110 / 218
[
] 110
T
HERE FOLLOWS A
summary of results from a study of
the economic value of short-range forecasts of snow-
storms in winter road-maintenance decisions in
Sweden.
1
The study had two objectives – to assess the value
of these forecasts in economic terms, and to assess ways of
improving decision-making and all aspects of the service
quality. Thus, when quality is mentioned it should be under-
stood to be technical quality (TQ).
When viewed as weather-related decision-making problems
(DMPs), three types of Swedish road-maintenance DMPs can
be identified:
• Snowstorms
• Black-ice
• Frost problems.
Each type of problem arises under distinct meteorological
conditions and involves different road-maintenance strategies.
The case study described here is concerned exclusively with
snowstorm problems.
This decision-analytic approach involves several steps,
including the structuring of the DMP, the quantification of the
relevant costs and losses, and the specification of probabilities
of the relevant weather events given the forecast information
in question. Optimal strategies and forecast-value estimates
are determined here under the assumption that the overall goal
of Swedish road authorities is to minimize total expenditure
where these expenses consist of both road-maintenance costs
and snowstorm losses.
The case study reported here focuses on snowstorm-related
winter road-maintenance activities on major highways in a
district in south-central Sweden. The relationship between fore-
cast quality and forecast value in the context of this
snowstorm/road-maintenance DMP is briefly examined.
Basic structure
A decision tree depicting the basic structure of the snow-
storm/road-maintenance DMP is presented in the diagram
below. The tree identifies the actions and events included in
the model, as well as the outcomes associated with the various
sequences of actions and events. Two sequential decisions are
considered:
• The wake-up decision made by central road-maintenance
authorities
• The pre-salt decision made by local road-maintenance
authorities.
Both decisions are assumed to involve a choice between binary
actions: namely, wake-up (W)/don’t wake-up (W’) in the case
The economic value of snowstorm forecasts
in winter road-maintenance decisions
Erik Liljas, Swedish Meteorological and Hydrological Institute
Basic decision tree for snowstorm/road-maintenance DMP
Source: Liljas, E., and Murphy, A.H.
Wake up
decision
Wake up
period
Pre-salt
decision
Pre-salt
period
O
1
O
2
O
3
O
4
O
5
O
6
O
7
O
8
O
9
O
10
O
11
Maintenance
period
Decision node
Event node
Timeline for decision-making
Source: Liljas, E., and Murphy, A.H.
Wake up
decision
Wake up
period
Pre-salt
decision
Pre-salt
period
Maintenance
period 1
Maintenance
period 2
Time (hours)
0
1
4
6
12