Forage supply of savanna grasslands plays a crucial role for local food security and consequently, a reliable monitoring system could help to better manage vital forage resources. To help installing such a monitoring system, we investigated whether in-situ hyperspectral data could be resampled to match the spectral resolution of multi- and hyperspectral satellites; if the type of sensor affected model transfer; and if spatio-temporal patterns of forage characteristics could be related to environmental drivers. We established models for forage quantity (green biomass) and five forage quality proxies (metabolisable energy, acid/neutral detergent fibre, ash, phosphorus). Hyperspectral resolution of the Hyperion satellite mostly resulted in higher accuracies (i.e. higher R-2, lower RMSE). When applied to satellite data, though, the greater quality of the multispectral Sentinel-2 satellite data leads to more realistic forage maps. By analysing a three-year time series, we found plant phenology and cumulated precipitation to be the most important environmental drivers of forage supply. We conclude that none of the investigated satellites provide optimal conditions for monitoring purposes. Future hyperspectral satellite missions like EnMAP, combining the high information level of Hyperion with the good data quality and resolution of Sentinel-2, will provide the prerequisites for installing a regular monitoring service.

A marine sediment record from the central Bering Sea, spanning the last 20 thousand years (ka), was studied to unravel the depositional history with regard to terrigenous sediment supply and biogenic sedimentation. Methodic approaches comprised the inference of accumulation rates of siliciclastic and biogenic components, grain-size analysis, and (clay) mineralogy, as well as paleoclimatic modelling. Changes in the depositional history provides insight into land-ocean linkages of paleoenvironmental changes. During the finale of the Last Glacial Maximum, the depositional environment was characterized by hemipelagic background sedimentation. A marked change in the terrigenous sediment provenance during the late Heinrich 1 Stadial (15.7-14.5 ka), indicated by increases in kaolinite and a high glaciofluvial influx of clay, gives evidence of the deglaciation of the Brooks Range in the hinterland of Alaska. This meltwater pulse also stimulated the postglacial onset of biological productivity. Glacial melt implies regional climate warming during a time of widespread cooling on the northern hemisphere. Our simulation experiment with a coupled climate model suggests atmospheric teleconnections to the North Atlantic, with impacts on the dynamics of the Aleutian Low system that gave rise to warmer winters and an early onset of spring during that time. The late deglacial period between 14.5 and 11.0 ka was characterized by enhanced fluvial runoff and biological productivity in the course of climate amelioration, sea-level rise, seasonal sea-ice retreat, and permafrost thaw in the hinterland. The latter processes temporarily stalled during the Younger Dryas stadial (12.9-11.7 ka) and commenced again during the Preboreal (earliest Holocene), after 11.7 ka. High river runoff might have fertilized the Bering Sea and contributed to enhanced upper ocean stratification. Since 11.0 ka, advanced transgression has shifted the coast line and fluvial influence of the Yukon River away from the study site. The opening of the Bering Strait strengthened contour currents along the continental slope, leaving behind winnowed sand-rich sediments through the early to mid-Holocene, with non-deposition occurring since about 6.0 ka.

With the present study, we introduce a fast and robust method to calculate the source displacement spectra of small earthquakes on a local to regional scale. The work is based on the publicly available Qopen method of full envelope inversion, which is further tuned for the given purpose. Important source parameters-seismic moment, moment magnitude, corner frequency, and high-frequency fall off-are determined from the source spectra by fitting a simple earthquake source model. The method is demonstrated by means of a data set comprising the 2018 West Bohemia earthquake swarm. We report moment magnitudes, corner frequencies, and centroid moment tensors inverted from short-period body waves with the Grond package for all earthquakes with a local magnitude larger than 1.8. Moment magnitudes calculated by envelope inversion show a very good agreement to moment magnitudes resulting from the probabilisitc moment tensor inversion. Furthermore, source displacement spectra from envelope inversion show a good agreement with spectra obtained by multiple taper analysis of the direct onsets of body waves but are not affected by the large scatter of the second. The seismic moments obtained with the envelope inversion scale with corner frequencies according to M-0 proportional to f(c)(-4.7). Earthquakes of the present data set result in a smaller stress drop for smaller magnitudes. Self-similarity of earthquake rupture is not observed. In addition, we report frequency-dependent site amplification at the used stations.

Geochemical homogeneity in shale is often assumed when tracing subsurface fluids and characterizing sedimentary basins. This study presents measurements of the bulk gas composition, stable isotopes, and noble gas volume fraction and isotopes for shale gas samples collected from gas wells in the Wufeng-Longmaxi Shale, the southern Sichuan Basin, China. The dryness [C-1 /(C-2 + C-3)] ranging from 166.3 to 251.2, combined with delta C-13(1) and delta DC1 that vary from -28.8 to -27.3 parts per thousand and - 153 to -145 parts per thousand, respectively, point to a late mature thermogenic origin of hydrocarbon gas. He-3/He-4 ratios of gas samples are around 0.01 times the air value suggesting dominantly crust-derived He. Ne-21/Ne-22 and Ar-40/Ar-36 ratios of many gas samples are higher than the corresponding air values indicating the mixing of crustal and atmospheric noble gases. Multiple dichotomous patterns are observed in noble gas signatures of forelimb and backlimb samples, and depression and crest samples. Ne-20/Ne-22 ratios of some crest samples are higher than that of depression samples in the backlimb, pointing to the presence of diffusion-driven fractionation that is likely caused by the long-distance migration from depression to crest. Elemental ratios of air-derived noble gas isotopes - Ne-22/Ar-36, Kr-84/Ar-36, and Xe-132/Ar-36 are compared to the recharge water values, suggesting the interactions of oil, gas, and water phases in the shale over geologic time. Forelimb samples generally display older ages than backlimb samples, indicating a larger flux of external radiogenic He-4 due to the higher density of deep faults in the forelimb area caused by the basementinvolved deformation. The basement-involved deformation also causes pore collapse especially in the forelimb leading to a lower porosity that results in a more pristine noble gas signature in the forelimb due to the reduced impact of younger recharge water.

Landslide hazard models aim at mitigating landslide impact by providing probabilistic forecasting, and the accuracy of these models hinges on landslide databases for model training and testing. Landslide databases at times lack information on the underlying triggering mechanism, making these inventories almost unusable in hazard models. We developed a Python-based unique library, Landsifier, that contains three different machine-Learning frameworks for assessing the likely triggering mechanisms of individual landslides or entire inventories based on landslide geometry. Two of these methods only use the 2D landslide planforms, and the third utilizes the 3D shape of landslides relying on an underlying digital elevation model (DEM). The base method extracts geometric properties of landslide polygons as a feature space for the shallow learner - random forest (RF). An alternative method relies on landslide planform images as an input for the deep learning algorithm - convolutional neural network (CNN). The last framework extracts topological properties of 3D landslides through topological data analysis (TDA) and then feeds these properties as a feature space to the random forest classifier. We tested all three interchangeable methods on several inventories with known triggers spread over the Japanese archipelago. To demonstrate the effectiveness of developed methods, we used two testing configurations. The first configuration merges all the available data for the k-fold cross-validation, whereas the second configuration excludes one inventory during the training phase to use as the sole testing inventory. Our geometric-feature-based method performs satisfactorily, with classification accuracies varying between 67 % and 92 %. We have introduced a more straightforward but data-intensive CNN alternative, as it inputs only landslide images without manual feature selection. CNN eases the scripting process without losing classification accuracy. Using topological features from 3D landslides (extracted through TDA) in the RF classifier improves classification accuracy by 12 % on average. TDA also requires less training data. However, the landscape autocorrelation could easily bias TDA-based classification. Finally, we implemented the three methods on an inventory without any triggering information to showcase a real-world application.