Applications models are the models used to translate the concent=
rations of radionuclides to the parameter of interest. In most of our appli=
cations, the concentrations of radionuclides serve as a *proxy* and =
most parameters are measured indirectly, only some parameters are measured =
directly.

The study of sensing defining soil application models for (geophysical) =
sensor technologies is also called pedometrics.

### Direct measurements

The gamma spectrometer gives a direct measurement of concentrations of r=
adionuclides. Most common are the the concentrations of natural radionuclid=
es (40K, 238U and 232Th) or other man-made radionuclides (or fission produc=
ts) as 137Cs.

### Indirect measurements

Many applications use the concentration of natural radionuclides as a pr=
oxy for the parameter is interest. These indirect measurements are based on=
validated statistical models or pure statistical models.

**Validated statistical models** correlate the concent=
rations of the radionuclides to the parameter of interest. In these validat=
ed models, the principle why these data are correlated is understood. Depen=
ding on the mechanism, the models can be used on a regional scale, or shoul=
d be determined for every site. Examples are parameters as median grainsize=
(for sediments) and clay content.The model for mapping clay, for example, =
depends on geology. This model can be used for the Netherlands as a whole (=
which is one geological region), but models for clay that are found to be v=
alid for dutch soils are not necessarily valid for soils in Spain. The actu=
al translation between radionuclides and soil properties therefore needs to=
be re-established again for each new, geologically different region.

**Pure statistical models** try to correlate the concentrat=
ions of the radionuclides to the parameter of interest. These models are no=
t based on a real understanding of the principle why the radionuclides are =
correlated to the parameter of interest, but just use a statistical correla=
tion. This might work well for a specific site, but there is no guarantee t=
hat this approach works for all situations. These statistical models have t=
o be proven each site that was not part of the dataset used for the statist=
ical analysis. Examples are nutrients of soils and pollution of soil.

### Application models

------=_Part_1197_1409341781.1603543208585--