Page 172 - Freshwater-Biology-and-Ecology-Handbook

CHAPTER 3 4.3 4.3 Probabilistic classification

• RICT2 also takes error into account to indicate the
The WFD requires that a high level of confidence
relative probability of the site being in each of the 5-status
and precision of the classifications should
classes. This provides information about the confidence
be achieved. This is done by taking error into
of class. Where confidence is low, you will need more
evidence before taking expensive remedial action. RICT
account. RICT2 does this in two ways:
is able to provide a probabilistic classification because it
uses Monte Carlo simulation to take errors into account
• RICT2 indicates the suitability of the RIVPACS (Figure 3.22). The resulting probabilistic classification
predictive model to the site in question and indicates enables RICT2 to compare two classifications
when the site is beyond the model’s capability because statistically in order to indicate the statistical certainty
it is not covered adequately by the typologies included that a classification has changed. These outputs
in the reference samples, ie the combination of are described in Section 4.4 and information about
environmental parameters is different to that of any of interpreting them is included in Section 5.
the sites in the reference database.

Option 2 River invertebrate classification with RICT2

Ch2 S6 & S7
RIVPACS sampling
Ch2 S12 RIVPACS
laboratory analysis
Ch2 S7.6 Sample Ch2 S13 & Ch3 S2.5
environmental data Laboratory analytical
annual average error Ch3 S2.4 Calculate
(preferably long-term) Bias from audit WHPT indices

Ch3 S3.2 & S3.3 Ch3 S3.2 & S3.3
RIVPACS Prediction Calculate EQR
Ch3 S2.5 & 4.4 100,000 simulations
Biological and
Ch3 S3.2 Map Ch3 S4.4 Suitability environmental error
environmental data (of RIVPACS for the site)
to vary input Average spring and
From research studies autumn EQRs
Ch3 S4.1 WFD status
class boundaries Optional for acid
Ch3 S4.2 Classify sites
Number of sites EQRs
Ch3 S6 Acidification
classification
Number of years Ch3 S4.4 Combine
data WFD AWICsp
classifications from
Monte Carlo all simulations
simulation Probabilities of class
Key User action or data
Numbers refer RICT software
to chapters and Ch3 S4.1 Combine Combine MINTA
sections of this ASPT and NTaxa class and WFD AWICsp
handbook WFD AWIC software Invertebrate status = WFD status class class =
MINTA class Combined class
RICT2
Outside RICT2 Acidification
General degradation (Scotland & Wales only)

Figure 3.22
A more detailed outline of the classification process than that provided in Figure 3.18, indicating
the role of Monte Carlo simulation to consider error in calculating a probabilistic classification

172 | Freshwater Biology and Ecology Handbook
–