Systems Analysis, Integrated Assessment and Modelling
In SIAM, we develop and apply models and formal techniques in order to understand, demonstrate, and predict the behavior of natural, technical, social and economical systems that pertain to water and other natural resources. Read more
Global phosphorus losses from croplands under future precipitation scenarios
Phosphorus (P) losses from fertilized croplands to inland water bodies cause serious environmental problems. During wet years, high precipitation disproportionately contributes to P losses. We combine simulations of a gridded crop model and outputs from a number of hydrological and climate models to assess global impacts of changes in precipitation regimes on P losses during the 21st century. Under the baseline climate during 1991-2010, median P losses are 2.7 ± 0.5 kg P ha-1 year-1 over global croplands of four major crops, while during wet years, P losses are 3.6 ± 0.7 kg P ha-1 year–1. By the end of this century, P losses in wet years would reach 4.2 ± 1.0 (RCP2.6) and 4.7 ± 1.3 (RCP8.5) kg P ha-1 year-1 due to increases in high annual precipitation alone. The increases in P losses are the highest (up to 200%) in the arid regions of Middle East, Central Asia, and northern Africa. Consequently, in three quarters of the world’s river basins, representing about 40% of total global runoff and home up to 7 billion people, P dilution capacity of freshwater could be exceeded due to P losses from croplands by the end of this century.
Liu, W.; Ciais, P.; Liu, X.; Yang, H.; Hoekstra, A. Y.; Tang, Q.; Wang, X.; Li, X.; Cheng, L. (2020) Global phosphorus losses from croplands under future precipitation scenarios, Environmental Science and Technology, doi:10.1021/acs.est.0c03978, Institutional Repository
A review of long-term pesticide monitoring studies to assess surface water quality trends
Aquatic pesticide pollution from both agricultural and urban pest control is a concern in many parts of the world. Making an accurate assessment of pesticide exposure is the starting point to protecting aquatic ecosystems. This in turn requires the design of an effective monitoring program. Monitoring is also essential to evaluate the efficacy of mitigation measures aimed to curb pesticide pollution. However, empirical evidence for their efficacy can be confounded by additional influencing factors, most prominently variable weather conditions. This review summarizes the experiences gained from long-term (>5 years) pesticide monitoring studies for detecting trends and provides recommendations for their improvement. We reviewed articles published in the scientific literature, with a few complements from selected grey literature, for a total of 20 studies which fulfill our search criteria. Overall, temporal trends of pesticide use and hydrological conditions were the two most common factors influencing aquatic pesticide pollution. Eighteen studies demonstrated observable effects to surface water concentrations from changes in pesticide application rates (e.g., use restriction) and sixteen studies from interannual variability in hydrological conditions during the application period. Accounting for seasonal- and streamflow-related variability in trend analysis is important because the two factors can obscure trends caused by changes in pesticide use or management practices. Other mitigation measures (e.g., buffer strips) were only detectable in four studies where concentrations or loads were reduced by > 45%. Collecting additional agricultural (e.g., pesticide use, mitigation measures) and environmental (e.g., precipitation, stream flow) data, as well as establishing a baseline before the implementation of mitigation measures have been consistently reported as prerequisites to interpret water quality trends from long-term monitoring studies, but have rarely been implemented in the past.
Towards an integrated surface water quality assessment: aggregation over multiple pollutants and time
Surface water quality management requires foresighted decision making regarding long-term investments. It should consider multiple objectives (e.g. related to different pollutants and costs), integrate multiple sources of pollution (point and diffuse sources), and external conditions that change over time (climate, population and land-use changes). Multi-attribute value theory can support such decisions, especially the development of an assessment method. Integrated surface water quality assessment methods including micropollutants are currently lacking or in development in many countries. Important steps for the development of such an immission oriented and integrated surface water quality assessment method are discussed in this paper and exemplified for organic micropollutants. The proposed assessment method goes beyond simple pass-fail criteria for single substances. It provides a continuous assessment on a scale from zero to one based on five color-coded water quality classes and suggestions for the visualization of assessment results. It takes into account the toxicity of the micropollutants and their mixture to aquatic organisms by comparing measured concentrations to environmental quality standards (EQS). The focus of this paper is on aggregation over multiple substances and time. Advantages and disadvantages of different aggregation methods are discussed as well as their implications for practice. The consequences of different aggregation methods are illustrated with didactical examples and by an application of the proposed water quality assessment method to pesticide monitoring data from Switzerland. Recommendations are provided that account for the purpose of the assessment. Furthermore, the paper illustrates how the proposed method can facilitate dealing with uncertainty and a transparent communication of monitoring results to support water quality management decisions.
Effects of site selection and taxonomic resolution on the inference of stream invertebrate responses to environmental conditions
Key decisions in the design of biomonitoring programs include taxonomic resolution, geographic extent, and site selection, each of which can affect our ability to infer human impacts on biodiversity from biomonitoring data. These decisions are constrained by monitoring goals and budget limitations, which may require trade-offs between them. In this study, we use species distribution models (SDMs) to assess the effects of key decisions in biomonitoring design on our ability to infer the effects of natural and anthropogenic environmental conditions on the occurrence of benthic macroinvertebrates in streams. We compared 4 datasets that differ in their site-selection strategy, geographic extent, and taxonomic resolution using data from Swiss federal and cantonal biomonitoring programs. We used individual SDMs with 3-fold cross validation to identify the environmental variables that best predict the probability of taxa occurrence across the datasets. We then used a hierarchical multi-species distribution model (hmSDM) to identify how key aspects of biomonitoring design influence the relative importance of the selected explanatory (predictor) variables in the model as well as the model’s predictive performance. The relative importance of the explanatory variables in the hmSDMs was lowest for the dataset with a grid-based site-selection approach and family-level resolution. An increase in predictive performance was achieved by either using a species-level taxonomic resolution for Ephemeroptera, Plecoptera, and Trichoptera or by combining different biomonitoring programs at the family level to increase the number of sites and improve the coverage of environmental conditions. Selecting monitoring sites to provide a good coverage of environmental conditions, while also targeting sites with rare combinations of environmental conditions, could further improve biomonitoring program data. Models based on finer taxonomic resolution revealed that widespread families consist of species and genera with different and stronger responses to environmental conditions. However, many families include species that are too rare to allow inference of significant responses to environmental conditions. We show that hmSDMs of stream invertebrates can contribute to the selection of specific taxa for identification at finer taxonomic resolution. This strategy could facilitate the standardization and combination of multiple biomonitoring datasets and improve the identifiability of stream invertebrate responses to environmental conditions in biomonitoring programs.
Caradima, B.; Reichert, P.; Schuwirth, N. (2020) Effects of site selection and taxonomic resolution on the inference of stream invertebrate responses to environmental conditions, Freshwater Science, 39(3), 415-432, doi:10.1086/709024, Institutional Repository
Characterizing fast herbicide transport in a small agricultural catchment with conceptual models
Herbicide pollution in headwater streams due to agricultural practices is a major environmental concern and is characterized by episodic peak concentrations from fast transport paths. We rely on previous experimental studies in a small (1.2km2) agricultural catchment in the Swiss Plateau, to model dynamic diffuse herbicide pollution with emphasis on fast transport paths in a conceptual modelling framework at the catchment scale. We show how experimentalists’ understanding of the fate of herbicides (perceptual model) can be translated into conceptual models considering sorption, degradation, and fast transport of water and chemicals facilitated by impervious surfaces, tile drains and artificial shortcuts. Different types of experimental data (streamflow, high-frequency concentration measurements, and soil–water distribution coefficients) are used in a joint Bayesian inference of model parameters. We assess the ability of different spatial configurations of hydrological response units in explaining observed heterogeneity in transport behaviour of two corn herbicides. Thereby, we find that (1) relatively simple conceptual models can provide a realistic description of herbicide fate in small agricultural catchments, (2) accounting for spray drift onto hard surfaces is necessary to avoid a severe model bias during the first rainfall event after application, and (3) including catchment-specific experimentalist knowledge about important elements like artificial shortcuts and tile drains leads to a reduction in uncertainty of 30% compared to the more conventional conception of the proximity to the stream as the dominant risk factor.
Since spring 2018, the newly developed underwater camera Aquascope has been recording a wide variety of plankton species in Lake Greifen. These sensitive organisms can thus, for the first time, be observed undisturbed in their natural habitat – an important step towards automated monitoring of water quality and aquatic biodiversity.