Methodology
A winter maintenance cost model was established to estimate the materials, labor and equipment for a given maintenance unit (e.g., shed, garage, cost center, or patrol yard) as a function of the following factors:

• Winter maintenance cost for the maintenance unit per winter season
• Lane miles of roadway maintained by the maintenance unit
• Level of service of the roadways maintained by the maintenance unit, often characterized by the pavement condition
• Winter severity index for the area managed by the maintenance unit
• Weather information usage (frequency and accuracy) by the maintenance unit
• Level of anti-icing used by the maintenance unit

Sensitivity analysis and an artificial neural network were applied to the model to provide estimates of winter maintenance costs if the state had not purchased weather observations (base case) as compared with the real world winter maintenance operating during the winter season. A third scenario estimated the benefits if usage and accuracy of weather information increased.

Findings
For each state, the benefit cost analysis shows that the use of weather information brings more benefits than costs. The case studies collectively showed that winter maintenance costs decreased as the use of weather information increased or its accuracy improved. The results are summarized in the table below.

Case Study State	Winter Season	Winter Maintenance Cost ($ 000s)	Benefits ($ 000s)	Weather Information Cost ($ 000s)	Benefit-Cost Ratio	Benefits/Maintenance Costs (%)
Iowa	2006-07	14,634	814	448	1.8	5.6
Nevada	2006-07	8,924	576	181	3.2	6.5
Michigan	2007-08	31,530	272	7.4	36.7	0.9

The Michigan case had the highest benefit-cost ratio due to low costs in weather service. The percentages of benefits over total winter maintenance costs were 5.6 percent (Iowa), 6.5 percent (Nevada), and 0.9 percent (Michigan). The benefit-cost ratios of Iowa and Nevada may be more representative because the costs associated with weather information in these two states were based on statewide numbers, while the Michigan case was not.

Note that amortized RWIS capital costs were excluded when calculating the cost of weather information for winter maintenance. The costs of maintaining the RWIS sensors were included as part of the maintenance contract; however, the in-house equipment and personnel costs related to RWIS maintenance were not included.

The benefit cost analysis only considered state transportation agency benefits and did not include benefits to motorist and society. The case studies show that the use of weather information is able to reduce resource usage, which in turn can reduce degradation of the surrounding environment, corrosion on vehicles, and infrastructure damage. In addition, it will benefit motorists by reducing delay and improving safety.