Forecasting transitions in the state of food security with machine learning using transferable features

Impact-based seasonal rainfall forecasting to trigger early action for droughts

The Horn of Africa faces an ongoing multi-year drought due to five consecutive failed rainy seasons, a novel climatic event with unpreceded impacts. Beyond the starvation of millions of livestock, close to 23 million individuals in the region are currently facing high food insecurity in Kenya, Somalia and Ethiopia alone. The severity of these impacts calls for the urgent upscaling and optimisation of early action for droughts. However, drought research focuses mainly on meteorological and hydrological forecasting, while early action triggered by forecasts is seldom addressed. This study investigates the potential for early action for droughts by using seasonal forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF) SEAS5 system for the March-April-May (MAM) and October-November-December (OND) rainy seasons. We show that these seasonal rainfall forecasts reflect major on-the-ground impacts, which we identify from drought surveillance data from 21 counties in Kenya. Subsequently, we show that the SEAS5 drought forecasts with short lead times have substantial potential economic value (PEV) when used to trigger action before the OND season across the region (PEVmax = 0.43). Increasing lead time to one or two months ahead of the season decreases PEV, but the benefits persist (PEVmax = 0.2). Outside of Kenya, MAM forecasts have limited value. The existence of opportunities for early action during the OND season in Kenya and Somalia is demonstrated by high PEV values, with some regions recording PEVmax values close to 0.8. To illustrate the practical value of this research, we point to a dilemma that a pastoralist in the Kenyan drylands faces when deciding whether to adopt early livestock destocking. This study underscores the importance to determine the value of early actions for forecast users with different action characteristics, and to disseminate this value alongside the standard forecasts themselves. This allows users to trigger effective actions before drought impacts develop.

On the forecastability of food insecurity

Food insecurity, defined as the lack of physical or economic access to safe, nutritious and sufficient food, remains one of the main challenges included in the 2030 Agenda for Sustainable Development. Near real-time data on the food insecurity situation collected by international organizations such as the World Food Programme can be crucial to monitor and forecast time trends of insufficient food consumption levels in countries at risk. Here, using food consumption observations in combination with secondary data on conflict, extreme weather events and economic shocks, we build a forecasting model based on gradient boosted regression trees to create predictions on the evolution of insufficient food consumption trends up to 30 days in to the future in 6 countries (Burkina Faso, Cameroon, Mali, Nigeria, Syria and Yemen). Results show that the number of available historical observations is a key element for the forecasting model performance. Among the 6 countries studied in this work, for those with the longest food insecurity time series, that is Syria and Yemen, the proposed forecasting model allows to forecast the prevalence of people with insufficient food consumption up to 30 days into the future with higher accuracy than a naive approach based on the last measured prevalence only. The framework developed in this work could provide decision makers with a tool to assess how the food insecurity situation will evolve in the near future in countries at risk. Results clearly point to the added value of continuous near real-time data collection at sub-national level.

An integrated approach of remote sensing and geospatial analysis for modeling and predicting the impacts of climate change on food security

The agriculture sector provides the majority of food supplies, ensures food security, and promotes sustainable development. Due to recent climate changes as well as trends in human population growth and environmental degradation, the need for timely agricultural information continues to rise. This study analyzes and predicts the impacts of climate change on food security (FS). For 2002-2021, Landsat, MODIS satellite images and predisposing variables (land surface temperature (LST), evapotranspiration, precipitation, sunny days, cloud ratio, soil salinity, soil moisture, groundwater quality, soil types, digital elevation model, slope, and aspect) were used. First, we used a deep learning convolutional neural network (DL-CNN) based on the Google Earth Engine (GEE) to detect agricultural land (AL). A remote sensing-based approach combined with the analytical network process (ANP) model was used to identify frost-affected areas. We then analyzed the relationship between climatic, geospatial, and topographical variables and AL and frost-affected areas. We found negative correlations of - 0.80, - 0.58, - 0.43, and - 0.45 between AL and LST, evapotranspiration, cloud ratio, and soil salinity, respectively. There is a positive correlation between AL and precipitation, sunny days, soil moisture, and groundwater quality of 0.39, 0.25, 0.21, and 0.77, respectively. The correlation between frost-affected areas and LST, evapotranspiration, cloud ratio, elevation, slope, and aspect are 0.55, 0.40, 0.52, 0.35, 0.45, and 0.39. Frost-affected areas have negative correlations with precipitation, sunny day, and soil moisture of - 0.68, - 0.23, and - 0.38, respectively. Our findings show that the increase in LST, evapotranspiration, cloud ratio, and soil salinity is associated with the decrease in AL. Additionally, AL decreases with a decreasing in precipitation, sunny days, soil moisture, and groundwater quality. It was also found that as LST, evapotranspiration, cloud ratio, elevation, slope, and aspect increase, frost-affected areas increase as well. Furthermore, frost-affected areas increase when precipitation, sunny days, and soil moisture decrease. Finally, we predicted the FS threat for 2030, 2040, 2050, and 2060 using the CA-Markov method. According to the results, the AL will decrease by 0.36% from 2030 to 2060. Between 2030 and 2060, however, the area with very high frost-affected will increase by about 10.64%. In sum, this study accentuates the critical impacts of climate change on the FS in the region. Our findings and proposed methods could be helpful for researchers to model and quantify the climate change impacts on the FS in different regions and periods.

Probabilistic forecasting of remotely sensed cropland vegetation health and its relevance for food security

In a world where climate change, population growth, and global diseases threaten economic access to food, policies and contingency plans can strongly benefit from reliable forecasts of agricultural vegetation health. To inform decisions, it is also crucial to quantify the forecasting uncertainty and prove its relevance for food security. Yet, in previous studies both these aspects have been largely overlooked. This paper develops a methodology to anticipate the agricultural Vegetation Health Index (VHI) while making the underlying prediction uncertainty explicit. To achieve this aim, a probabilistic machine learning framework modelling weather and climate determinants is introduced and implemented through Quantile Random Forests. In a second step, a statistical link between VHI forecasts and monthly food price variations is established. As a pilot implementation, the framework is applied to nine countries of South-East Asia (SEA) with consideration of national monthly rice prices. Model benchmarks show satisfactory accuracy metrics, suggesting that the probabilistic VHI predictions can provide decision-makers with reliable information about future cropland health and its impact on food price variation weeks or even months ahead, albeit with increasing uncertainty as the forecasting horizon grows. These results - ultimately allowing to anticipate the impact of weather shocks on household food expenditure - contribute to advancing the multidisciplinary literature linking vegetation health, probabilistic forecasting models, and food security policy.

Machine learning can guide food security efforts when primary data are not available

Estimating how many people are food insecure and where they are is of fundamental importance for governments and humanitarian organizations to make informed and timely decisions on relevant policies and programmes. In this study, we propose a machine learning approach to predict the prevalence of people with insufficient food consumption and of people using crisis or above-crisis food-based coping when primary data are not available. Making use of a unique global dataset, the proposed models can explain up to 81% of the variation in insufficient food consumption and up to 73% of the variation in crisis or above food-based coping levels. We also show that the proposed models can nowcast the food security situation in near real time and propose a method to identify which variables are driving the changes observed in predicted trends-which is key to make predictions serviceable to decision-makers.

Do Livelihood Strategies Affect the Livelihood Resilience of Farm Households in Flooded Areas? Evidence From Hubei Province, China

Hubei Province, located in the middle reaches of the Yangtze River, is a complex area of fragile ecological environment and traditional agricultural production in China. With the further intensification of the impact of global warming, flood disasters have brought a more severe threat to the sustainable development of farmers’ livelihoods. This paper therefore examines the livelihood resilience of farmers with different livelihood strategies in the region by constructing a livelihood resilience evaluation system based on three target levels: buffering capacity, Adaptation and restoration, and using a contribution model to identify the main contributing factors affecting the livelihood resilience of fa rmers. The following three conclusions were found: (1). The overall level of livelihood resilience of farmers in flood-affected areas in Hubei Province is not high, and the difference in livelihood resilience indices between farmers with different livelihood strategies is large; (2). Farming-led farmers and part-time balanced farmers can better adapt to external shocks brought about by floods; (3). The main contributing factors affecting the livelihood resilience of various types of farmers have Convergence.

Geospatial and Machine Learning Regression Techniques for Analyzing Food Access Impact on Health Issues in Sustainable Communities

Food access is a major key component in food security, as it is every individual’s right to proper access to a nutritious and affordable food supply. Low access to healthy food sources influences people’s diet and activity habits. Guilford County in North Carolina has a high ranking in low food security and a high rate of health issues such as high blood pressure, high cholesterol, and obesity. Therefore, the primary objective of this study was to investigate the geospatial correlation between health issues and food access areas. The secondary objective was to quantitatively compare food access areas and heath issues’ descriptive statistics. The tertiary objective was to compare several machine learning techniques and find the best model that fit health issues against various food access variables with the highest performance accuracy. In this study, we adopted a food-access perspective to show that communities that have residents who have equitable access to healthy food options are typically less vulnerable to health-related disasters. We propose a methodology to help policymakers lower the number of health issues in Guilford County by analyzing such issues via correlation with respect to food access. Specifically, we conducted a geographic information system mapping methodology to examine how access to healthy food options influenced health and mortality outcomes in one of the largest counties in the state of North Carolina. We created geospatial maps representing food deserts—areas with scarce access to nutritious food; food swamps—areas with more availability of unhealthy food options compared to healthy food options; and food oases—areas with a relatively higher availability of healthy food options than unhealthy options. Our results presented a positive correlation coefficient of R2 = 0.819 among obesity and the independent variables of transportation access, and population. The correlation coefficient matrix analysis helped to identify a strong negative correlation between obesity and median income. Overall, this study offers valuable insights that can help health authorities develop preemptive preparedness for healthcare disasters.

Managing the Energy Transition through Discourse. The Case of Poland

Since energy transition has become a necessity, many countries are developing decarbonization plans and looking for ways to reduce CO2 emissions. The process takes different forms, most often being affected by local conditions, which means that it takes place in each country differently, due to the country’s social and economic specifics. One of the countries that has just started its energy transition process is Poland. The case of Poland is extremely interesting, as it is a country whose economy is still heavily dependent on coal and there are very strong traditions related to this particular source of energy. On the other hand, Poland has two very good reasons to move away from coal. The complexity and comprehensiveness of the energy transition results, among other things, from the diversity of actors involved in this issue and the multiplicity of their, often conflicting, interests. Using a critical discourse analysis, this paper attempts to show how those actors attempt to shape the transition through the discourses which they generate. The study covers the media discourse that took place in Poland between September 2020 and May 2021, i.e., a time when the Polish government negotiated with representatives of the miners’ trade unions the terms of Poland’s transition from coal. A critical analysis of the texts that appeared in the media during this time provides a picture of how different groups of social actors perceive the transition and how they try to manage that process.