This study details the creation of a gridded snowfall dataset for North America, with focus on the quality of the interpolated product. Daily snowfall amounts from National Weather Service Cooperative Observer Program stations and Meteorological Service of Canada surface stations are interpolated to 1° by 1° grids from 1900 to 2009 and examined for data record length and quality. The interpolation is validated spatially and temporally through the use of stratified sampling and k-fold cross-validation analyses. Interpolation errors average around 0.5 cm and range from less than 0.01 to greater than 2.5 cm. For most locations, this is within the measurement sensitivity. Grid cells with large variations in elevation experience higher errors and should be used with caution. A new gridded snowfall climatology is presented based on in situ observations that capture seasonal and interannual variability in monthly snowfall over most of the North American land area from 1949 to 2009. The Community Collaborative Rain, Hail and Snow Network is used as an independent set of point data that is compared to the gridded product. Errors are mainly in the form of the gridded data underestimating snowfall compared to the point data. The spatial extent, temporal length, and resolution of the dataset are unprecedented with regard to observational snowfall products and will present new opportunities for examining snowfall across North America.

The current network of weather surveillance radars within the United States readily detects flying birds and has proven to be a useful remote-sensing tool for ornithological study. Radar reflectivity measures serve as an index to bird density and have been used to quantitatively map landbird distributions during migratory stopover by sampling birds aloft at the onset of nocturnal migratory flights. Our objective was to further develop and validate a similar approach for mapping wintering waterfowl distributions using weather surveillance radar observations at the onset of evening flights. We evaluated data from the Sacramento, CA radar (KDAX) during winters 1998–1999 and 1999–2000. We determined an optimal sampling time by evaluating the accuracy and precision of radar observations at different times during the onset of evening flight relative to observed diurnal distributions of radio-marked birds on the ground. The mean time of evening flight initiation occurred 23 min after sunset with the strongest correlations between reflectivity and waterfowl density on the ground occurring almost immediately after flight initiation. Radar measures became more spatially homogeneous as evening flight progressed because birds dispersed from their departure locations. Radars effectively detected birds to a mean maximum range of 83 km during the first 20 min of evening flight. Using a sun elevation angle of −5° (28 min after sunset) as our optimal sampling time, we validated our approach using KDAX data and additional data from the Beale Air Force Base, CA (KBBX) radar during winter 1998–1999. Bias-adjusted radar reflectivity of waterfowl aloft was positively related to the observed diurnal density of radio-marked waterfowl locations on the ground. Thus, weather radars provide accurate measures of relative wintering waterfowl density that can be used to comprehensively map their distributions over large spatial extents.

This study examines the regional variations in the frequency of snowfall across the conterminous United States from 1930 to 2007. Principal components analysis and cluster analysis are used to group stations together based on the main modes of variation in snowfall frequency. Results indicate the existence of seven unique snowfall regions, which correspond to predominant storm tracks across the United States. These are the southeast, the south central Plains and southwest, the Ohio River Valley and mid-Atlantic, the Pacific Northwest, and three sub-regions in the Upper-Midwest. Quantile regression reveals that the distribution functions of each region's snowfall frequency are different and in some regions, changing over time. The northern part of the Upper-Midwest is experiencing increasing trends in all percentiles of snowfall frequencies, the Pacific northwest is experiencing declines in greater than median snowfall frequencies, and the southeast is seeing a decline in extreme frequency years. Correlation analysis between large-scale teleconnection patterns and regionally averaged snowfall frequencies corroborate previous findings and indicate specific forcing mechanisms for snowfall frequency in each region.

On the basis of snowfall observations from 1929 to 1999, positive (negative) snowfall anomalies are associated with wetter (drier) than normal conditions during the summer [July-August (JJA)] in the northern Great Plains. The five driest summers are associated with negative snowfall anomalies during the preceding winter (266.7 mm) and spring (262.4 mm) that cover most of the study region (;85%). Snowfall anomalies during the late spring (April-May) are more important for determining summer moisture conditions than snowfall anomalies in fall [September-November (SON)] or winter [December-February (DJF)]. The link between snowfall anomalies and summer moisture conditions appears to be, at least partly, through soil moisture since positive (negative) snowfall anomalies are associated with wetter (drier) soils, a later (earlier) date of snowmelt, cooler (warmer) air temperatures, and more (less) evaporation during spring and summer. However, the relationship between spring snowfall and summer moisture conditions is only statistically significant when the moisture anomaly index (Z), which accounts for both temperature and precipitation, is used to characterize summer moisture conditions and the signal is weak when just considering precipitation (e.g., standardized precipitation index). Results also indicate that the strength of the relationship between winter/spring snowfall and summer moisture varies significantly over space and time, which limits its utility for seasonal forecasting.

This study seeks to determine the skill of multiple discriminant analysis for predicting seasonal snowfall. Winter total snowfall amount and frequency of snowfall events are examined for 440 stations in the United States from 1930 to 2006. The independent variables used to create the forecast include ocean-atmosphere teleconnection patterns [such as the Pacific Decadal Oscillation (PDO) and El Niño Southern Oscillation (ENSO)], large-scale atmospheric patterns [such as the Arctic Oscillation (AO), North Atlantic Oscillation (NAO) and Pacific/North American (PNA)], land cover (such as Arctic sea ice extent and Eurasian snow cover extent), and temperature. Based on a jackknife analysis, forecasts are correct 20-80% of the time for categories of 'below normal', 'near normal', and 'above normal'. When broader categories are used of 'normal or below', 'near normal', and 'normal or above' the forecasts are correct as much as 90% of the time at some stations. The Central United States, Ohio River Valley, Great Lakes, and Upper Midwest regions show the highest level of skill. Results not only confirm relationships previously documented between atmospheric phenomena and US snowfall (such as with the PNA, NAO, and ENSO), but also expand our understanding of factors that influence decadal-scale snowfall variation (such as Arctic sea ice extent and Eurasian snow cover extent).