Back

 

Improving INL Wind Forecasting with Cluster Analysis of Wind Patterns 

Background

NOAA ARL-FRD has provided INL Site climatology monitoring and specialized weather forecasting such as wind forecasts for nearly 60 years. Understanding wind patterns and forecasting winds are of interest at the INL Site for various applications. Short-term wind forecasts are used during emergency operations to track the potential transport of hazardous substances and in predicting the spread of wildfires. Wind forecasts are also important for the safety of personnel and the effi ciency of many routine INL Site activities. For example, some operations at the INL Site can only be conducted when the wind speed remains under specific thresholds. Climatological wind patterns are also one factor that must be considered as part of INL Site’s efforts to ensure it meets regulatory requirements for public safety.

The meteorologists working at NOAA ARL-FRD have long known that a few typical wind patterns recur frequently across the INL Site and the surrounding area. For example, meteorologists have learned to expect northeast, down-valley winds on summer mornings and up-valley, southwesterly flows on summer afternoons. Cluster analysis is one mathematical approach used to identify common patterns in data. A cluster analysis of the NOAA INL Site Mesonet wind observations was completed to better understand the frequent wind patterns and to exploit them in weather forecasting.

This analysis identified eight clusters or typical wind patterns which seem to be well correlated to commonly observed meteorological conditions.

Objectives

The main objective of this project is to formally identify recurring wind patterns and create a method to categorize wind fields according to these patterns. Three goals of the project once these clusters are defined are to:

  1. Understand how the wind fields evolve over time and obtain a better understanding of the physical processes driving the wind fields.
  2. Improve wind forecasting, both short and long term at the INL Site.
  3. Investigate whether the wind patterns are correlated with other factors such as precipitation coverage and wildfire frequency.

Accomplishments Through 2007

A cluster analysis was conducted using Mesonet data from November 1993 to February 1999 which identified eight wind field clusters. These were further refined by taking all data available from November 1993 to March 2006 and assigning them to clusters and refining the cluster centers. The clusters are numbered from 1 to 8, with 1 being the most common and 8 the least common. A number of statistics for each cluster were calculated including frequency of occurrence by season and time of day, average duration, times when each cluster was most likely to be observed, and the probability of transitions from one cluster to another. The evolution of the clusters has also been studied. A number of software tools have been developed that allow forecasters to examine current and historical wind fields in the context of which cluster they belong to and the expected changes in cluster membership over time.

Results

A detailed description of each cluster is beyond the scope of this report. As an example, Figure 9-11 shows the INL Site wind fields representing clusters 1 and 3. The left map shows the most frequent pattern at INL Site, namely a nocturnal drainage flow from the northeast. It is most common during summer nights and early mornings. The right map shows cluster 3, the third most frequent pattern representing moderate up-valley flow. It is often observed during summer afternoons. Overall, the first 5 clusters are more common and are related to normal diurnal trends due to terrain and atmospheric stability conditions. Clusters 6-8 occur less frequently and are associated with large-scale forcing from passing weather systems. People working at INL Site may be somewhat surprised that strong southwest winds are not the most common pattern. However, it must be remembered that most INL Site workers are at the site only during daylight hours, whereas the cluster analysis is based on data from all hours. Also, people usually remember extreme weather events more than the intervening quiescent periods. A more in-depth description of the cluster analysis is found in Clawson et al. (2007).

Plans for Continuation

We plan to continue improving our understanding of the physical processes, such as terrain effects, related to each wind cluster. We also plan to improve the cluster forecasting tools. This will allow us to incorporate the clusters into daily forecasting and also to work with the Pocatello National Weather Service in improving short term wind forecasting across SE Idaho and the INL Site.

Eventually we would like to look at whether the clusters are correlated with other spatial factors including precipitation and vegetation distributions and wildfi re probability.


Investigators and Affiliations

Roger Carter, Physical Scientist, NOAA Air Resources Laboratory Field Research Division, Idaho Falls, Idaho

Jason D. Rich and Neil Hukari, Research Meteorologists, NOAA Air Resources Laboratory Field Research Division, Idaho Falls, Idaho

Funding Sources

U.S. Department of Energy Idaho Operations Office

References

Clawson, K. L., R. M. Eckman, N. F. Hukari, J. D. Rich, N. R. Ricks, 2007: Climatography of the Idaho National Laboratory 3rd Edition, NOAA Technical Memorandum OAR ARL-259, Idaho Falls, Idaho, 249 pp.


Home | Background | Surveillance |  Land Management | Education | Research | Risk Assessment | Publications |  Links | Feedback | Opportunities