Accuracy
versus precision
Soil carbon should be measured as accurate and precise as possible. The concept of accuracy and precision is illustrated in Figure 1,2 by a simple example.
Figure 1. Accuracy is the proximity of
measurement results to the true value; precision, the repeatability, or
reproducibility of the measurement (http://en.wikipedia.org/wiki/Accuracy_and_precision).
Figure 2. Soil carbon measurement should
target high precision and accurate measurements (Case 1). Case 1: all
measurements are close to the true value
(accurate) and all measurements are close to each other or less
variation among measurements (precise);
Case 2: the measurement is precise but all measurements are far from the actual
value (center) which makes it biased and inaccurate; Case 3: most measurements
are reassembly close to the true value (accurate) but not precise; Case 4: all
measurements far from the true value (less accurate) and there is high
variation between measurements (less precise).
See table
1 for major sources of uncertainties in SOC measurement.
Table 1. Source of uncertainty in SOC measurement
Activity |
Sources
of uncertainty |
Sampling |
Sampling design (random, stratified random) |
Sample
size |
|
SOC
measurement |
Natural
variability (spatial) |
Sample
preparation (e.g. contamination, subsampling) |
|
Lab
method used (instrument resolution) |
|
Human
error |
|
Field
data collection (e.g. soil mass, volume) |
|
SOC
prediction using IR |
Imported
uncertainties (from reference data) |
Model
(assumption) |
|
Instrument
and human errors |
|
Mapping
SOC |
Covariates
used |
Image
pre -processing (geometric and radiometric corrections) |
|
Scale/resolution
(e.g. farm vs landscape) |
|
Model
(assumption, strength) |