Glacier evolution models rely on mass balance observations to reproduce past changes and project future trends, but observational biases undermine accuracy. We investigate biases in snow cover minima observations, often used to constrain glacier mass balance, due to late-summer (~September) snapshot observations. We create weekly snow cover time series for 200 glaciers in western North America using an automated workflow applied to Landsat and Sentinel-2 images (2013–2023). Snow minima occurred up to three months earlier at higher-latitude sites (~August) than lower-latitude sites (~October). September-based observations overestimated AARs by up to 0.5, emphasizing the need to observe the full melt season across diverse climates. Median accumulation area ratios (AARs) indicate most glaciers are unhealthy (AAR<0.5). Comparisons with a global glacier model revealed overestimated model snowline altitudes, potentially biasing projections. Our results underscore the importance of capturing temporal variability in snow minima to improve glacier mass balance estimates and future projections.