There is scarce evidence on whether and how assistance in humanitarian emergencies and conflict settings impacts household well-being and behaviour. Conducting rigorous impact evaluations in such settings poses multiple challenges in design and data collection. In SEEDS, we evaluate the impact of a complex large-scale multi-arm agricultural intervention on productivity, food security, and resilience in the context of an on-going humanitarian crisis in Syria. Specifically, we identify the causal impacts of agricultural asset transfers over various time horizons (the short-, medium-, and long-run), and across different conditions and subgroups (gender and conflict intensity) at the household-level. We evaluate the effectiveness of irrigation rehabilitation separately at the community-level. We use and combine various data sources, including a unique survey panel dataset collected over a period of four years from multiple governorates in Syria, satellite remote-sensing data, and publicly available violent conflict incidence and weather data. Our findings from using cutting-edge machine and deep learning approaches together with innovative balancing and analytical methods can be summarised as follows: For average treatment effects at the household-level, we find that the provision of agricultural asset support leads to significant improvements in food security in the short- and long-term, three years after the intervention. The positive and significant effect on food security is driven mainly by the increased consumption of healthy food items such as vegetables. In the long-run, livestock support reduces the use of harmful coping strategies households employ to deal with food shortages. Interestingly, we find that households who received vegetable kits are not just less likely to sell their productive assets in the long-term but also are less likely to marry off their young daughters or send their children to work. Overall, we find that both agricultural and livestock asset support is key to improving households’ resilience in the long-term. The irrigation rehabilitation interventions at the community-level positively affected agricultural productivity compared to the pre-intervention and pre-conflict periods. However, these effects were only significantly pronounced in the spring season. As for the heterogeneity analysis, we find that female-headed households benefit remarkably more in terms of food security in the medium-term compared to male-headed families. Moreover, households residing in areas that are moderately affected by violent conflict show stronger food security improvements compared to households from peaceful or conflict-intense settings. Overall, we draw three overarching lessons from our findings in SEEDS: First, agricultural support in protracted conflict settings effectively improves the long-term welfare and resilience of vulnerable households. In fact, the presence of an ongoing humanitarian operation acts as a social safety net if circumstances deteriorate suddenly. Second, not all interventions are equally effective, and not all households equally benefit, underscoring the need to design and implement inclusive context-specific interventions with detailed targeting. Third, methodologically, using multiple remote data sources and machine learning methods help overcome challenges in conducting rigorous impact evaluations in hard-to-reach humanitarian emergency settings.
Recent literature shows increasing interest in analyzing causes of what was referred to as “unusual” fires in Iraq and Syria in 2019. Here, we examine the causes of uncontrolled and irregular fires within farmlands in parts of the two countries in 2019–2020 and quantify their extent spatiotemporally using a combination of medium and coarse-resolution satellite imagery, land cover, precipitation, temperature data, and multiple sources of armed violence data. Our analysis reveals the extent of the deliberate arson of farmland in Iraq and Northeast Syria in 2019–2020. In comparison, only a few fires of farmland were observed in 2007–2013 in Iraq, and Al-Hassakeh, Syria, that became a stronghold of the Islamic State (IS) in 2013. In 2014–2017, we find evidence for increased farmland fires in various locations that closely tracked the military withdrawal of IS between 2015/2016 and 2017 when it lost most of its territory. We find that although the burned area in wheat-producing regions of Iraq (Ninewa and Kirkuk) and Syria have recently increased, the increase in the percentage burned area of agricultural lands in Iraq’s Ninewa and Kirkuk was not unprecedented. MODIS overestimates burned areas when in low fire activity while it underestimates it when fire activity is high, compared to a Landsat–Sentinel-2 combination. A significant positive relationship (r = 0.83) exists between the number of IS-related incidents and the percent burned agricultural area during 2019, which raises questions about the future strategy of the IS terror group and its use and targeting of the water-food complex.
Runoff modelling is a crucial element in hydrologic sciences. However, a global runoff database is not currently available at a resolution higher than 0.1°. We use the recently developed Global Curve Number dataset (GCN250) to develop a dynamic runoff application (2015 – present) and that can be accessed via a Google Earth Engine application. We also provide a global mean monthly runoff dataset for April 2015-2021 in GeoTIFF format at a 250-meter resolution. We utilize soil moisture and GPM rainfall to dynamically retrieve the appropriate curve number and generate the corresponding runoff anywhere on Earth. Mean annual global runoff ratio results for 2021 were comparable to the runoff ratio from the Global Land Data Assimilation System (0.079 vs. 0.077, respectively). Mean annual global runoff from GCN and GLDAS were within 11% each other for 2020–2021 (0.18 vs. 0.16 mm/day, respectively). The GCN250 runoff application and the dataset are useful for many water applications such hydrologic design, land management, water resources management, and flood risk assessment.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.