Continuous data on spatial and temporal patterns of snowmelt rates are essential for hydrological studies, but are commonly not available, especially in the subarctic, mainly due to high monitoring costs. In this study, temperature loggers were used to measure local and microscale variations in snowpack temperature, in order to understand snowmelt processes and rates in subarctic northern Finland. The loggers were deployed on six test plots along a hillslope with varying topography (elevation and aspect) and vegetation (forest, transitional zone and mires, i.e. treeless peatlands) during two consecutive winters (2014 and 2015). At each test plot, the sensors were installed in five locations, at two heights in a snow profile. Algorithms were developed to analyse the snowmelt rates from high-resolution snowpack temperature data. The validity of the results was evaluated using snow depth and soil moisture data from adjacent reference sensors and the results were tested using an empirical degree-day snow model calibrated for each test plot. Snowmelt rates were relatively similar in mires (median 2.3 mm d−1 °C−1) and forests (median 2.6 mm d−1 °C−1) with apparent inter-annual variation. The observed melt rates were highest in the highest elevation plots, in transition zone in 2014 (median 4.6 mm d−1 °C−1) and southwest-facing forest line in 2015 (median 3.2 mm d−1 °C−1). The timing of the modelled meltwater outflow and snowpack ablation showed good agreement with the snowpack temperature-derived estimates and the soil moisture and snow depth measurements. The simple approach used represents a novel and cost-effective method to improve the spatial accuracy of in situ snow cover ablation measurements and melt rates and the precision of snowmelt models in the subarctic. An open-access R-based model is provided with this paper for analysis of high-frequency snow temperature data.