Search

Virtually all climate monitoring and forecasting efforts concentrate on hazards rather than on impacts, while the latter are a priority for planning emergency activities and for the evaluation of mitigation strategies. Effective disaster risk management strategies need to consider the prevailing "human terrain" to predict who is at risk and how communities will be affected. There has been little effort to align the spatiotemporal granularity of socioeconomic assessments with the granularity of weather or climate monitoring. The lack of a high-resolution socioeconomic baseline leaves methodical approaches like machine learning virtually untapped for pattern recognition of extreme climate impacts on livelihood conditions. While the request for "better" socioeconomic data is not new, we highlight the need to collect and analyze environmental and socioeconomic data together and discuss novel strategies for coordinated data collection via mobile technologies from a drought risk management perspective. A better temporal, spatial, and contextual understanding of socioeconomic impacts of extreme climate conditions will help to establish complex causal pathways and quantitative proof about climate-attributable livelihood impacts. Such considerations are particularly important in the context of the latest big data-driven initiatives, such as the World Bank's Famine Action Mechanism (FAM).

IntroductionFood insecurity has a bidirectional relationship with HIV infection, with hunger driving compensatory risk behaviors, while infection can increase poverty. We used a laboratory recency assay to estimate the timing of HIV infection vis-à-vis the timing of severe food insecurity (SFI).MethodsData from population-based surveys in Zambia, Eswatini, Lesotho, Uganda, and Tanzania and Namibia were used. We defined SFI as having no food ≥three times in the past month. Recent HIV infection was identified using the HIV-1 LAg avidity assay, with a viral load (>1000 copies/ml) and no detectable antiretrovirals indicating an infection in the past 6 months. Logistic regression was conducted to assess correlates of SFI. Poisson regression was conducted on pooled data, adjusted by country to determine the association of SFI with recent HIV infection and risk behaviors, with effect heterogeneity evaluated for each country. All analyses were done using weighted data.ResultsOf 112,955 participants aged 15-59, 10.3% lived in households reporting SFI. SFI was most common in urban, woman-headed households. Among women and not men, SFI was associated with a two-fold increase in risk of recent HIV infection (adjusted relative risk [aRR] 2.08, 95% CI 1.09-3.97), with lower risk in high prevalence countries (Eswatini and Lesotho). SFI was associated with transactional sex (aRR 1.28, 95% CI 1.17-1.41), a history of forced sex (aRR 1.36, 95% CI 1.11-1.66), and condom-less sex with a partner of unknown or positive HIV status (aRR 1.08, 95% CI 1.02-1.14) in all women, and intergenerational sex (partner ≥10 years older) in women aged 15-24 (aRR 1.23, 95% CI 1.03-1.46), although this was heterogeneous. Recent receipt of food support was protective (aRR 0.36, 95% CI 0.14-0.88).ConclusionSFI increased risk for HIV acquisition in women by two-fold. Worsening food scarcity due to climactic extremes could imperil HIV epidemic control.SUMMARYWhat is already knownThe link between food insecurity and the adoption of high-risk sexual behaviors as a coping mechanism has been shown in several settings.HIV infection can also drive food insecurity due to debilitating illness reducing productivity, the costs of treatment diverting money from supplies, and potentially reduced labor migration.Food insecurity has been associated with chronic HIV infection, but it has not been linked with HIV acquisition.What are the new findingsThis study of 112,955 adults across six countries in sub-Saharan Africa provides unique information on the association between acute food insecurity and recent HIV infection in women, as well as the potential behavioral and biological mediators, including community viremia as a measure of infectiousness.The data enabled a comprehensive analysis of factors associated with risk of infection, and how these factors differed by country and gender. Women living in food insecure households had a two-fold higher risk of recent HIV acquisition, and reported higher rates of transactional sex, early sexual debut, forced sex, intergenerational sex and sex without a condom with someone of unknown or positive HIV status. This pattern was not seen in men.This study is also the first to demonstrate a protective association for food support, which was associated with a lower risk of recent HIV infection in women.What do the new findings implyIn light of worsening food insecurity due to climate change and the recent COVID-19 pandemic, our results support further exploration of gender-specific pathways of response to acute food insecurity, particularly how women’s changes in sexual behavior heighten their risk of HIV acquisition.These and other data support the inclusion of food insecurity in HIV risk assessments for women, as well as the exploration of provision of food support to those households at highest risk based on geographic and individual factors.

To support livelihoods who rely on agricultural activities against increasing climate and food insecurity risks, the World Food Programme is implementing Forecast-based Financing (FbF) for drought management in Mozambique. FbF is an approach in which humanitarian financing and anticipatory action are automatically made available based on a certain likelihood of a drought event.To enable the implementation of FbF projects, the World Food Programme has developed and implemented probabilistic seasonal forecasts of Standardized Precipitation Index (SPI) covering Mozambique’s rainfall season (October-April). The system produces forecast of the probability of the SPI to be less than -1, a threshold that identifies significant drought events at time scales of 2 and 3 months. These are derived from ECMWF ensemble seasonal daily precipitation forecasts, available monthly and processed from August to February to forecast drought occurrence one to six months ahead of time in four Mozambican districts.Operational usage of probabilistic SPI forecasts requires both the derivation of a trigger (a probability value above which drought is assumed to take place) and an assessment of forecast skill. The trigger is a probability value above which drought is assumed to take place and its exceedance leads to the implementation of anticipatory actions. Forecast skill determines if the forecast system for a specific SPI time frame is usable. Both forecast skill and triggers are derived jointly through a forecast verification analysis based on a comparison between historical time series of SPI forecasts (1993-2019) and SPI values derived from CHIRPS satellite rainfall estimates used as a reference precipitation data set.The outcomes of this analysis are compiled into manageable tables of forecast analysis that were readily applied for decision-making process in four districts in Mozambique. In addition, a preliminary cost loss analysis for some of the Forecast-based Financing interventions against droughts and food insecurity demonstrated a potential to generate large socio-economic gains for both WFP and the beneficiaries of the anticipatory actions.The goal of this abstract is to present WFP’s on-going and previous technical activities linked to the development of Forecast-based Financing projects for drought risk management to the broader scientific community. Whereas this system is being consolidated and still under review, next technical developments will comprise the better integration of hazard indicators with “impact levels” and risk metrics, adequate bias correction and benchmarking with other existing forecasting systems. Finally, WFP is committed in producing evidences that can protect livelihoods and save lives through the great window of opportunity generated by actionable forecasts.

Accurate spatial information of agricultural fields is important for providing actionable information to farmers, managers, and policymakers. On the other hand, the automated detection of field boundaries is a challenging task due to their small size, irregular shape and the use of mixedcropping systems making field boundaries vaguely defined. In this paper, we propose a strategy for field boundary detection based on the fully convolutional network architecture called ResU-Net. The benefits of this model are two-fold: first, residual units ease training of deep networks. Second, rich skip connections within the network could facilitate information propagation, allowing us to design networks with fewer parameters but better performance in comparison with the traditional U-Net model. An extensive experimental analysis is performed over the whole of Denmark using Sentinel-2 images and comparing several U-Net and ResU-Net field boundary detection algorithms. The presented results show that the ResU-Net model has a better performance with an average F1 score of 0.90 and average Jaccard coefficient of 0.80 in comparison to the U-Net model with an average F1 score of 0.88 and an average Jaccard coefficient of 0.77.

Crop type mapping represents one of the most challenging problems in remote sensing. Spatial, spectral, and temporal information are required in order to obtain a unambiguous distinction among the types of crop. This paper presents a multi-sensor approach, where labelled high-resolution images from drones, limited to small areas, are used to enhance the classification ability of machine learning models based on Sentinel 2 time series. The project described in this paper is organized into three major activities. The first part focused on the exploitation of RGB drone images by using transfer learning and convolutional networks, and it has already been described in a previous work by the team. The second part deals with preliminary analysis of multi-spectral Sentinel 2 time-series using the labelled data from the drones campaign and trees-based machine learning algorithms. Finally, the third ongoing part deals with the combination of drones and satellite data in order to show how drones data can help the Sentinel 2 classification by reducing the effort needed to collect reference crop type information.

Background: We assessed the associations between severe food insecurity (SFI) and HIV risk in six sub-Saharan African countries. Methods: Data from nationally representative Population-based HIV Impact Assessments (PHIAs) in Zambia, Eswatini, Lesotho, Uganda, and Tanzania and Namibia were used. SFI was defined as having no food in the house at least three times in the past month. Recent HIV infection was identified using the HIV-1 LAg avidity assay, with viral load (>1000 copies/ml) and antiretroviral data. Logistic regression was conducted to assess correlates of SFI. Poisson regression was conducted on pooled data, stratified by sex and adjusted by country, urbanicity, wealth quintile and age, to determine the association of SFI with recent HIV infection and risk behaviors. All analyses were done using weighted data. Findings: Of the 112,955 survey participants aged 15-59, 10·3% lived in households reporting SFI. SFI was most common in poor, urban, woman-headed households with many dependents. Among women, SFI was associated with a two-fold increase in risk of recent HIV infection (adjusted relative risk [aRR] 2·04, 95% CI 1·04-3·98), and receipt of food support was protective (aRR 0·36, 95% CI 0·14-0·91), associations not observed among men. SFI was also associated with transactional sex (aRR 1·29, 95% CI 1·17-1·41), a history of forced sex (aRR 1·42, 95% CI 1·16-1·74), and condom-less sex with a partner of unknown or positive HIV status (aRR 1·08, 95% CI 1·02-1·14) in all women, and intergenerational sex (sexual partner ≥10 years older) in women aged 15-24 (aRR 1·23, 95% CI 1·03-1·46). Interpretation: The findings demonstrate that food insecurity increases risk for HIV acquisition. Worsening food scarcity due to climactic extremes could imperil HIV epidemic control, while food support programs might mitigate this risk. Funding Information: President’s Emergency Plan for AIDS Relief (PEPFAR) through the US Centers for Disease Control and Prevention under the terms of cooperative agreement #U2GGH001226. Declaration of Interests: The authors declare that they have no conflicts of interest. Ethics Approval Statement: Written informed consent/assent was documented via electronic signature, with witnesses verifying consent for illiterate individuals. The PHIA protocol and data collection tools were approved by national ethics committees for each country, and the institutional review boards at Columbia University Irving Medical Center, the US Centers for Disease Control and Prevention (CDC) and the University of California, San Francisco in the case of Namibia.

In this chapter, we present an overview of the role of satellite-based rainfall estimates (SREs) in drought risk management applications, ranging from simple anomaly and index-based approaches to cross-cutting drought early warning systems (EWS) and financial instruments such as weather index-based insurance (WII) schemes. We contend that meteorological, hydrological, agricultural, and socioeconomic are aspects – not types – of drought, and a universally acceptable drought definition is not a prerequisite for the effective and efficient assessment of the impacts of drought using SREs and other satellite-based datasets and/or models. This is illustrated through examples from the work of the co-authors, as well as the wider community. The chapter concludes with a synthesis of the challenges for SREs and the current trends in the development and application of SREs in drought risk management, including an outlook of the priorities for future research and applications.

Crop type mapping currently represents an important problem in remote sensing. Accurate information on the extent and types of crops derived from remote sensing can help managing and improving agriculture especially for developing countries where such information is scarce. In this paper, high-resolution RGB drone images are the input data for the classification performed using a transfer learning (TL) approach. VGG16 and GoogLeNet, which are pre-trained convolutional neural networks (CNNs) used for classification tasks coming from computer vision, are considered for the mapping of the crop types. Thanks to the transferred knowledge, the proposed models can successfully classify the studied crop types with high overall accuracy for two considered cases, achieving up to almost 83% for the Malawi dataset and up to 90% for the Mozambique dataset. Notably, these results are comparable to the ones achieved by the same deep CNN architectures in many computer vision tasks. With regard to drone data analysis, application of deep CNN is very limited so far due to high requirements on the number of samples needed to train such complicated architectures. Our results demonstrate that the transfer learning is an efficient way to overcome this problem and take full advantage of the benefits of deep CNN architectures for drone-based crop type mapping. Moreover, based on experiments with different TL approaches we show that the number of frozen layers is an important parameter of TL and a fine-tuning of all the CNN weights results in significantly better performance than the approaches that apply fine-tuning only on some numbers of last layers.

The increasing frequency and severity of extreme climatic events and their impacts are being realized in many regions of the world, particularly in smallholder crop and livestock production systems in Sub-Saharan Africa (SSA). These events underscore the need for timely early warning. Satellite Earth Observation (EO) availability, rapid developments in methodology to archive and process them through cloud services and advanced computational capabilities, continue to generate new opportunities for providing accurate, reliable, and timely information for decision-makers across multiple cropping systems and for resource-constrained institutions. Today, systems and tools that leverage these developments to provide open access actionable early warning information exist. Some have already been employed by early adopters and are currently operational in selecting national monitoring programs in Angola, Kenya, Rwanda, Tanzania, and Uganda. Despite these capabilities, many governments in SSA still rely on traditional crop monitoring systems, which mainly rely on sparse and long latency in situ reports with little to no integration of EO-derived crop conditions and yield models. This study reviews open-access operational agricultural monitoring systems available for Africa. These systems provide the best-available open-access EO data that countries can readily take advantage of, adapt, adopt, and leverage to augment national systems and make significant leaps (timeliness, spatial coverage and accuracy) of their monitoring programs. Data accessible (vegetation indices, crop masks) in these systems are described showing typical outputs. Examples are provided including crop conditions maps, and damage assessments and how these have integrated into reporting and decision-making. The discussion compares and contrasts the types of data, assessments and products can expect from using these systems. This paper is intended for individuals and organizations seeking to access and use EO to assess crop conditions who might not have the technical skill or computing facilities to process raw data into informational products.

Global and regional scale agricultural monitoring systems aim to provide up-to-date information regarding food production to different actors and decision makers in support of global and national food security. To help reduce price volatility of the kind experienced between 2007 and 2011, a global system of agricultural monitoring systems is needed to ensure the coordinated flow of information in a timely manner for early warning purposes. A number of systems now exist that fill this role. This paper provides an overview of the eight main global and regional scale agricultural monitoring systems currently in operation and compares them based on the input data and models used, the outputs produced and other characteristics such as the role of the analyst, their interaction with other systems and the geographical scale at which they operate. Despite improvements in access to high resolution satellite imagery over the last decade and the use of numerous remote-sensing based products by the different systems, there are still fundamental gaps. Based on a questionnaire, discussions with the system experts and the literature, we present the main gaps in the data and in the methods. Finally, we propose some recommendations for addressing these gaps through ongoing improvements in remote sensing, harnessing new and innovative data streams and the continued sharing of more and more data.

Governments, aid organizations and people affected by drought are struggling to mitigate the resulting impact on both water resources and crops. In this paper we focus on improved decision-support for agricultural droughts that threaten the livelihoods of people living in vulnerable regions. We claim that new strategic partnerships are required to link scientific findings to actual user requirements of governments and aid organizations and to turn data streams into useful information for decision-support. Furthermore, we list several promising approaches, ranging from the integration of satellite-derived soil moisture measurements that link atmospheric processes to anomalies on the land surface to improved long-range weather predictions and mobile applications. The latter can be used for the dissemination of relevant information, but also for validating satellite-derived datasets or for collecting additional information about socio-economic vulnerabilities. Ideally, the consequence is a translation of early warning into local action, strengthening disaster preparedness and avoiding the need for large-scale external support.