The devil is in the details, especially for remote sensing where the scale of the imagery does not always match the scale of interpretation. Iceland, a subarctic volcanic island in the north Atlantic, has a long history of dynamic vegetation changes. However, the scale at which these changes occur may be smaller than the resolution of datasets most commonly used to study arctic vegetation trends. In this study, I used MODIS Aqua 250m satellite imagery to evaluate nationwide trends in Normalized Difference Vegetation Index (NDVI) over the past twenty years. Then, for a handful of sites r epresenting land cover classifications of interest, I used 30-meter Harmonized Landsat and Sentinel-2 (HLS) imagery to quantify information loss between the 30m trend map and several synthetically upscaled trend maps. Over Iceland as a whole, the trend in NDVI was found to be minimal. However, the higher resolution imagery revealed dynamic trends which are lost at lower resolutions. The greatest information loss occurred in highly heterogeneous land cover classes, with a maximum information loss of 63% from 30m to 240m and 90% from 30m to 1000m. Additionally, I found that the way in which land cover classifications are defined has the potential to impact interpretation. Neither of these complexities have been addressed in previous studies of Icelandic vegetation trends and have rarely been addressed in studies of Arctic vegetation trends as a whole. My findings support the need for greater consideration of ecosystem classification and scale dependencies in any work aimed at using sa tellite remote sensing to better understand Arctic greening.