Abstract. We present a simple method that allows snow depth measurements to be converted to snow water equivalent (SWE) estimates. These estimates are useful to individuals interested in water resources, ecological function, and avalanche forecasting. They can also be assimilated into models to help improve predictions of total water volumes over large regions. The conversion of depth to SWE is particularly valuable since snow depth measurements are far more numerous than costlier and more complex SWE measurements. Our model regresses SWE against snow depth (h), day of water year (DOY) and climatological (30-year normal) values for winter (December, January, February) precipitation (PPTWT), and the difference (TD) between mean temperature of the warmest month and mean temperature of the coldest month, producing a power-law relationship. Relying on climatological normals rather than weather data for a given year allows our model to be applied at measurement sites lacking a weather station. Separate equations are obtained for the accumulation and the ablation phases of the snowpack. The model is validated against a large database of snow pillow measurements and yields a bias in SWE of less than 2 mm and a root-mean-squared error (RMSE) in SWE of less than 60 mm. The model is additionally validated against two completely independent sets of data: one from western North America and one from the northeastern United States. Finally, the results are compared with three other models for bulk density that have varying degrees of complexity and that were built in multiple geographic regions. The results show that the model described in this paper has the best performance for the validation data sets.

Western United States snowpacks are generated by cold-season storms, yet the vast majority of snow trend studies utilize undifferentiated air temperature records. Previous studies do not distinguish between days with and without precipitation, which effectively dilutes temperature trends relevant for snowpack monitoring. We examined trends in cold-season precipitation and impacts on snow in nine mountain regions across the western United States. Using 33 years of daily meteorological data (1984-2016) from 567 Snow Telemetry (SNOTEL) sites and a homogenized daily temperature dataset (TopoWx), we investigated seasonal and regional trends in storm day temperatures, storm day frequency, and resulting snowpack fate. We found widespread warming of 0.4°C-1.2°C per decade, especially during the fall and in the Interior West using tests for statistically significant trends. Disaggregation showed that days with precipitation are warming nearly twice as fast as dry days in the fall and winter. We also observed increases in November dry days, increased melt on dry days throughout the accumulation season, spring cooling and declines in daily snowmelt during snowdepleting storm days. These findings demonstrate the importance of disaggregating temperature data to elucidate trends (in storm day frequency, accumulation and melt, and warming and cooling) and their impacts on snow in the western US.

Atmospheric rivers (ARs) are regions of high water vapor transport in the lower atmosphere. When these air masses encounter mountain ranges, they can produce significantly enhanced orographic precipitation. AR events may substantially influence seasonal mountain snow, an important ecologic and economic resource for Washington, Oregon, and California. To better understand how ARs affect the montane snowpack of these U.S. West Coast states, we used 33 years of Snow Telemetry and Topography Weather data to examine AR and non‐AR storm day temperatures and impacts on snow water equivalent (SWE). We found mean daily minimum temperatures of AR storm days to be 1 to 4 °C warmer and mean daily increases in SWE to be 23% to 57% higher than non‐AR storm days. AR storm days have contributed an average of 23% of effective snowfall to seasonal SWE in the California Sierra Nevada and 34% of effective snowfall to seasonal SWE in the Washington and Oregon Cascades.

Road extraction is important for road network renewal, intelligent transportation systems and smart cities. This paper proposes an effective method to improve road extraction accuracy and reconstruct the broken road lines caused by ground occlusion. Firstly, an attention mechanism-based convolution neural network is established to enhance feature extraction capability. By highlighting key areas and restraining interference features, the road extraction accuracy is improved. Secondly, for the common broken road problem in the extraction results, a heuristic method based on connected domain analysis is proposed to reconstruct the road. An experiment is carried out on a benchmark dataset to prove the effectiveness of this method, and the result is compared with that of several famous deep learning models including FCN8s, SegNet, U-Net and D-Linknet. The comparison shows that this model increases the IOU value and the F1 score by 3.35%-12.8% and 2.41-9.8, respectively. Additionally, the result proves the proposed method is effective at extracting roads from occluded areas.