Data Scarce Contexts Research Articles

Burden assessment of antimicrobial use and resistance in livestock in data-scarce contexts.

Misuse and overuse of antimicrobials in livestock production are identified as drivers for antimicrobial resistance (AMR). To improve decision-making concerning livestock health, it is important to understand the impact of AMR in livestock and aquaculture, within and beyond farm level, as well as expenditure on antimicrobial use (AMU). Such understanding provides grounds for systematic disease prioritisation and establishes a baseline for understanding the value of different strategies to mitigate animal health problems and for the monitoring and evaluation of the impact of those strategies. Yet limited data availability and quality surrounding AMU and AMR create barriers to furthering the knowledge of such impact. These data constraints are also more prevalent in contexts that lack the necessary resources to develop and maintain systematic and centralised data collection and collation systems. Even in regions with robust AMU and AMR monitoring systems in place, data limitations remain, such that the expenditure on antimicrobials and impacts of AMR remain unclear. Additionally, the current research funding strategies have been less focused on primary data collection, adding further barriers to filling the data void and reducing the global AMU/AMR knowledge gap. To work around the data scarcity and leverage previous and ongoing research efforts, it is vital to gain comprehensive knowledge of the people, projects and research consortia dedicated to the topic of AMU/AMR.

Watershed scale soil moisture estimation model using machine learning and remote sensing in a data-scarce context

Soil moisture content can be used to predict drought impact on agricultural yield better than precipitation. Remote sensing is viable source of soil moisture data in instrument-scarce areas. However, space-based soil moisture estimates lack suitability for daily and high-resolution agricultural, hydrological, and environmental applications. This study aimed to assess the potential of the random forest machine learning technique to enhance the spatial resolution of remote soil moisture products from the SMAP satellite. Models were built using random forest for spatial downscaling of SMAP-L3-E, then visually and statistically evaluated for disaggregation quality. The impact of topography, soil properties, and precipitation on the downscaled soil moisture was examined. The relationship between downscaled soil moisture and in-situ soil moisture was analyzed. The results indicate that the proposed method demonstrated spatial and hydrological coherence, along with a satisfactory downscaling quality. Statistical validation indicated suitable generalization error for scientific and practical use (RMSE < 0.05 cm3 cm-3). Random forest effectively achieved spatial downscaling of SMAP-L3-E in the study area. Principal component and spatial analysis revealed dependence of downscaled soil moisture on elevation, soil organic carbon content, clay content, and saturated hydraulic conductivity, mainly under near-saturation conditions. Regarding validation against in-situ data, downscaled soil moisture explained in-situ soil moisture well under low soil water content ( = 0.624). Downscaling performance deteriorates for water contents between 0.40 to 0.50 cm3 cm-3, suggesting inadequacy under near saturation conditions at a daily temporal frequency. However, coarser temporal aggregations (7 to 10 days) yielded an average 0.98 correlation coefficient, regardless of saturation conditions. These results could potentially be applied in irrigation planning, soil physics studies and hydrology monitoring, to forecasting the occurrence of droughts, leaching of contaminants, surface runoff modeling, carbon cycle studies, soil's capacity to store and provide nutrients.

Validation of the intrinsic vulnerability to pollution of fractured siliciclastic aquifers using natural background levels

Aquifers are important water reserves in water-scarce areas. However, uncontrolled human activities and lack of knowledge of aquifer characteristics increase their vulnerability to pollution. Assessing intrinsic vulnerability is essential for improving groundwater management. However, traditional tools, such as DRASTIC, ignore relevant parameters that, in particular contexts, lead to underestimating vulnerability. This work assesses the intrinsic vulnerability to pollution of a siliciclastic aquifer system (Shallow and Upper) by comparing the DRASTIC and modified DRASTIC methods. Modified DRASTIC included parameters such as Land Use and fracturing patterns relevant in populated areas where hydrogeological configurations involve highly cemented and fractured siliciclastic rocks with secondary porosity that define the main pathways for transporting contaminants from the surface. The intrinsic vulnerability was validated using Natural Background Levels (NBLs) of contaminant indicator ions (NO3−, PO43− and SO₄2-) to identify areas where the concentration of these indicators surpassed the NBLs. These areas were compared with the vulnerability classes from DRASTIC methods. NBLs were: 2.12 mg/l, 0.05 mg/L and 26.77 mg/L (Shallow aquifer), and 3.17 mg/L, 0.17 mg/L and 5.06 ml/L (Upper aquifer) for the ions of NO3−, PO43− and SO₄2-, respectively. The ion concentrations surpassing NBLs were spatially distributed, being more notable in the dry season and showing a greater coincidence with the high vulnerability class of the Modified DRASTIC. Thus, the modified DRASTIC better represented the high vulnerability of the aquifer system to pollution and the negative effect of anthropic activities. Additionally, Pearson's coefficients were calculated between the concentrations of contaminant indicator ions and intrinsic vulnerability indices of DRASTIC methods. However, the coefficients lack statistical significance. These results support the relevance of adding land use and fracturing patterns to assess the intrinsic vulnerability of fractured, siliciclastic aquifers in populated regions and show NBLs as a better validation method than Pearson's coefficients in data-scarce contexts.

Data-driven estimates for the geostatistical characterization of subsurface hydraulic properties

Abstract. The geostatistical characterization of the subsurface is confronted with the double challenge of large uncertainties and high exploration costs. Making use of all available data sources is consequently very important. Bayesian inference is able to mitigate uncertainties in such a data-scarce context by drawing on available background information in the form of a prior distribution. To make such a prior distribution transparent and objective, it should be calibrated against a data set containing estimates of the target variable from available sites. In this study, we provide a collection of covariance and/or variogram functions of the subsurface hydraulic parameters from a large number of sites. We analyze this data set by fitting a number of widely used variogram model functions and show how they can be used to derive prior distributions of the parameters of said functions. In addition, we discuss a number of conclusions that can be drawn for our analysis and possible uses for the data set.

Towards the Improvement of Soil Salinity Mapping in a Data-Scarce Context Using Sentinel-2 Images in Machine-Learning Models

Several recent studies have evidenced the relevance of machine-learning for soil salinity mapping using Sentinel-2 reflectance as input data and field soil salinity measurement (i.e., Electrical Conductivity-EC) as the target. As soil EC monitoring is costly and time consuming, most learning databases used for training/validation rely on a limited number of soil samples, which can affect the model consistency. Based on the low soil salinity variation at the Sentinel-2 pixel resolution, this study proposes to increase the learning database’s number of observations by assigning the EC value obtained on the sampled pixel to the eight neighboring pixels. The method allowed extending the original learning database made up of 97 field EC measurements (OD) to an enhanced learning database made up of 691 observations (ED). Two classification machine-learning models (i.e., Random Forest-RF and Support Vector Machine-SVM) were trained with both OD and ED to assess the efficiency of the proposed method by comparing the models’ outcomes with EC observations not used in the models´ training. The use of ED led to a significant increase in both models’ consistency with the overall accuracy of the RF (SVM) model increasing from 0.25 (0.26) when using the OD to 0.77 (0.55) when using ED. This corresponds to an improvement of approximately 208% and 111%, respectively. Besides the improved accuracy reached with the ED database, the results showed that the RF model provided better soil salinity estimations than the SVM model and that feature selection (i.e., Variance Inflation Factor-VIF and/or Genetic Algorithm-GA) increase both models´ reliability, with GA being the most efficient. This study highlights the potential of machine-learning and Sentinel-2 image combination for soil salinity monitoring in a data-scarce context, and shows the importance of both model and features selection for an optimum machine-learning set-up.

Dry Sanitation Technologies: Developing a Simplified Multi-Criteria Decision Analysis Tool

Safely managed sanitation is indispensable for societies to ensure public health, environmental protection, and economic and social development. This could be achieved, in large areas of the world, through dry sanitation systems. Dry sanitation systems are especially used in water-scarce regions and low-income households. In dense peri-urban areas, the achievement of safe sanitation necessitates a comprehensive fecal sludge management (FSM) service chain, surpassing the mere provision of latrines. This research introduces an automatic Multi-Criteria Decision Analysis (MCDA) approach, which focuses on the particular interface/storage stage of the FSM service chain. The tool aims to support the decision-making process and may be especially useful in the early stages of sanitation planning as it compares different technologies with potential application in low-income countries. It includes different criteria and parameters for the Social, Economic, Technical, and Environmental dimensions of dry sanitation options (SETEds), being adaptable to different contexts and to different priorities. The main key strengths of the tool were found to be its minimal data requirements and ability to customize operation and maintenance cost parameters. These features are particularly relevant in data-scarce contexts, where traditional models may lead to unreliable recommendations or lack of solution ownership by users. The tool was applied to the Ambriz case study, a coastal town in Northern Angola, in West Africa. The obtained results are analyzed and show the tool’s application provides technology recommendations aligned with the site and community characteristics.

Perceptions, trends and adaptation to climate change in Yala wetland, Kenya

Purpose Tropical wetland ecosystems are threatened by climate change but also play a key role in its mitigation and adaptation through management of land use and other drivers. Local-level assessments are needed to support evidence-based wetland management in the face of climate change. This study aims to examine the local communities’ knowledge and perception of climate change in Yala wetland, Kenya, and compare them with observed data on climate trends. Such comparisons are useful to inform context-specific climate change adaptation actions. Design/methodology/approach The study used a mixed methods approach that combined analysis of climate data with perceptions from the local community. Gridded data on temperature and rainfall for the period from 1981 to 2018 were compared with data on climate change perceptions from semi-structured questionnaires with 286 key informants and community members. Findings Majority of the respondents had observed changes in climate parameters – severe drought (88.5%), increased frequency of floods (86.0%) and irregular onset and termination of rains (90.9%) in the past 20 years. The perceptions corresponded with climate trends that showed a significant increasing trend in the short rains and the average maximum temperature, high incidence of very wet years and variability in onset and termination of rainfall between 1981 and 2018. Gender, age and education had little influence on knowledge and awareness of climate change, except for frequency of floods and self-reported understanding of climate change. The community perceived the wetland to be important for climate change adaptation, particularly the provision of resources such as grazing grounds during drought. Research limitations/implications The study faced challenges of low sample size, use of gridded climate data and reproducibility in other contexts. The results of this study apply to local communities in a tropical wetland in Western Kenya, which has a bi-modal pattern of rainfall. The sample of the study was regional and may therefore not be representative of the whole of Kenya, which has diverse socioeconomic and ecological contexts. Potential problems have been identified with the use of gridded data (for example, regional biases in models), although their usefulness in data scarce contexts is well established. Moreover, the sample size has been found to be a less important factor in research of highly complex socio-ecological systems where there is an attempt to bridge natural and social sciences. Practical implications This study addresses the paucity of studies on climate change trends in papyrus wetlands of sub-Saharan Africa and the role of local knowledge and perceptions in influencing the management of such wetlands. Perceptions largely influence local stakeholders’ decisions, and a study that compares perceptions vs “reality” provides evidence for engagement with the stakeholders in managing these highly vulnerable ecosystems. The study showed that the local community’s perceptions corresponded with the climate record and that adaptation measures are already ongoing in the area. Originality/value This study presents a case for the understanding of community perceptions and knowledge of climate change in a tropical wetland under threat from climate change and land use change, to inform management under a changing climate.

Comparing spatially explicit approaches to assess social vulnerability dynamics to flooding

Madagascar is one of the poorest countries in the world, and therefore highly vulnerable to tropical cyclone-induced coastal and inland flooding. This study aims to assess Madagascar's social vulnerability to flooding resulting from tropical storms, in a spatially explicit manner ,accounting for dynamics in vulnerability over time. For this, the research applied three different social vulnerability models in a data-scarce context, for the years 2009 and 2018. Around these years, two tropical cyclones hit the island (in 2008 and 2018). The three models are 1) inductive (social vulnerability index: SoVI), 2) deductive (Weighted Median, WM), and 3) social vulnerability profiling (SVP). Our results show that the most vulnerable regions are in the south of Madagascar, which is consistent across all three models. While the calculated vulnerability score (based on hazard and exposure data) indicated a decrease in vulnerability over time, only the SVP predicted a similar vulnerability decrease, but this comparison is surrounded by uncertainty since the 2008 and 2018 events differ in flood hazard characteristics. The application of the SoVI method in such regions has some limitations, for example, introducing subjective modeling decisions. The application of the WM method could be suitable, but only if relations with variables and social vulnerability are known and understood. The SVP model seems a suitable approach for a first scoping study of social vulnerability, but it provides less insight into spatial variation. The main recommendation from this study is to further focus future research on model validation in data-scarce regions and to assess model accuracy by exploring validation data.

Leveraging graph convolutional networks for semi-supervised fault diagnosis of HVAC systems in data-scarce contexts

Leveraging graph convolutional networks for semi-supervised fault diagnosis of HVAC systems in data-scarce contexts

Mapping and assessing ecosystem services for sustainable policy and decision-making in Eritrea

The mapping and assessment of ecosystems and their services (MAES) is key to inform sustainable policy and decision-making at national and sub-national levels. Responding to the paucity of research in sub-Saharan Africa, we conduct a pilot study for Eritrea that aims to map and assess the temporal dynamics of key ecosystems and their services. We reviewed policy and legal documents, analyzed land cover changes and estimated the potential for ecosystem services supply through an expert-based matrix approach. Our results showed that from 2015 to 2019, the potential supply of the ecosystem services analyzed (e.g., crop provisioning, water supply and recreation) increased, with the exception of wood supply. Overall, our study presents policy-relevant insights as to where to conserve, develop, or restore ecosystem services supply in Eritrea. Our approach is transferable to similar data scarce contexts and can thereby support policies toward more sustainable land development for people and nature.

A census from heaven: Unraveling the potential of deep learning and Earth Observation for intra-urban population mapping in data scarce environments

Urban population distribution maps are vital elements for monitoring the Sustainable Development Goals, appropriately allocating resources such as vaccination campaigns, and facilitating evidence-based decision making. Typically, population distribution maps are derived from census data from the region of interest. Nevertheless, in several low- and middle-income countries, census information may be unreliable, outdated or unsuitable for spatial analysis at the intra-urban level, which poses severe limitations in the development of urban population maps of adequate quality. To address these shortcomings, we deploy a novel framework utilizing multisource Earth Observation (EO) information such as Sentinel-2 and very-high-resolution Pleiades imagery, openly available building footprint datasets, and deep learning (DL) architectures, providing end-to-end solutions to the production of high quality intra-urban population distribution maps in data scarce contexts. Using several case studies in Sub-Saharan Africa, namely Dakar (Senegal), Nairobi (Kenya) and Dar es Salaam (Tanzania), our results emphasize that the combination of DL and EO data is very potent and can successfully capture relationships between the retrieved image features and population counts at fine spatial resolutions (100 meter). Moreover, for the first time, we used state-of-the-art domain adaptation methods to predict population distributions in Dar es Salaam and Nairobi (R 2 = 0.39, 0.60) that did not require national census or survey data from Kenya or Tanzania, but only a sample of training locations from Dakar. The DL architecture is based on a modified ResNet-18 model with dual-streams to analyze multi-modal data. Our findings have strong implications for the development of a new generation of urban population products that are an output of end-to-end solutions, can be updated frequently and rely completely on open data. • An EO-DL framework to model intra-urban population in the face of data scarcity. • Building footprints is a key variable to ensure transferability in areas with no local data. • A consistency-based domain adaptation enhances the transferability outcome. • Sentinel-2 imagery has good potential to retrieve fine-scale population patterns.

Estimating poverty for refugees in data-scarce contexts: an application of cross-survey imputation

The increasing growth of forced displacement worldwide has brought more attention to measuring poverty among refugee populations. However, refugee data remain scarce, particularly regarding income or consumption. We offer a first attempt to measure poverty among refugees using cross-survey imputation and administrative and survey data collected by the United Nations High Commissioner for Refugees (UNHCR). Employing a small number of predictors currently available in the UNHCR registration system, the proposed methodology offers out-of-sample predicted poverty rates that are not statistically different from actual poverty rates. These estimates are robust to different poverty lines, perform well according to targeting indicators, and are more accurate than those based on asset indexes or proxy means tests. They can also be obtained with relatively small samples. We additionally show that it is feasible to provide poverty estimates for one geographical region based on existing data from another similar region.

The geopolitical economy of state-led intelligence-commerce: two examples from Iraq and West Germany

ABSTRACT This article argues that intelligence agencies are actors in the global political economy. Two key arguments are advanced: firstly, that states use intelligence-commerce to secretly pursue economic, strategic aims, which they have to publicly shun for political or normative reasons. Secondly, that states' specific use of intelligence-commerce is related to their structural position within the international political economy. To demonstrate that the use of intelligence-commerce exceeds regime-type and straddles the presumed, strict boundary of ‘democratic' and ‘authoritarian' intelligence, this article uses Iraqi and West German intelligence-commerce as two, related empirical examples. The article applies an understanding of the state as a bundle of practices organized around the hegemonic principles of nationalism and raison d’etat, which manifest both discursively and materially in intelligence agencies, and discusses research methodologies in the data-scarce context of intelligence scholarship.

How widespread is fluoride contamination of Ghana’s groundwater?

Spatial modelling using machine-learning methods to predict fluoride occurrence in groundwater has vast potential in data-scarce contexts such as Ghana. Hazard risk maps can help inform health policy targeted towards communities at high risk.

Data-centric or algorithm-centric: Exploiting the performance of transfer learning for improving building energy predictions in data-scarce context

Data-driven methods have gained increasing popularity due to their high-convenience and high-accuracy in practice. Considering the wide discrepancies in data availability across different buildings, transfer learning can be applied to improve the feasibility and robustness of data-driven solutions for individual buildings. In principle, the performance of transfer learning can be enhanced from two perspectives, i.e., the algorithm-centric and data-centric perspectives. The algorithm-centric perspective highlights the adoption of advanced learning algorithm, while the data-centric perspective emphasizes the preparation of proper data for cross-building sharing. At present, there is a lack of studies to systematically compare the performance of the above-mentioned strategies for building energy predictions in a broad range of building types. This study, therefore, investigates the actual performance of transfer learning in data-scarce context, i.e., target buildings have insufficient/extremely limited operational data for model calibrations and domain adaptations. Various transfer learning methods, using different learning algorithms and source data utilization schemes, have been developed and applied for performance comparisons. Comprehensive data experiments have been designed using 600 actual buildings to draw statistically significant conclusions. The results are helpful for quantifying the behavioral patterns of transfer learning, and providing practical guidelines to develop cost-effective data-driven solutions for building energy predictions.

Calculating Mexico City’s Food Supply: Methodological Insights for Regionalizing Food Data at the Urban Scale

Tackling the environmental impacts of food systems can play a crucial role in achieving urban sustainability. Robust accounts of urban food supply must inform policymaking. Current research on urban food supply often relies on extrapolating national data sets to the urban scale, obscuring the diversity of food consumption patterns across regions and across the urban–rural divide within countries. One illustration of this is the case of Mexico, where biocultural diversity has shaped diverse diets across the country. In this article, we present a method to estimate urban food supply in the data-scarce context of Mexico City. We combine national data on food supply with food consumption data from a spatially explicit food questionnaire to estimate urban food supply. We estimate the food supply for the Mexico City Metropolitan Area, breaking it down by food group, and present key differences with national-scale data. In particular, consumption of animal products (meat and milk) is higher than the national average, and corn consumption is lower. We conclude that it is crucial to produce precise accounts of urban diets to enable robust analyses of the environmental impacts of urban food supply, and reflect on the limitations and opportunities of our method.

Estimating water input in the mining industry in Brazil: A methodological proposal in a data-scarce context

Metal mining plays a significant role in the Brazilian economy since its foundation as an overseas colony. The rapid increase in ore extraction brings along pressures on the country’s water resources, as mining is a particularly water-intensive activity. However, site-specific data on water input and management are scarce. We propose a methodology for estimating water input in mining at a high geographical resolution. We focus on the three key metals mined in Brazil: iron, aluminum (i.e. bauxite ore), and copper, and derive water input coefficients for all mines from governmental and corporate sources as well as from the literature. We estimate that overall, the sum of the water inputs estimated for Brazilian bauxite, copper, and iron ore mining decreased by 15% from an average of 506.5±62.4 hm3 in 2014 to an average of 408.4±67.2 hm3 in 2017. The regions where most water was appropriated were Northern (Pará state) and Southeast (Minas Gerais) for iron, Northern (Pará) for aluminum, and Northern (Pará) and Central West (Goiás) for copper. We show that there are still significant consistency and data availability gaps, and that further work is still necessary to improve site-specific reporting and open access to data collected by public institutions.

When Enough Is Really Enough? On the Minimum Number of Landslides to Build Reliable Susceptibility Models

Mapping existing landslides is a fundamental prerequisite to build any reliable susceptibility model. From a series of landslide presence/absence conditions and associated landscape characteristics, a binary classifier learns how to distinguish potentially stable and unstable slopes. In data rich areas where landslide inventories are available, addressing the collection of these can already be a challenging task. However, in data scarce contexts, where geoscientists do not get access to pre-existing inventories, the only solution is to map landslides from scratch. This operation can be extremely time-consuming if manually performed or prone to type I errors if done automatically. This is even more exacerbated if done over large geographic regions. In this manuscript we examine the issue of mapping requirements for west Tajikistan where no complete landslide inventory is available. The key question is: How many landslides should be required to develop reliable landslide susceptibility models based on statistical modeling? In fact, for such a wide and extremely complex territory, the collection of an inventory that is sufficiently detailed requires a large investment in time and human resources. However, at which point of the mapping procedure, would the resulting susceptibility model produce significantly better results as compared to a model built with less information? We addressed this question by implementing a binomial Generalized Additive Model trained and validated with different landslide proportions and measured the induced variability in the resulting susceptibility model. The results of this study are very site-specific but we proposed a very functional protocol to investigate a problem which is underestimated in the literature.

Nature-based solutions for flood risk reduction: A probabilistic modeling framework

<h2>Summary</h2> Flooding is the most frequent and damaging natural hazard globally. While nature-based solutions can reduce flood risk, they are not part of mainstream risk management. We develop a probabilistic risk analysis framework to quantify these benefits that (1) accounts for frequent small events and rarer large events, (2) can be applied to large basins and data-scarce contexts, and (3) quantifies economic benefits and reduction in people affected. Measuring benefits in terms of avoided losses enables the integration of nature-based solutions in standard cost-benefit analysis of protective infrastructure. Results for the Chindwin River basin in Myanmar highlight the potential consequences of deforestation on long-term flood risk. We find that loss reduction is driven by small but frequent storms, suggesting that current practice relying on large storms may underestimate the benefits of nature-based solutions. By providing average annual losses, the framework helps mainstream nature-based solutions in infrastructure planning or insurance practice.

Methodological and empirical challenges of SEEA EEA in developing contexts: Towards ecosystem service accounts in the Kyrgyz Republic

The System of Environmental-Economic Accounting Experimental Ecosystem Accounting (SEEA EEA) represents a crucial approach to incorporate the assessment of the sustainable use of natural resources and ecosystems into decision- and policy-making. However, its application is constrained by challenges distinct across specific implementation contexts, including those present in developing nations. In this paper, we focus on a pilot SEEA EEA application in a local-scale case study in Kyzyl Unkur, Jalal-Abad region, the Kyrgyz Republic, characterized by a unique natural walnut forest. We summarize key methodological and empirical challenges identified through collaboration with local experts and stakeholders during the compilation of Supply and Use tables for selected ecosystem services (ES) relevant at local, national and global levels. Specifically, we focus on the methodological challenges related to a) defining and assigning benefits for own consumption; b) delineating the chain of ES flows (e.g., fodder for farm animals); c) uncovering the relevance of carbon sequestration in developing nation contexts which are often minor greenhouse gas (GHG) emitters and demand for the service lies mostly beyond their boundaries. Among empirical challenges, we highlight the issues of data collection and availability. The aim of this communication is to provide lessons learnt from building SEEA EEA accounts in a developing, data-scarce context, potentially transferable to other similar applications.