Physics-informed machine learning algorithms for forecasting sediment yield: an analysis of physical consistency, sensitivity, and interpretability.

The sediment transport, involving the movement of the bedload and suspended sediment in the basins, is a critical environmental concern that worsens water scarcity and leads to degradation of land and its ecosystems. Machine learning (ML) algorithms have emerged as powerful tools for predicting sediment yield. However, their use by decision-makers can be attributed to concerns regarding their consistency with the involved physical processes. In light of this issue, this study aims to develop a physics-informed ML approach for predicting sediment yield. To achieve this objective, Gaussian, Center, Regular, and Direct Copulas were employed to generate virtual combinations of physical of the sub-basins and hydrological datasets. These datasets were then utilized to train deep neural network (DNN), conventional neural network (CNN), Extra Tree, and XGBoost (XGB) models. The performance of these models was compared with the modified universal soil loss equation (MUSLE), which serves as a process-based model. The results demonstrated that the ML models outperformed the MUSLE model, exhibiting improvements in Nash-Sutcliffe efficiency (NSE) of approximately 10%, 18%, 32%, and 41% for the DNN, CNN, Extra Tree, and XGB models, respectively. Furthermore, through Sobol sensitivity and Shapley additive explanation-based interpretability analyses, it was revealed that the Extra Tree model displayed greater consistency with the physical processes underlying sediment transport as modeled by MUSLE. The proposed framework provides new insights into enhancing the accuracy and applicability of ML models in forecasting sediment yield while maintaining consistency with natural processes. Consequently, it can prove valuable in simulating process-related strategies aimed at mitigating sediment transport at watershed scales, such as the implementation of best management practices.

Water use efficiency across scales: from genes to landscapes.

Water scarcity is already set to be one of the main issues of the 21st century, because of competing needs between civil, industrial, and agricultural use. Agriculture is currently the largest user of water, but its share is bound to decrease as societies develop and clearly it needs to become more water efficient. Improving water use efficiency (WUE) at the plant level is important, but translating this at the farm/landscape level presents considerable challenges. As we move up from the scale of cells, organs, and plants to more integrated scales such as plots, fields, farm systems, and landscapes, other factors such as trade-offs need to be considered to try to improve WUE. These include choices of crop variety/species, farm management practices, landscape design, infrastructure development, and ecosystem functions, where human decisions matter. This review is a cross-disciplinary attempt to analyse approaches to addressing WUE at these different scales, including definitions of the metrics of analysis and consideration of trade-offs. The equations we present in this perspectives paper use similar metrics across scales to make them easier to connect and are developed to highlight which levers, at different scales, can improve WUE. We also refer to models operating at these different scales to assess WUE. While our entry point is plants and crops, we scale up the analysis of WUE to farm systems and landscapes.

COVID-19 response in Africa: impacts and lessons for environmental management and climate change adaptation.

The COVID-19 pandemic adds pressure on Africa; the most vulnerable continent to climate change impacts, threatening the realization of most Sustainable Development Goals (SDGs). The continent is witnessing an increase in intensity and frequency of extreme weather events, and environmental change. The COVID-19 was managed relatively well across in the continent, providing lessons and impetus for environmental management and addressing climate change. This work examines the possible impact of the COVID-19 pandemic on the environment and climate change, analyses its management and draws lessons from it for climate change response in Africa. The data, findings and lessons are drawn from peer reviewed articles and credible grey literature on COVID-19 in Africa. The COVID-19 pandemic spread quickly, causing loss of lives and stagnation of the global economy, overshadowing the current climate crisis. The pandemic was managed through swift response by the top political leadership, research and innovations across Africa providing possible solutions to COVID-19 challenges, and redirection of funds to manage the pandemic. The well-coordinated COVID-19 containment strategy under the African Centers for Disease Control and Prevention increased sharing of resources including data was a success in limiting the spread of the virus. These strategies, among others, proved effective in limiting the spread and impact of COVID-19. The findings provide lessons that stakeholders and policy-makers can leverage in the management of the environment and address climate change. These approaches require solid commitment and practical-oriented leadership. Supplementary Information: The online version contains supplementary material available at 10.1007/s10668-023-02956-0.

A new index on agricultural land greenhouse gas emissions in Africa.

Africa emits the lowest amounts of greenhouse gases (GHGs) into the global GHG budget. However, the continent remains the most vulnerable continent to the effects of climate change. The agricultural sector in Africa is among the most vulnerable sectors to climate change. Also, as a dominant agricultural sector, African agriculture is increasingly contributing to climate change through GHG emissions. Research has so far focused on the effects of GHG emissions on the agricultural and other sectors with very little emphasis on monitoring and quantifying the spatial distribution of GHG emissions from agricultural land in Africa. This study develops a new index: African Agricultural Land Greenhouse Gas Index (AALGGI) that uses scores and specific scale ranges for carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O) to map the spatial variations in regional GHG emissions across Africa. The data for the three main GHGs (CO2, CH4, and N20) were downloaded from FAOSTAT. The data were analyzed through the newly developed African Agricultural Land Greenhouse Gas Index (AALGGI). This is an empirical index with scores ranging from 0 to 10, with higher scores indicating higher levels of emissions. The results show that Southern and North African regions have the lowest amounts of agricultural land GHG emissions, with AALGGIs of 3.5 and 4.5, respectively. East Africa records the highest levels of GHG emissions, with an AALGGI of 8 followed by West Africa with an AALGGI of 7.5. With the continental mean or baseline AALGGI being 5.8, East and Middle Africa are above the mean AALGGI. These results underscore the fact that though Africa, in general, is not a heavy emitter of GHGs, African agricultural lands are increasingly emitting more GHGs into the global GHG budget. The low AALGGIs in the more developed parts of Africa such as Southern and North Africa are explained by their domination in other GHG emitting sectors such as industrialization and energy. The high rates of emissions in East Africa and Middle Africa are mainly linked to intensive traditional farming practices/processes and deforestation. These findings underscore the need to further leverage climate change mitigation actions and policy in Africa and most importantly the co-benefits of mitigation and adaptations in the most vulnerable regions.

Water quality index modeling using random forest and improved SMO algorithm for support vector machine in Saf-Saf river basin.

The water quality index is one of the prominent general indicators to assess and classify surface water quality, which plays a critical role in river water resources practices. This research constructs a hybrid artificial intelligence model namely sequential minimal optimization-support vector machine (SMO-SVM) along with random forest (RF) as a benchmark model for predicting water quality values at the Wadi Saf-Saf river basin in Algeria. The fifteen input water quality datasets such as biochemical oxygen demand (BOD), oxygen saturation (OS), the potential for hydrogen (pH), chemical oxygen demand (COD), chloride (Cl-), dissolved oxygen (DO), electrical conductivity (EC), total dissolved solids (TDS), nitrate-nitrogen (NO3-N), nitrite-nitrogen (NO2-N), phosphate (PO43-), ammonium (NH4+), temperature (T), turbidity (NTU), and suspended solids (SS) were employed for constructing the predictive models. Different input data combinations are evaluated in terms of predictive performance, using a set of statistical metrics and graphical representation. Results show that less than 40% of samples were observed to be poor quality water during the dry season in downstream northeastern part of the basin. The findings also show that the RF model mostly generates more precise water quality index predictions than the SMO-SVM model for both training and testing stages. Although thirteen input parameters attain the optimal predictive performance (R2 testing = 0.82, RMSE testing = 5.17), a couple of five input parameters, e.g., only pH, EC, TDS, T, and saturation, gives the second optimal predictive precision (R2 test = 0.81, RMSE testing = 5.55). The sensitivity analysis results indicate a greater sensitivity by the all input variables chosen except NO2- of the predictive outcomes to the earlier influencing water quality parameters. Overall, the RF model reveals an improvement on earlier tools for predicting water quality index, according to predictive performance and reducing in the number of input variables.

Predicting daily pore water pressure in embankment dam: Empowering Machine Learning-based modeling.

Dam safety assessment is important to implement the appropriate measures to avoid a dam break disaster as part of the water reservoirs management process. Prediction-based approaches are valuable to compare the actual measurements with the simulated values to proactively detect anomalies. However, the application of the conventional hydrostatic seasonal time (HST) has some limitations related to an instantaneous response of the dam to environmental factors, which can lead to inaccurate prediction and interpretation, especially for daily measurements. Besides, the generalization ability (GA) of these models is not analyzed enough despite its crucial importance in selecting the appropriate models. In this study, the multiple linear regression (MLR), artificial neural network (ANN), support vector regression (SVR), and adaptive boosting (AdaBoost) models with nonlinear autoregressive exogenous (NARX) inputs are proposed to incorporate the response delay of the dam to the hydraulic load. Thus, these models were evaluated and compared with the HST model for predicting the daily pore water pressure in an embankment dam. Moreover, we proposed a classification method of the models into four categories, namely perfect, excellent, good, and poor according to the GA. Results show that, except for the AdaBoost, the other ML models outperformed the traditional statistical approach (HST) in terms of prediction accuracy as well as the GA. Overall, the study results provide new insights in enhancing the monitoring processes and dam safeties by detecting the anomalies early through the comparison of the measurements and simulated results produced by the best-fitted models from the confidence interval (CI) perspective.

Morocco's coastal aquifers: Recent observations, evolution and perspectives towards sustainability.

During the last decades, the coastal areas of Morocco have witnessed an intense socioeconomic development associated with a continuous population growth and urban extension. This has led to an overexploitation of coastal aquifers leading to a degradation of their water quality. In order to obtain large scale overview on the quality status of Morocco's coastal aquifers (MCA) to assist national water managers to make informed decisions, a comprehensive scrutinization of the MCA against common indicators and using unified methods is essential. In this study, databases from thirteen MCA were analyzed, using multivariate statistical approaches and graphical methods in order to investigate the degree of mineralization in each aquifer and to identify the main salinization processes prevailing in groundwater. The results showed that the dominant groundwater types are Na-Cl, Ca-Mg-Cl, Ca-Mg-SO4, Ca-Mg-HCO3 and Ca-HCO3-Cl. The Gibbs diagram and the seawater contribution (0-37%) indicate that the mineralization is mainly due to the seawater intrusion and water-rock interaction. The salinity degree diagram illustrates that almost all groundwater samples are located in the moderate to very saline zone, indicating that MCA are recharged by water from variable sources. The groundwater quality assessment shows a deterioration, particularly by seawater intrusion and significant nitrate pollution. The temporal evolution confirm that the MCA are influenced by seawater namely in the Atlantic part. The Wilcox and USSL diagram indicate that the majority of sampled water are unsuitable for irrigation uses. In addition, and by referring to the WHO and the Moroccan standards for water potability, large number of samples from the groundwaters of the MCA is not fully adequate for drinking purposes. A set of management actions (e,g., artificial recharge) are proposed in order to mitigate the effect of groundwater overexploitation and seawater intrusion to ensure the sustainability of MCA.

Vulnerability of maize, millet, and rice yields to growing season precipitation and socio-economic proxies in Cameroon.

In sub-Saharan Africa growing season precipitation is affected by climate change. Due to this, in Cameroon, it is uncertain how some crops are vulnerable to growing season precipitation. Here, an assessment of the vulnerability of maize, millet, and rice to growing season precipitation is carried out at a national scale and validated at four sub-national scales/sites. The data collected were historical yield, precipitation, and adaptive capacity data for the period 1961-2019 for the national scale analysis and 1991-2016 for the sub-national scale analysis. The crop yield data were collected for maize, millet, and rice from FAOSTAT and the global yield gap atlas to assess the sensitivity both nationally and sub-nationally. Historical data on mean crop growing season and mean annul precipitation were collected from a collaborative database of UNDP/Oxford University and the climate portal of the World Bank to assess the exposure both nationally and sub-nationally. To assess adaptive capacity, literacy, and poverty rate proxies for both the national and regional scales were collected from KNOEMA and the African Development Bank. These data were analyzed using a vulnerability index that is based on sensitivity, exposure, and adaptive capacity. The national scale results show that millet has the lowest vulnerability index while rice has the highest. An inverse relationship between vulnerability and adaptive capacity is observed. Rice has the lowest adaptive capacity and the highest vulnerability index. Sub-nationally, this work has shown that northern maize is the most vulnerable crop followed by western highland rice. This work underscores the fact that at different scales, crops are differentially vulnerable due to variations in precipitation, temperature, soils, access to farm inputs, exposure to crop pest and variations in literacy and poverty rates. Therefore, caution should be taken when transitioning from one scale to another to avoid generalization. Despite these differences, in the sub-national scale, western highland rice is observed as the second most vulnerable crop, an observation similar to the national scale observation.