Researchers at the University of Debrecen's Faculty of Agriculture, Food Science and Environmental Management and the UD Agricultural Research Institutes and Agricultural Economics are working on the monitoring of agricultural and hydrological drought trends and the development of a monitoring and forecasting system. The results could help develop effective drought mitigation plans. Details of the research have been published in the prestigious scientific journal Computers and Electronics in Agriculture.

Drought prediction is one of the biggest challenges for climate scientists and hydrologists. Researchers from the Institute of Earth Utilization, Engineering and Precision Technology (EET) of the Faculty of Agriculture, Food and Environmental Sciences (MÉK) at the University of Debrecen (DE) and the Agricultural Research Institutes and Tangible Industries (AKIT) of UD have been monitoring drought patterns in the Eastern Mediterranean basin, including Syria.

- The Mediterranean region is one of the most critical areas for climate change, where winter precipitation has fallen by up to 40 percent. Increasingly intense droughts will have a significant impact on groundwater levels and the water supply for reservoirs and lakes. Water scarcity can have negative impacts on several economic sectors, wild biodiversity and agricultural productivity, especially in countries that rely heavily on rainfed agriculture," Safwan Mohammed, a researcher at the Visual Crop Experimental Station of DE AKIT, who started his research as a PhD student at the Kálmán Kerpely PhD School of DE MÉK, told

The researchers used machine learning (Mashine Learning-ML) algorithms to develop a model system to support the prediction of agricultural and hydrological droughts

- ML methods are now widely used in areas such as flood forecasting, dust pollution determination, soil and landscape modelling, and landslide susceptibility assessment. These algorithms outperform traditional statistical methods. In addition, ML systems can handle large data sets and provide more accurate results. In this study, we used the Serial Peripheral Interface (SPI) index for drought computation and four machine learning algorithms (BG, RSS, RF and RT) for drought prediction," Safwan Mohammed explained.

For the studies, rainfall data were collected from the Syrian Meteorological Service, but the information from the 15 stations was not chronologically coordinated.

- There is still scarce literature on the drought in the Eastern Mediterranean, and the current conflict in Syria has caused disruption to many monitoring and measuring stations in the country. The use of ML algorithms could be one solution to fill in and bridge the gaps in climate descriptions based on historical data. Our results show that the use of algorithms improves the accuracy of drought research prediction and this can be further improved by using additional climate components such as soil evaporation and soil moisture data," the DE AKIT researcher added.

Measurements suggest that the Mediterranean basin is one of the world's climate 'hotspots' where even more severe consequences of climate change are expected. Both meteorological and hydrological forecasts suggest that drought events in this region will become more frequent and severe in the future. Thus a drier future climate is predicted for Syria as a whole, which will have an impact on the agricultural sector.

Experts will also use the experience gained from the development of the monitoring and forecasting system to increase the resilience of Hungarian agriculture to climate change. The researchers will also share their professional knowledge and results with the students of the Kálmán Kerpely Doctoral School of the DE, thus helping them to advance their professional development.

The details and results of the research have been published in the journal Computers and Electronics in Agriculture, and the paper by Safwan Mohammed has won the Publication Award of the Gróf Tisza István Foundation for the University of Debrecen.

Press Office - ÉE