N and 64 meter for S at the first sampling date (Figure 2).

 

 

 

Figure 2. Semi-variogram for Nmin. contents in the soil in Mariensee and SO4 contents in Neuenkirchen.

 

Neither in Mariensee, nor in Neuenkirchen could a spatial correlation be verified for the second date of sampling (Haneklaus et al., 1998; Bloem, 1998). This means that even with sampling distances of 30 meters, spatial correlations need not necessarily be found so that the local fertilizer demand cannot be prognosed with sufficient accuracy. It is well known from SO4 which is as mobile as NO3 in soil that sampling distances of £25 meters are required to find a spatial correlation (Schnug and Haneklaus, 1998). The required high sampling density, high temporal fluctuance and high analytical costs therefore exclude its practical utilization for fertilizer recommendations.

 

Remote sensing

The utilization of remote sensed images from satellites and airborne platforms for the prognosis of the N demand of crops carry both the problem of availability and also unsuitable survey conditions (mostly due to surface coverage by clouds). Further details are given by Panten et al. (1998). Additionally, remote sensing applied to crops has the general problem that differences in the spectral  reflectance of the canopy can have more than one reason. Changes in the "green" color of crops are most likely due to changes in the chlorophyll content and the nitrogen content of the plant. This is most often attributed to the nitrogen supply of the crop and thus claimed as a suitable information for the design of nitrogen dressings (Baret and Fourty 1997). But there are many more reasons why the canopy´s green color can change - for instance the S supply of crops has become the major explanation for reasons of "green" color intensity of crops in the northern hemisphere (Schnug and Haneklaus 1998). ‘Surf-eyes’ may overcome these problems efficiently. Surf-eyes are fixed scanner systems on elevated positions scattered in the landscape (Panten et al., 1998). By means of surf-eyes, real time images of crop and soil surfaces can be continuously recorded and automatically rectified and geo-referenced in a GIS (Geographical Information System). Thus sampling points for directed sampling (see below (Schnug et al., 1998)) can easily be localized and results more or less instantly transferred into variable rate management on the fields.

 

Self-surveying and directed sampling

Basic long-term variable parameters such as texture, organic matter content and landscape geomorphology highly influence the productivity of soils (Franko, 1996; Thompson and Robert 1995). The ‘self-surveying’ approach which combines GPS navigation and positioning with human sensory capabilities, efficiently provides geocoded information of these time-constant soil parameters in optional density (Haneklaus et al., 1998).

With knowledge of the spatial variability of these key variables, equifertiles can be established. Equifertiles are areas in the field with identical or similar productive capacity in space and time (Schnug et al. 1994). Within equifertiles permanent monitoring points are located which reflect temporal changes in the corresponding equifertile areas and thus provide information on factors causing differences in productivity. If these points are continuously chosen for directed sampling, they reflect the variability of soil parameters including short-term variable factors such as nitrate without losing the spatial correlation. A detailed description of directed sampling is given by Schnug et al. (1998).

 

Evaluation of VARs for N and S

Optimum VARs have been calculated for N and S for sampling locations along a transect in Kassow (Figure 1; Table 1). Geomorphology has a strong influence on N fluxes as the variation in topography indicates (Figure 1). Variation of the N rate in dependence on the variability of geomorphology has already been successfully applied under farming conditions in Kassow (Schnug and Haneklaus, 1999). Along the transect a significant negative correlation was determined for the relationship between organic matter contents and seed yield (r = -0.64). Therefore optimum variable rates were calculated in dependence on the variation of the organic matter content (relative variation of N rate: 60 – 135%) and geomorphology (relative variation of N rate: 40 – 120%) assuming an equal impact on the N supply (Table 1).

The investigations further showed that there was a close relationship between clay and water content in the deeper soil layers (0.6 – 1.5 m) and the S supply of the oilseed rape crop (clay: r=0.71 (p<0.05); water: r=0.80 (p<0.01)). The S contents in younger leaves of oilseed rape ranged from 0.28% S (severe S deficiency) to 0.53% S (moderate S deficiency). Aiming at a total S level of at least 0.65% S for achieving maximum yields, corresponding optimum fertilizer rates would be varied between 40 and 70 kg ha-1 S (Table 1).

The spatial variability of hydrological and physical soil parameters on a larger scale (1:100.000) can be derived from topographic soil maps which allow a reliable evaluation of the variation of the S supply within a larger area (Bloem, 1998). No topographic soil maps providing this information are, however, available on field level (1:5.000). Access to data concerning hydrological and physical soil parameters in soil layers of up to 1.5m depth will therefore be the major obstacle for a VAR of S in precision agriculture.

 

Table 1. Optimum VARs for N and S based on the variability of key variables (mean values) – a case study for sampling locations along a transect in Kassow.

 

Sample location

Geo-morphology

OMC1

Clay

OMC

Clay

Rel. seed yield2

VARs

 

 

 (0 – 0.3 m)

(0.6 – 1.5 m)

 

N3

S4

 

 

------------------------------------  (%) ------------------------------------

1

Summit

1.4

  2.9

11.0

14.4

  32

120

100

2

Slope

1.3

  6.5

15.7

14.7

122

  80

  90

3

Plateau

1.3

  3.9

  6.5

11.8

118

  95

135

4

Depression

1.3

  5.3

  4.5

10.2

  76

  50

140

5

Slope

1.7

10.4

11.6

11.7

  46

  70

125

6

Summit

1.5

  9.0

15.1

14.3

  84

120

100

7

Plateau

1.0

11.4

19.1

16.5

118

100

  80

8

Plateau

1.4

12.8

14.7

12.1

  97

  95

110

9

Slope

1.4

  9.9

13.5

13.3

  93

  75

105

10

Depression

1.2

  6.0

  6.9

13.7

  84

  50

120

11

Summit

1.2

  9.2

12.6

11.9

  89

125

110

note: 1OMC = organic matter content; 2100% = 2.4 t/ha;  2100% = 150 kg ha-1 N; 3100% = 50 kg ha-1 S

 

Conclusion

 

A balanced N fertilization is a major need in order to achieve sustainable oilseed rape production. Under conditions of severe S deficiency the N uptake of oilseed rape was reduced up to 56% at sampling locations along a transect in Kassow (Figure 1). Decision making strategies for highly mobile nutrients such as N and S, however, cannot be based practically on grid soil sampling and analysis of available nutrient contents as both parameters show a high spatial and temporal variability which would require sampling distances < 30 meters. Other sources of soil and crop information e.g. remote sensed data miss practicability under present conditions. An efficient alternative may become surf-eyes which could provide farmers with a continuous information on the N and S nutritional status of the oilseed rape crop.

At present, a successful decision making strategy for VARs of N which has been verified over three years (Haneklaus et al., 1998) is by collecting information of the spatial variability of long-term stable key parameters such as organic matter, soil texture and geomorphology via self-surveying. Though hydrological and physical soil parameters proved to be major factors determining the S supply of oilseed rape, an integration of these relationships into precision agriculture will not be practicable due to too intensive sampling efforts.

 

Acknowledgement

 

The authors cordially thank Dr. K. C. Walker (SAC, Aberdeen) for the linguistic improvement of this paper.

 

 

 

 

 

References

 

Anon 1991. Bestimmung von mineralischem (Nitrat-) Stickstoff in Bodenprofilen (Nmin.-Labormethode). Ed. Hoffman. G. Methodenbuch I. VDLUFA Verlag. Darmstadt. Kap. A.9.1.4.1

Baret, F. and Th. Fourty. 1997. Radiometric Estimates of Nitrogen Status of Leaves  and Canopies. 201 - 227. In G. Lemaire (ed.), Diagnosis of the Nitrogen Status in  Crops. Springer Verlag, Heidelberg.

Bloem. E. 1998. Schwefel-Bilanz von Agrar-Oekosystemen unter besonderer Beruecksichtigung hydrologischer und bodenphysikalischer Standorteigenschaften. Sonderheft Landbauforschung Voelkenrode, ISBN 3-933140-13-7.

Franko, U. 1996. Simulation der Kohlenstoff-Stickstoff-Dynamik in Agrarlandschaften. Landbauforschung Völkenrode 3/1996. 114-120.

Haneklaus. S. and E. Schnug 1998. Impacts of precision agriculture technologies on fertilization. Proc. 11th Int. Symposium of CIEC. “Codes of good fertilizer practice and balanced fertilization”, eds E. Schnug and M. Fotyma, 95-107.

Haneklaus, S., Paulsen, H. M., Schroeder, D., Leopold, U. and E. Schnug 1998. Self-Surveying - A Strategy for Efficient Mapping of the Spatial Variability of Time Constant Soil Parameters. Commun. Soil Sci. Plant Anal. 29 (11):1593-1601.

Haneklaus. S.. Schroeder. D. and E. Schnug. 1998. Decision Making Strategies for Fertilizer Use in Precision Agriculture. Proc. of the 4th Int. Conf. on Precision Agriculture. ASA-CSSA-SSSA Madison. (in press).

Kohrs. K. 1999. Beziehungen zwischen N-Versorgung und Freisetzung von N2O-N in unterschiedlichen Fruchtfolgen. Sonderheft Landbauforschung Voelkenrode, ISBN 3-933140-15-3.

Murphy. D. P.. Schnug. E. and S. Haneklaus. 1995. Yield mapping - a guide to improve techniques and strategies Proc. of the 3rd Int. Conf. on Precision Agriculture. Minneapolis. ASA-CSSA-SSSA Madison. 33-47.

Panten, K., Haneklaus, S., Vanoverstraaten, M., Schroeder, D.,. and E. Schnug. 1998. Remote sensing  as an aid for the spatial management of nutrients. Proc. of the 4th Int. Conf. on Precision Agriculture. ASA-CSSA-SSSA Madison. (in press).

Schnug, E. and S. Haneklaus. 1998. Diagnosis of sulphur nutrition. In: Schnug, E. and Beringer, H. (ed.): Sulphur in Agro-Ecosystems. Vol. 2 of the series ´Mineral Nutrition in Ecosystems´,  Kluwer Academic Publ. Dordrecht. 1-38.

Schnug, E., Panten, K. and S. Haneklaus. 1998. Sampling and nutrient recommendations – the future. Commun. Soil Sci. Plant Anal. 29 (11-14), 1455-1462

Schnug. E.. Haneklaus. S. and D. Murphy. 1993. Impact of sulphur fertilisation on fertiliser nitrogen efficiency. Sulphur in Agriculture 17: 8-12.

Schnug, E., Haneklaus, S. and Murphy, D. E. P. 1994. Equifertiles - an innovative concept for efficient sampling in the local resource management of agricultural soils.  Aspects of Applied Biology 37 (1994), 63-72.

Schnug, E., Panten, K. and Haneklaus, S. 1998. Soil sampling and nutrient recommendations - the future. Comm. Soil Sci. Plant Anal. 29 (11-14): 1455-1462.

Schumann. M.. Kücke. M. und E. Schnug. 1997. Fallstudien und Konzeption zur Einführung bilanzorientierter Düngung in der deutschen Landwirtschaft. Landbauforschung Voelkenrode. Sonderheft 180.

Simchen. H. and E. Schnug. 1998. From serf to satellite. AgroPrecise 1: 14-15.

Thompson, W. H. and Robert, P. C. 1995. Evaluation of Mapping Strategies for Variable Rate Applications. Site-Specific management for Agricultural Systems, Proc. of the 2nd International Conference, March 27-30, 1994. ASA, CSSA, SSSA. 303-323.