Site-specific fertilizer application technique requires geographical position of features linked with information on amount of fertilizer to be applied within these features. In principle, three approaches to obtain the required information are conceivable. An application map for nutrients on grassland can be calculated on previously determined spatial yield distribution with the intention to simply substitute extracted nutrients, but with the restriction that the quality of prediction of historical yield data might be limited. Small forage harvesters carrying an integrated weighing system have been developed for grassland research that can rapidly cut samples and automatically separate aliquots for laboratory analyses. From such plant samples, the amount of nutrients to be substituted with fertilizers has then to be calculated either from additional nutrient analyses or through estimates of critical nutrient concentration (Lemaire and Gastal, 1997). Alternatively to yield measurements, a diagnosis of plant nutrient status based on small samples has been developed for highly productive
perennial ryegrass (Beaufils, 1973; Walworth and Sumner, 1986; Bailey et al., 1997a, Bailey et al., 1997b; Bailey et al., 2000). This technique has successfully been tested on grassland silage fields to predict the spatial distribution of nutrient sufficiency status, thus avoiding laborious and time consuming clipping of voluminous sward samples. The third approach is the one that uses N sensors (www.yara.de) aiming to obtain reflectance properties above canopies that allow the detection of local N deficiencies. However, this technique has only been tested successfully in cereals but not yet calibrated on species rich and diverse grassland. So, there is a need to verify that this technique is applicable on grassland where manifold interactions exist of spectral properties with floristic composition, standing biomass, sward density, height, and plant morphology.
At present there is no rapid technique available or sufficiently tested that allows to measure and map spatial yield distribution within fields without manual or mechanical clipping, except one development that is provided by a New Zealand enterprise, the rapid pasture meter (www.c-dax.com, www.farmworkspfs.co.nz). This instrument reads the pasture height by sensors that are mounted in a metal frame pulled through the pasture. However, the rapid pasture meter does not allow measurements in taller grass stands, e.g. in silage fields.
Further, calibration is required to convert sensor data adequately into dry matter data depending on species composition and density of the sward. In conclusion, there is an important gap in technological development that hinders the site-specific application of fertilizers due to missing “ground truth information”.
Provided that the spatial and temporal heterogeneity of soil nutrient status and potential nutrient extraction could be either mapped or detected on-the-go, the problem remains how to calculate local N demand without knowing future N extraction of the momentary growing grass crop. Even with grassland growth models, the simulation of yield performance and N extraction would implicate intolerable uncertainty of prediction, as growing conditions cannot be anticipated over a long term. The response of yield and N extraction to fertilizer application and its variation throughout multiple years can be proven by long-term field trials (Hejcman and Schellberg, this issue). On grassland, such annual yield variation can mainly be explained by the response to changes in actual precipitation that is difficult or even impossible to predict.
Unless the above discussed uncertainties, the development of site-specific fertilizer application techniques has been considered attractive by research groups on species rich and frequently utilized grassland. With chemical-synthetic fertilizer, application technique will be the same on grassland as on arable land. However, with slurry as the main nutrient source on grassland farms, new application techniques need to be developed. Requirements of site-specific slurry application on grassland are listed in table 1. Albeit current difficulties to determine local nutrient demand, technical realisation of site-specific application is possible.
For rapid determination of organic N and NH4+
content in slurry, the accuracy of traditional analyses have been compared in quick tests (Van Kessel and Reeves, 2000), where Quantofix-N-Volumeter (Klasse and Werner, 1987) performed best. Recently, a near-infrared reflectance spectroscopy (NIRS) sensor has been developed that allows rapid and comfortable measurements of N content during filling of the slurry tank (Dolud, 2005). This technique can potentially be applied also on the running tank trailer after calibration.
Table 1. Requirements of site-specific slurry application technique and a ranking of its technical standard and precision
technical standard1 precision2
detection of nutrient demand
DM yield or nutrient yield map3 4 2
online detection of crop nutritional status3 5 n.a.
slurry application
determination of nutrient content in slurry 2 2
permanent homogenisation of slurry 1 n.a.
spreading technique 2 2
rapid slurry flow rate control 2 2
measurement of flow rate 2 4
navigation and mapping
measurement of travelling speed 1 1
high-resolution GNSS monitoring 1 1
as-applied-map 2 n.a.
1 ranking from easily achievable = 1 to laborious, time consuming or technically not yet available = 6
2 ranking from sufficient = 1 to insufficient = 6
3alternatively required
n.a. = information not available
With the development of trailing hose systems and injectors, precision of slurry distribution has been optimized. Transverse distribution has been tested on a test bench, indicating that most spreaders available in Germany and Austria showed good distribution even on slopes (Sauter, 2004). The measurement of actual flow rate is as well possible with good precision through electronic flow meters installed between tank and spreader.
Navigation and mapping during application is not a challenge too, but the comparison of as-applied maps with application maps is still missing.
Based on the above discussed technical and methodological requirements, a flow diagram demonstrating the course of action in site-specific slurry application has been laid out in figure 2. The procedure follows current research activities at Bonn University (Germany). The technical realization of slurry flow control is based on three main components, (i) a control software that reads the amount of slurry to be applied in the digital map depending on the GNSS position in the field, (ii) a control valve linked to the micro-controller that steers the outlet of the slurry container and (iii) the flow meter that is checking the actual flow in a feedback loop. The design of an existing technique is given in figure 3.
Preliminary testing of the system demonstrated that the development of the electronic control was most challenging.
Figure 2. Scheme of a site-specific slurry application system based either on (A) “hard” data acquisition on crop status and yield through sampling or harvest mapping and (B) “soft” data acquisition of present crop status by means of close range or remotely sensed crop detection technology. Dotted lines indicate data flow, solid lines indicate flow of slurry.
Figure 3. Design of a site-specific slurry application technique based on a priori calculation of an application map on grassland.