Maximizing Interactive Decision Support Services for Road Maintenance and Visitors at Yosemite National Park by Identifying Biases in HRRR Snow Level Predictions

The number of tourists visiting Yosemite National Park (YNP) each year has been increasing, and YNP road maintenance has sought ways to be more proactive on snow removal and road closures, to prevent accidents. Messaging the information required to meet this need can be complicated, since a higher peak in the park may require closure while a lower elevation could remain open, depending on the snow level. The Physical Sciences Division (PSD) of NOAA has strategically placed snow level radars around the Sierra Nevada to measure the height of the atmosphere at which snow turns to rain. Using the data from these radars, snow level verification was completed comparing actual snow levels to the High-Resolution Rapid Refresh (HRRR) model at the time steps of initial, 3-hrs, 6-hrs, 12-hrs and 18-hrs.

Forecasters at NWS Hanford have anecdotally observed that the HRRR would place snow levels too high compared to actual events. Our analysis of the verification data provided by Allen White and Dan Gottas from PSD has proven this assumption true, while also providing crucial details on the accuracy of the different time steps from the HRRR. This will help allow our snow level forecasts to move from deterministic to probabilistic, which YNP can then use for snow removal preparation. Given the means and standard deviations of this data from the two sites, a range of the snow level may now be predicted with improved skill.

Two locations were assessed, New Exchequer Dam, 40 miles to the west of Yosemite, and Pine Flat Dam, 60 miles south of Yosemite, for snow level verification above 4000 ft and 9000 ft. Yosemite Valley, the most visited site of the park, sits at about 4000 ft while Tioga Pass sits above 9000 ft. Tioga Pass is of extreme importance, since it is one of the most scenic roads over the Sierra Nevada and provides international travelers access to tourist destinations on the East side of the mountain range. The seasonal opening and closing dates are publicized across the state to help groups ensure unaffected travel. Safe travel along this road is a major concern for YNP and its visitors.

Our study from 2016-2019 for both locations showed overall that while the HRRR had a warm bias, snow level projections above 9000 ft at Tioga Pass exhibited an unexpected cold bias. The HRRR had a snow level mean error of -1460 ft for forecasts above 9000 ft for the initialization time step at Pine Flat Dam. The New Exchequer Dam location produced a cold bias of -458 ft at the same elevation. These inconsistencies in the HRRR result in the longevity of road closures within the park and impact the ability of transportation managers to plan appropriate times and funds for utilizing snow removal equipment. The research also helped clarify biases from the HRRR at different elevations, while also showcasing trends in the skill at different time steps. Finally, a line of best fit with exponential regression was developed which can allow NWS Hanford forecasters to adjust HRRR snow level forecasts. Based upon this work, Interactive Decision Support Services (IDSS) can be enhanced for this core partner by providing graphical depictions of snow level, which have found to be more palatable for roads maintenance crews at YNP.