Search

Landslide thickness is a key parameter in various types of models used to simulate landslide susceptibility. In this study, we developed a model providing improved predictions of potential shallow landslide thickness in Switzerland. We tested three machine learning models based on random forests, generalized additive model, and linear regression and compared the results to three existing models that link soil thickness to slope and elevation. The models were calibrated using data from two field inventories in Switzerland ('HMDB' with 709 records and 'KtBE' with 517 records). We explored 37 different covariates including metrics on terrain, geomorphology, vegetation height, and lithology at three different cell sizes. To train the machine learning models, 21 variables were chosen based on the variable importance derived from random forest models and expert judgement. Our results show that the machine learning models consistently outperformed the existing models by reducing the mean absolute error by at least 17 %. The random forests models produced a mean absolute error of 0.25 m for the HMDB and 0.19 m for the KtBE data. Models based on machine learning substantially improve the prediction of landslide thickness, offering refined input for enhancing the performance of slope stability simulations. [ABSTRACT FROM AUTHOR]

As peatlands are a major terrestrial sink in the global carbon cycle, gaining an understanding of their development and changes throughout time is essential in order to predict their future carbon budget and potentially mitigate the adverse outcomes of climate change. With this aim to understand peat development, many studies have investigated the paleoecological dynamics by analyzing various proxies, including pollen, macrofossil, elemental, and biomarker analyses. However, as each of these proxies is known to have its own benefits and limitations, examining them in parallel allows for a deeper understanding of these paleoecological dynamics at the peatland and a systematic comparison of the power of these individual proxies. In this study, we therefore analyzed peat cores from a peatland in Germany (Beerberg, Thuringia) to (a) characterize the vegetation dynamics over the course of the peatland development during the late Holocene and (b) evaluate to what extent the inclusion of multiple proxies, specifically pollen, plant macrofossils, and biomarkers, contributes to a deeper understanding of those dynamics and interaction among factors. We found that, despite a major shift in the regional forest composition from primarily beech to spruce as well as many indicators of human impact in the region, the local plant population in the Beerberg area remained stable over time following the initial phase of peatland development up until the last couple of centuries. Therefore, little variation could be derived from the paleobotanical data alone. The combination of pollen and macrofossil analyses with the elemental and biomarker analyses enabled further understanding of the site development as these proxies added valuable additional information, including the occurrence of climatic variations, such as the Little Ice Age, and more recent disturbances, such as drainage. [ABSTRACT FROM AUTHOR]

As peatlands are a major terrestrial sink in the global carbon cycle, gaining understanding of their development and changes throughout time is essential to predict their future carbon budget and potentially mitigate negative influences of climate change. With this aim to understand peat development, many studies have investigated the paleoecological dynamics through the analysis of various proxies, including pollen, macrofossil, elemental, and biomarker analyses. However, as each of these proxies are known to have their own benefits and limitations, examining them in parallel potentially allows for a deeper understanding of these paleoecological dynamics at the peatland and for a systematic comparison of the power of these individual proxies. In this study, we therefore analyzed soil cores from a peatland in Germany (Beerberg, Thuringia) to a) characterize the vegetation dynamics over the course of the peatland development during the late Holocene and b) evaluate to what extent the inclusion of multiple proxies, specifically pollen, macrofossil, and biomarkers, contributes to a deeper understanding of those dynamics and interaction among factors. We found that, despite a major shift in regional forest composition from primarily beech to spruce as well as many indicators of human impact in the region, the local plant population in the Beerberg area remained stable over time following the initial phase of peatland development up until the last couple of centuries. Therefore, little variation could be derived from the paleobotanical data alone. The combination of pollen and macrofossil analyses with the elemental and biomarker analyses enabled further understanding of the site development as these proxies added valuable additional information including the occurrence of climatic variations, such as the Little Ice Age, and more recent disturbances such as drainage and dust deposition. [ABSTRACT FROM AUTHOR]

Despite the importance of soil organic matter (SOM) in the global carbon cycle, there remain many open questions regarding its formation and preservation. The study of individual organic compound classes that make up SOM, such as lipid biomarkers including n -alkanes, can provide insight into the cycling of bulk SOM. While studies of lipid biomarkers, particularly n -alkanes, have increased in number in the past few decades, only a limited number have focused on the transformation of these compounds following deposition in soil archives. We performed a systematic review to consolidate the available information on plant-derived n -alkanes and their transformation from plant to soil. Our major findings were (1) a nearly ubiquitous trend of decreased total concentration of n -alkanes either with time in litterbag experiments or with depth in open plant–soil systems and (2) preferential degradation of odd-chain length and shorter chain length n -alkanes represented by a decrease in either carbon preference index (CPI) or odd-over-even predominance (OEP) with depth, indicating degradation of the n -alkane signal or a shift in vegetation composition over time. The review also highlighted a lack of data transparency and standardization across studies of lipid biomarkers, making analysis and synthesis of published data time-consuming and difficult. We recommend that the community move towards more uniform and systematic reporting of biomarker data. Furthermore, as the number of studies examining the complete leaf–litter–soil continuum is very limited as well as unevenly distributed over geographical regions, climate zones, and soil types, future data collection should focus on underrepresented areas as well as quantifying the transformation of n -alkanes through the complete continuum from plant to soil. [ABSTRACT FROM AUTHOR]

As a key component of the carbon cycle, soil CO2 efflux (SCE) is being increasingly studied to improve our mechanistic understanding of this important carbon flux. Predicting ecosystem responses to climate change often depends on extrapolation of current relationships between ecosystem processes and their climatic drivers to conditions not yet experienced by the ecosystem. This raises the question of to what extent these relationships remain unaltered beyond the current climatic window for which observations are available to constrain the relationships. Here, we evaluate whether current responses of SCE to fluctuations in soil temperature and soil water content can be used to predict SCE under altered rainfall patterns. Of the 58 experiments for which we gathered SCE data, 20 were discarded because either too few data were available or inconsistencies precluded their incorporation in the analyses. The 38 remaining experiments were used to test the hypothesis that a model parameterized with data from the control plots (using soil temperature and water content as predictor variables) could adequately predict SCE measured in the manipulated treatment. Only for 7 of these 38 experiments was this hypothesis rejected. Importantly, these were the experiments with the most reliable data sets, i.e., those providing high-frequency measurements of SCE. Regression tree analysis demonstrated that our hypothesis could be rejected only for experiments with measurement intervals of less than 11 days, and was not rejected for any of the 24 experiments with larger measurement intervals. This highlights the importance of high-frequency measurements when studying effects of altered precipitation on SCE, probably because infrequent measurement schemes have insufficient capacity to detect shifts in the climate dependencies of SCE. Hence, the most justified answer to the question of whether current moisture responses of SCE can be extrapolated to predict SCE under altered precipitation regimes is "no" - as based on the most reliable data sets available. We strongly recommend that future experiments focus more strongly on establishing response functions across a broader range of precipitation regimes and soil moisture conditions. Such experiments should make accurate measurements of water availability, should conduct high-frequency SCE measurements, and should consider both instantaneous responses and the potential legacy effects of climate extremes. This is important, because with the novel approach presented here, we demonstrated that, at least for some ecosystems, current moisture responses could not be extrapolated to predict SCE under altered rainfall conditions. [ABSTRACT FROM AUTHOR]

Soil respiration studies are increasingly undertaken with the aim of quantifying C fluxes and predicting changes for the future. The interpretation of field data into annual C loss predictions requires the use of modeling tools which generally include model variables related to the underlying drivers of soil respiration, such as soil temperature, soil moisture and plant activity. Very few studies have reported using model selection procedures in which structurally different models are calibrated, then validated on separate observation datasets and the outcomes critically compared. This study utilized thorough model selection procedures to determine soil heterotrophic (microbial) and autotrophic (root) respiration for a heathland chronosequence. The model validation process identified that none of the six measured plant variables explained any data variation when included in models with soil temperature, which contradicts many current studies. The best predictive model used a generalized linear mixed effect model format with soil temperature as the only variable. There were no heterotrophic respiration differences between the community ages. In contrast, autotrophic respiration was significantly greater on the youngest vegetation (55% of total soil respiration in summer) and decreased as the plants aged (oldest vegetation: 37% of total soil respiration in summer). Total annual soil C loss from the youngest and oldest communities was estimated to be 650 and 435 gCm-2 yr-1 respectively. Heathlands are cultural landscapes which are managed through cyclical cutting, burning or grazing practices. Understanding the C fluxes from these ecosystems provides information on the optimal management cycle-time to maximize C uptake and minimize C output. Inclusion of the predicted soil fluxes into a preliminary ecosystem C balance suggested that the youngest vegetation is a C sink while the oldest vegetation is a C source, indicating that shorter management cycles could reduce C emissions. [ABSTRACT FROM AUTHOR]

DEM error propagation methodology is extended to the derivation of vector-based objects (stream networks) using geostatistical simulations. First, point sampled elevations are used to fit a variogram model. Next 100 DEM realizations are generated using conditional sequential Gaussian simulation; the stream network map is extracted for each of these realizations, and the collection of stream networks is analyzed to quantify the error propagation. At each grid cell, the probability of the occurrence of a stream and the propagated error are estimated. The method is illustrated using two small data sets: Baranja hill (30m grid cell size; 16 512 pixels; 6367 sampled elevations), and Zlatibor (30m grid cell size; 15 000 pixels; 2051 sampled elevations). All computations are run in the open source software for statistical computing R: package geoR is used to fit variogram; package gstat is used to run sequential Gaussian simulation; streams are extracted using the open source GIS SAGA via the RSAGA library. The resulting stream error map (Information entropy of a Bernoulli trial) clearly depicts areas where the extracted stream network is least precise - usually areas of low local relief and slightly convex (0-10 difference from the mean value). In both cases, significant parts of the study area (17.3% for Baranja Hill; 6.2% for Zlatibor) show high error (H >0.5) of locating streams. By correlating the propagated uncertainty of the derived stream network with various land surface parameters sampling of height measurements can be optimized so that delineated streams satisfy the required accuracy level. Such error propagation tool should become a standard functionality in any modern GIS. Remaining issue to be tackled is the computational burden of geostatistical simulations: this framework is at the moment limited to small data sets with several hundreds of points. Scripts and data sets used in this article are available on-line via the www.geomorphometry.org website and can be easily adopted/adjusted to any similar case study. [ABSTRACT FROM AUTHOR]

DEM error propagation methodology is extended to the derivation of vector-based objects (stream networks) using geostatistical simulations. First, point sampled elevations are used to fit a variogram model. Next 100 DEM realizations are generated using conditional sequential Gaussian simulation; the stream network map is extracted for each of these realizations, and the collection of stream networks is analyzed to quantify the error propagation. At each grid cell, the probability of the occurrence of a stream and the propagated error are estimated. The method is illustrated using two small data sets: Baranja hill (30m grid cell size; 16 512 pixels; 6367 sampled elevations), and Zlatibor (30m grid cell size; 15 000 pixels; 2051 sampled elevations). All computations are run in the open source software for statistical computing R: package geoR is used to fit variogram; package gstat is used to run sequential Gaussian simulation; streams are extracted using the open source GIS SAGA via the RSAGA library. The resulting stream error map (Information entropy of a Bernoulli trial) clearly depicts areas where the extracted stream network is least precise - usually areas of low local relief, slightly concave. In both cases, significant parts of the study area (17.3% for Baranja Hill; 6.2% for Zlatibor) show high error (H > 0.5) of locating streams. By correlating the propagated uncertainty of the derived stream network with various land surface parameters sampling of height measurements can be optimized so that delin eated streams satisfy a required accuracy level. Remaining issue to be tackled is the computational burden of geostatistical simulations: this framework is at the moment limited to small to moderate data sets with several hundreds of points. Scripts and data sets used in this article are available on-line via the www.geomorphometry.org website and can be easily adopted/adjusted to any similar case study. [ABSTRACT FROM AUTHOR]

Root zone soil moisture is a key variable in many land surface hydrology models. Often, however, there is a mismatch in the spatial scales at which models simulate soil moisture and at which soil moisture is observed. This complicates model validation. The increased availability of detailed datasets on space-time variability of root-zone soil moisture allows for a posteriori analysis of the uncertainties in the relation between point-scale observations and the spatial mean. In this paper we analyze three comprehensive datasets from three different regions. We identify different strategies to select observation sites. For instance, sites can be located randomly or according to the rank stability concept. For each strategy, we present methods to quantify the uncertainty that is associated with this strategy. In general there is a large correspondence between the different datasets with respect to the relative uncertainties for the different strategies. For all datasets, the uncertainty can be strongly reduced if some information is available that relates soil moisture content at that site to the spatial mean. However this works best if the space-time dynamics of the soil moisture field are known. Selection of the site closest to the spatial mean on a single random date only leads to minor reduction of the uncertainty with respect to the spatial mean over seasonal timescales. Since soil moisture variability is the result of a complex interaction between soil, vegetation, and landscape characteristics, the soil moisture field will be correlated with some of these characteristics. Using available information, we show that the correlation with leaf area index or a wetness coefficient alone is insufficient to predict if a site is representative for the spatial mean soil moisture content. [ABSTRACT FROM AUTHOR]

Root zone soil moisture is a key variable in many land surface hydrology models. Often, however, there is a mismatch in the spatial scales at which models simulate soil moisture and at which soil moisture is observed. The limited spatial support of observations combined with the large spatial variability of the soil moisture field complicates model validation. The increased availability of detailed datasets on space-time variability of root-zone soil moisture allows for a posteriori analysis of the uncertainties in the relation between point-scale observations and the spatial mean. In this paper we analyze three comprehensive datasets from three different regions. We identify different strategies to select observation sites, and we present methods for quantifying the uncertainty that is associated with each strategy. In general there is a large correspondence between the different datasets with respect to the relative uncertainties for the different strategies. For all datasets, the uncertainty can be strongly reduced if some information is available that relates soil moisture at that site to the spatial mean. However this works best if the space-time dynamics of the soil moisture field are known. Selection of the site closest to the spatial mean on a single random date only leads to minor reduction of the uncertainty with respect to the spatial mean over seasonal timescales. Since soil moisture variability is the result of a complex interaction between soil, vegetation, and landscape characteristics, the soil moisture field will be correlated with some of these characteristics. Using available information, we show that the correlation with leaf area index or a wetness coefficient alone is insufficient to predict if a site is representative for the spatial mean soil moisture. [ABSTRACT FROM AUTHOR]