18.11.2012 Views

Cranfield University

Cranfield University

Cranfield University

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Field Study With Full Size Combine Harvesters<br />

4 FIELD STUDY WITH FULL SIZE COMBINE HARVESTERS<br />

In spring 2006 a field study was conducted aiming to verify soil bin results on a sandy<br />

loam and clay soils with full size combine harvesters. For the detailed field set up, refer to<br />

Section 2.1.2. A description of parameter measurement can be found in Section 2.2 for<br />

penetrometer resistance, rut depth and DBD which were measured similar to the soil bin.<br />

Soil displacement was measured using fishing hooks and a gamma ray probe was used as<br />

well to measure DBD as detailed in Sections 2.2.1.2 and 2.2.3.2, respectively.<br />

The scatter of the measurements in the field results is much larger than in the soil bin labo-<br />

ratory, with high probability due to the inhomogeneities, e.g. variations in water content,<br />

soil texture and structure, etc. in the field soil.<br />

4.1 Results of the field experiment<br />

4.1.1 Soil Displacement<br />

Due to the pattern of the observed soil displacement data from the fishhook experiments<br />

shown exemplarily for the clay soil in Figure 48, different regression functions were evalu-<br />

ated, quadratic, 1/x, logarithmic, and root functions, respectively, all including a linear<br />

term. Attempting to describe the data of all three soils with quadratic and 1/x functions<br />

including a linear term reduced mainly to linear functions. Fitting logarithmic and root<br />

functions, the linear terms disappeared in the majority of cases, but the differences of the<br />

R 2 for the linear, the logarithmic, and the square root functions are marginal with less than<br />

2 % for the clay and the subsoiled sandy loam. The R 2 is greater than 0.7 for the wheeled<br />

configurations and exceeds 0.45 for the tracked treatments. As Figure 48 shows the linear<br />

term is appropriate because it has zero displacement at a depth where the majority of data<br />

from this depth and below is zero. This confirms the findings discussed in 2.2.1.1 assuming<br />

to approximate soil displacement by a linear function.<br />

On the shallow tilled sandy loam plot, no coherent picture was gained. For the track data<br />

all models reduced to a linear regression function with R 2 greater than 0.7. But for the<br />

wheeled treatments, the full model (linear plus quadratic) fitted best with R 2 greater than<br />

0.75. However, as full models had never been chosen before, and the particular data distri-<br />

Ph.D. Thesis Dirk Ansorge (2007)<br />

61

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!