S193 An Inter-Model Comparison of Gridded Meteorological Datasets in Montana

Sunday, 6 January 2019
Hall 4 (Phoenix Convention Center - West and North Buildings)
Colin Brust, University of Montana, Missoula, MT; and K. Jencso, M. Sweet, and K. Bocinsky

Farmers, land managers, and scientists rely on gridded meteorological data sets (GMDs) to aid their decision-making processes. The fine spatial and temporal resolution of these data sets makes them convenient sources of temperature or precipitation data when weather station data are not available. However, differences in input weather station data, interpolation methods, and topoclimatic assumptions can lead to notable differences between data sets across a study area. In addition to these differences, the vast number of publicly available GMDs can make it difficult for stakeholders to select a data set that is accurate and reliable for their area of interest. Past comparisons of GMDs have generally had two main limitations:

1) They only compare data sets at pixels that are co-located with weather stations, limiting the area of study.

2) They resample and reproject data sets onto a common grid, distorting the original interpolations and introducing uncertainty into the comparisons.

To work around these limitations, we developed a new framework for comparing GMDs. We used this framework to compare maximum temperature (tmax), minimum temperature (tmin), and precipitation (ppt) interpolations from five GMDs across the state of Montana and then used unbiased Montana Mesonet weather station data to compare each GMD to measured ground data. Our findings show that at large temporal (30 years) and spatial (state of Montana) scales, all five data sets produce similar results for tmax, tmin, and ppt. At smaller temporal and spatial scales and at higher elevations, however, the predictions of data sets begin to diverge. Most notably, we found that interpolations of ppt between data sets begin to diverge at 1,500 meters and that daily interpolations of tmin and tmax can be biased by as much as 5°C relative to Montana Mesonet measurements in agricultural settings. Our results show that failing to acknowledge the differences between GMDs and physical weather measurements could significantly impact the final decisions and planning efforts of GMD users.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner