Frozen ground, a key indicator of climate change, profoundly influences ecological, hydrological, and carbon flux processes in cold regions. However, traditional monitoring methods, which rely on a binary 0 °C soil temperature threshold, fail to capture the complexities of soil freezing, such as freezing point depression and transitional states where water and ice coexist. This study introduces a framework that fits a theoretical Soil Freezing Characteristic Curve (SFCC) in permittivity–temperature space to site- and cycle-specific in situ measurements. This approach enables the quantification of the degree of soil freezing and the classification of soil states as frozen, unfrozen, or in transition (partially frozen). We analyzed 135 freezing cycles from 87 sites, each equipped with permittivity-based soil moisture probes. These sites are part of eight monitoring networks spanning diverse Canadian landscapes, including eastern boreal forests (Montmorency Forest, La Romaine, James Bay, Chapleau), western boreal forests (Candle Lake), prairies (Kenaston), and tundra regions (Trail Valley Creek and George River). On average, eastern boreal forest sites exhibited prolonged unfrozen and transitional states due to high soil moisture retention and insulation from snow and vegetation cover (23 frozen days, 46 transitional days). In contrast, western boreal forest sites experienced more extensive freezing under drier conditions (73 frozen days, 76 transitional days). Prairie sites displayed equal durations of frozen and transitional states (71 days each), while tundra sites had the longest frozen periods (145 frozen days, 52 transitional days). Notably, transitional periods lasted as long as – or even longer than – frozen ones, underscoring the limitations of binary classifications. Furthermore, the traditional 0 °C threshold misclassified transitional soil states, overestimating frozen days by over 87 % in prairie and western boreal regions, and unfrozen days by 86 % in the eastern boreal forest. In tundra, the bias was more balanced, with 64 % and 36 % of transitional days misclassified as unfrozen and frozen, respectively. This SFCC-based framework enhances seasonally frozen ground monitoring, offering deeper insights into soil freeze-thaw dynamics. These advancements have implications for improving climate change assessments, refining carbon flux models, and training and validating remote sensing products. Additionally, the resulting database of soil states from this study provides a valuable resource for advancing frozen ground research, particularly in remote sensing and ecosystem modeling efforts.