Small-area specific maintenance fertilisation - for good quality | 21.09.2010

There are huge differences in the supply of the basic nutrients (P, K, Mg) and lime between fields and also between parts of fields. In order to take account of this, sub-area specific fertilisation is carried out in farm practice. The quality of the soil sampling plays a central role here, as the overall planning for four years is based on its results. There are many things that can go wrong here.

figure 1

figure 2

The field trial of the University of Halle (Figure 1) shows how small scale the nutrient distribution can be on agricultural areas. Sixteen individual samples were taken and analysed in a 1 ha raster cell. It was clear that in this raster cell alone the distribution of phosphorous varied from class “A” – extreme under-supply- to class “E” – very well supplied. With point sampling there is a danger that extreme locations will be over-evaluated. The route taken in figure 2 (Z form) helps to avoid this over-evaluation if the mixed sample is made up of many individual samples along this line.

Such small-scale cultivation can surely not be adequately taken into account in the usual farm practice. It brings up the question as to what extent small-scale trends in nutrient supply can be sufficiently recorded in a replicable manner with soil sampling in 1, 3 and 5 ha rasters.

In order to follow up this question the Ministry for Environment, Agriculture and Geology of Saxony carried out sampling trials in 2006, together with Agri Con. A 70 ha area was assessed with the help of soil scanner measurements with respect to its heterogeneity. This provided the basic information for an intelligent sampling raster. Six samplers were given only this in order to enable the implementation to be as realistic and practice oriented as possible. The course of the sampling line was decided on by each participant. All samplers were equipped with similar, machine based sampling equipment.

Different people sample at different places

It was clear from the recorded sample trial that different people sample at different places. The sampling took place as mixed samples with 15 to 20 individual samples, in “N” or “Z” form with a track length of 150 to 200 m, GPS supported and using machinery. 

In 99 % identical results

The evaluation of the six nutrient distribution maps showed a high degree of agreement visually (figure 2). All six sampling persons delivered an identical result for 99% of the area. An agreement of 100 % was achieved with respect to the pH value.

Quality characteristics for the next sampling

The following quality characteristics should be noted in order to achieve such good results with the repeatability of the sampling results each time:

Raster size: A large sampling raster tends to lead to ever larger deviations from the real nutrient heterogeneity. The sampling rasters currently used in farm practice of 1, 3 and 5 ha are, in the end, a compromise of significance and cost. The following is true: the smaller the raster, the more precise are the results, but more expensive

Raster layout: the flexible adaptation of the sampling raster helps to avoid mixed samples being taken from sub-areas with very different growth conditions. Indirectly, the heterogeneity can be measured with the soil scanner. From a technical point of view, the values measured by this device are to be assigned to the appropriate quality of the location.

GPS-recording: the GPS recording of each individual point has two important functions. Firstly: only in this way can the results of the analysis be exactly spatially aligned to the corresponding area. Secondly: they guarantee a consistent, verifiable quality from the first to the last sample of the day.

Sampling: an area related sampling in the form of a „N“ or a „Z“ is definitely better than the point sampling. The explanation is given by the 1 ha raster cell in the trial by the University of Halle: if only one point sample was taken on this 1 ha, there would be a danger that an “A”or an “E” supply would be obtained as the result. Both of these extremes occur in this raster cell. The average, however, lies elsewhere.

Sampling depth: Usually a sampling depth of 0-20 cm is stipulated for arable land. In long-term non-turning cultivation on one location it can make sense to take a sample at a depth of up to 30 cm, in order to take into account possible layering of nutrients.

If all these quality criteria are to be adhered to – in particular the representative, area related sampling – then 70 to 80 samples per day appear to be the upper limit.