Soil moisture plays an important part in predicting different forest-related phenomena, such as tree growth or forest fire risk. As they influence the carbon storage capacity of boreal forest ecosystems, it is crucial to provide soil moisture information at high spatio-temporal scales. Current satellite-based soil moisture products often have high temporal resolution at the expense of spatial resolution. Therefore, we developed a machine-learning-based model to estimate soil moisture at high spatial resolution over boreal forested areas for the annual time period from May to October, while retaining the high temporal resolution. The basis data of the model is the 36 km spatial resolution soil moisture data from the Soil Moisture Active Passive (SMAP) mission. Additionally, vegetation properties, weather-related parameters, and measured in situ soil moisture data are used to guide the model construction process. The analysis of the developed model shows that it retains the temporal and large-scale spatial variability of SMAP soil moisture. Furthermore, comparisons with the independent in situ soil moisture data indicate that the model's predictions align more closely with in situ values than SMAP soil moisture, as RMSE decreases from 0.103 to 0.092 m3 m−3 , and correlation increases from 0.46 to 0.55 over forest sites. Therefore, this machine-learning-based model can be used to predict high-resolution soil moisture over boreal forested areas.

Small lakes are common across the Boreal-Arctic zone. Due to shallowness and high shoreline-surface area ratios, they are abundant in aquatic macrophytes. Vegetated littoral zones have been suggested to count as wetlands when quantifying carbon sinks and sources, but the actual magnitude of aquatic vegetation is seldom quantified. Remote sensing has potential in this quantification but has rarely been applied to macrophyte abundance and phenology in Boreal-Arctic lakes. We examined the abundance of emergent and floating-leaved vegetation in small, mainly oligotrophic lakes at the northern boreal-subarctic ecotone in Finland with remote sensing. We collected uncrewed aerial vehicle imagery to quantify aquatic macrophytes in five small lakes at total, growth-form, and genus levels in terms of 1) binary presence and 2) percentage coverage. We produced seasonal estimates across a 105 km² catchment by training stochastic gradient boosting models with Sentinel-1, Sentinel-2, and combined multi-sensor data. In a sequential classification-regression approach, the multi-sensor models identified 39-93 % of occurrences of aquatic vegetation and explained 26-93 % of variation in coverage of different target classes. Across 45 small lakes in the catchment, emergent and floating-leaved macrophytes were present in 9 % and 25 % of surface area, respectively, and both life forms covered 2 %. Total coverage by lake ranged 3-97 %. While denser Carex spp. and Nuphar lutea occurrences were well predicted, sparse and erect Equisetum fluviatile stands were challenging to capture. Dominant species varied across the lakes, although N. lutea was the most common. Multispectral Sentinel-2 data universally outperformed Sentinel-1 SAR data in estimating aquatic macrophytes, detecting coverages above ca. 1-5 %. Shoreline effects prevented accurate prediction in lakes with <0.25 ha areal extent. Based on our results, the seasonal mapping of aquatic macrophytes at the genus level with Sentinel data is feasible, though requires passive instrument data, which is only available intermittently due to cloud cover.

The deterioration of road conditions and increasing repair deficits pose challenges for the maintenance of reliable road infrastructure, and thus threaten, for example, safety and the fluent flow of traffic. Improved and more efficient procedures for maintenance are required, and these require improved knowledge of road conditions, i.e., improved data. Three-dimensional mapping presents possibilities for large-scale collection of data on road surfaces and automatic evaluation of maintenance needs. However, the development and, specifically, evaluation of large-scale mobile methods requires reliable references. To evaluate possibilities for close-range, static, high-resolution, three-dimensional measurement of road surfaces for reference use, three measurement methods and five instrumentations are investigated: terrestrial laser scanning (TLS, Leica RTC360), photogrammetry using high-resolution professional-grade cameras (Nikon D800 and D810E), photogrammetry using an industrial camera (FLIR Grasshopper GS3-U3-120S6C-C), and structured-light handheld scanners Artec Leo and Faro Freestyle. High-resolution photogrammetry is established as reference based on laboratory measurements and point density. The instrumentations are compared against one another using cross-sections, point–point distances, and ability to obtain key metrics of defects, and a qualitative assessment of the processing procedures for each is carried out. It is found that photogrammetric models provide the highest resolutions (10–50 million points per m²) and photogrammetric and TLS approaches perform robustly in precision with consistent sub-millimeter offsets relative to one another, while handheld scanners perform relatively inconsistently. A discussion on the practical implications of using each of the examined instrumentations is presented.

Context: The spectral variation hypothesis (SVH) suggests that remotely sensed spectral variation correlates with biodiversity. However, conflicting evidence has been presented regarding the SVH, it has not been widely tested in high-latitude landscapes, and its assessments mainly employ species richness, omitting other aspects of biodiversity. Objectives: We examine plant diversity, spectral variation and the SVH across a northern boreal landscape utilizing high-spatial-resolution satellite imagery. We incorporate taxonomic, phylogenetic, functional, and indicator-based plant diversity metrics and visualize spatial differences between proposed plant diversity proxies. Methods: We surveyed plant community composition in peatlands, forests, and tundra in northwest Finland and calculated 14 diversity metrics for 355 sites. We compared observed diversity with (1) spatial, (2) temporal, (3) multivariate spatiotemporal spectral variation, and (4) multivariate machine learning. We used statistical relationships and machine learning models to create continuous maps of plant diversity. Results: While forests were generally more biodiverse than peatlands or tundra, each of these ecosystems obtained the highest values in some metrics. Spectral variation had positive correlations (0.14–0.72) across most dimensions of plant diversity, but relationships varied between ecosystems and were sometimes insignificant or negative. For example, plant diversity in forests was largely uncorrelated with spatial spectral variation. Temporal spectral variation was more strongly correlated with plant diversity than spatial spectral variation in a majority (41 of 56) of cases. Conclusions: While spectral variation was usually correlated with plant diversity, correlations varied among peatlands, forests, and tundra. Thus, methods for estimating plant diversity should be tailored to different biodiversity metrics and ecosystem types.