**1. Introduction**

Climate change over the Northern Hemisphere high latitudes and boreal forest zone has affected the snow and vegetation cover and, thus, the surface albedo [1–12]. Previous studies have shown that the duration and timing of the snow melt season has changed differently in different areas [13–17]. The snow cover extent has decreased especially significantly in the spring [3,5,7], and the surface albedo during the melt season months has also decreased, largely due to a decline in the area covered by snow [10]. These changes affect the local and global energy budgets [6]. The changing climate also changes the vegetation. The increased size of vegetation decreases wintertime albedo by covering the land surface and casting larger shadows on snow-covered surfaces. This is particularly relevant in the late winter. Changes in vegetation affect the local climate and the scattering properties of the forest (with larger shadows and increased multiple scattering due to increased forest height or density) [11]. The changes in the vegetation are also linked to changes in the spatial coverage of permafrost.

Changes in the timing and duration of the melt season, as well as in the surface albedo, are important parameters for climate models [18,19]. They can be used as comparison data for model parameters during the run. More accurate information on these parameters is needed to improve the models. In particular, the albedo of vegetated surfaces overlain by snow, such as boreal forests, introduce uncertainty to the model outputs [20–22].

Changes in the albedo of snow-covered surfaces can be caused by changes in a number of variables, such as climate, impurities in snow, vegetation, permafrost and changes in the properties of the snow surface [23–27]. Some of these factors also interact with each other. For example, changes in air temperature, precipitation, and wind speed affect both the snow cover extent and snow surface properties, which in turn change the snow surface albedo [10,24,27]. The optical scattering properties of a snow surface are most heavily determined by grain size and shape [28], with grain size being the main physical factor responsible for snow-albedo variations [24]. These, together with climatic factors such as wind, air temperature, and the existence of vegetation affect snow surface roughness, which also affects the brightness of the surface [29,30].

The vast area covered by seasonal snow provides a reason to study it at global scale. This can be characterized using satellite remote sensing. Snow cover and surface albedo have been analyzed from satellite data for decades [31–33]. The first studies covered small areas and short periods of time, and the available data was limited in coverage and resolution. These studies form the basis for the new satellite-derived data records, which offer better spatial and temporal coverage and, thus, enable climatic studies of various key parameters. The timing of the melt season has typically been estimated using passive microwave satellite data [14,15], which are sensitive to the presence and amount of liquid water. Therefore, microwave data are good at detecting changes in the snow moisture content, but do not react to thin snow covers, which affect the surface albedo significantly, but have very low liquid water content. Moreover, microwave data cannot easily differentiate between wet snow and wet ground. Surface albedo data have also been used to determine the timing of melt season [34], but the spatial and temporal coverage of these studies has been limited. There are many different definitions for the start and end of melt [35]. One way to define these dates is to use the time when the open areas are less than half covered with snow [35]. The choice of definition depends on the intended application of the data.

Previous studies on snow cover have shown a decrease in the area covered by snow, as in the melt season albedo [2,3,5,10], but whether or not climate change has caused the albedo of the snow surface to change prior to the onset of melt has so far been unclear. Studies of changes in snow season surface albedo have typically been based on either specific calendar months assumed to represent the melt season [10], or on the maximum albedo of the snow season [36]. This paper presents a study of the changes in the surface albedo of the land areas of the Northern Hemisphere between latitudes 40◦N and 80◦N prior to the melt season. This has an effect on the global energy budget, as well as the length of the melt season and the surface albedo during the melt season. The study also investigates the changes in the timing of the melt season. The analysis utilizes 5-day mean surface albedo data, derived from optical satellite data for the years 1982 to 2015, to determine the start and end dates of the melt season and the corresponding surface albedo levels. The trends of these parameters over the 34-year period and their relationships to land use and trends in climatic parameters are also investigated.
