Context

Yemen, a nation at the southern tip of the Arabian Peninsula (Figure 1), boasts a remarkably diverse geography – from rugged mountains to arid plains. This varied landscape, coupled with its predominantly arid climate, profoundly shapes agriculture. Farmers have historically adapted, developing three main irrigation methods: ancient highland terraces, natural and artificial channels in wadis and river valleys, and groundwater fed irrigation.

However, rapid population growth since the 1970s, coupled with the widespread increase of diesel and, more recently, solar-powered pumps, has led to extensive groundwater exploitation. Groundwater levels have been dropping faster than they can be replenished. All this has been compounded by civil unrest since 2014 and the threat of climate change.

To address these challenges, we conducted an innovative analysis of agricultural land’s geographic context and production intensity. This crucial knowledge will help identify communities most vulnerable to water scarcity, assess the impact of groundwater depletion on food security, and enable agencies to develop targeted interventions.

What we did

We investigated nine areas of interest (AOIs) representing significant agricultural land in Yemen for analysis. For these, we combined four years of satellite data with hydrological modelling to classify agricultural land by whether it’s more likely to be terraced, spate or groundwater irrigated.

Firstly, we describe agriculture based on slope (how steep the land is) and distance to a flow network (e.g., river, stream, or seasonal waterway). If the land is very steep, it is likely to be terraced. If close to a flow network, the land likely uses spate irrigation. This is where farmers harness seasonal floods to water crops.

Secondly, production intensity identifies what’s actually growing and for how long (Figure 2). We measured the duration of green, healthy leaves—known as the ‘photosynthetically active canopy’—to do this. Crops growing for a short period (a few months) suggest cereals such as wheat or millet. Year-round green crops might include fruit trees or khat. This provides a clear picture of what farmers cultivate and their likely annual water needs.

Results

In total, we mapped agricultural productivity over 21,500 square kilometres (an area greater than the size of Wales). We showed that the total area of agriculture varied over time. Agriculture was most prevalent in 2020; known for being a particularly wet year.

Three quarters of the area we mapped had a short productive season (2-5 months). These likely include cereal crops such as millet or wheat. The remaining areas have a 9–12 month productive season. This can be associated with year-round canopy crops like fruit trees or khat.

Our evidence is feeding into groundwater models to predict how water levels in aquifers are changing over time. This type of independent information is invaluable for water resources authorities. It helps them identify areas where water depletion threatens food security and livelihoods, enabling better planning of interventions.