7 August 1998

How accurate is your data?

By Ian Marshall

ACCURACY of on-combine yield monitors is critical if they are to prove effective, says Germanys DLGtesting station.

With the publication of its latest test report in Germany, farmers can now compare manufacturers claims with figures obtained from independent, long-term field trials.

In the tests, practical work was supplemented by bench tests which show sloping ground has the greatest influence on monitor accuracy.

Carried out between 1991 and 1995, the field trials covered four of the most popular systems on the market: RDSs Ceres 2 and Claas Quantimeter II, MFs Datavision Flowcontrol and LH Agros LH 565. Bench tests took place in 1997.

Ceres 2 and Quantimeter II (which uses RDS sensors), operate on a volumetric measuring principle. Flowcontrol and the LH565 determine the mass of grain flow; the former by means of radiometric measuring; the latter through a force measuring cell.

The monitors were evaluated on a number of different models of combine harvester in different cereal types in light to medium yielding soils. The area covered with each system ranged from 130-140ha (321 – 346 acres), which produced 132 – 182 tank loads of grain, with each monitors level of accuracy established by weighing tank loads on calibrated platform scales.

Results indicate that, in the field, there is not much to choose between the systems (table 1). When interpreting the figures, mean relative calibration errors (MRCE) – representing the accuracy of calibration, which should measure, or be close to, zero – was achieved by all the meters.

Standard deviation of the relative error (SDRE) – the indication of measuring accuracy – shows the range of error within which about two-thirds of all measurements lie. Here, too, with an approximately equal range of errors of  3.5% –   4%, the meters were similar, despite the different principles of measurement used.

Test bench studies were conducted to determine the accuracy of the measuring systems under identical, clearly defined conditions, with particular consideration given to the effect of different throughput levels and of lateral and longitudinal tilt.

On level ground simulations with throughputs from 10-35t/hour, MRCE ranged from +0.56% (Claas Quantimeter II) to -1% (RDS Ceres 2), with the an SDRE range of +/- 0.94% (RDS Ceres 2) – +/-3.15% (LH 565).

But, it was found, at a constant throughput of 20t/hour the angle of work of the combine exerted the greatest influence on the accuracy of the monitors.

In bench studies at this throughput at 5, 10 and 15% of left and right lateral tilt, and forward and back longitudinal tilt, the radiometric principle was least affected and the volumetric most (table 2).

The two volumetric systems have sensors to counteract the influence of slope, points out the DLG, but, it says, the errors caused by lateral and longitudinal tilt cannot be successfully compensated under all conditions. &#42

Table 1:In-field errors of yield monitors

Monitor Length of Combines/ Area Tank MRCE SDRE

study (yrs) types of crop (ha) loads

Ceres 2/Quantimeter II 3 3/4 140 179 -0.14 +/-3.43

Flowcontrol 2 2/2 140 132 -1.01 +/-4.07

LH 565 3 3/4 130 182 -1.83 +/-4.06

Table 2:Errors of yield

measuring systems at

different tilts

Monitor MRCE (%) DRE (%)

Ceres 2 3.20 +/-6.04

Quantimeter II -1.03 +/-4.27

Flowcontrol -0.71 +/-1.53

LH 565 -2.84 +/-1.66