Machine learning and geographic information systems-based framework for multidimensional analysis of cascading...

Area of Science:

• Environmental Science
• Agricultural Science
• Climate Science

Background:

• Drought poses significant risks to agricultural sectors globally.
• Understanding cascading impacts is crucial for sustainable water and food security.
• The Konya Closed Basin in Türkiye is a drought-sensitive agricultural region.

Purpose of the Study:

• To propose and apply a multidimensional conceptual framework for assessing drought's cascading impacts on agriculture.
• To analyze drought dynamics and land use transitions in the Konya Closed Basin.
• To identify maladaptive agricultural practices and their implications for water and food security.

Main Methods:

• Integration of remote sensing (NDVI, NDWI, LST, land cover) and ground-based data (precipitation, temperature, groundwater).
• Application of statistical trend analysis (Mann-Kendall) for drought and land use characterization.
• Utilized machine learning algorithms (SVM, ANN, RF) to model land use change based on drought indicators (NDVI, NDWI, LST, PDSI).

Main Results:

• Significant conversion of irrigated land (approx. 510,000 ha) to non-irrigated areas between 1990 and 2018.
• Continued rise in production of water-intensive crops (maize, sugar beet) despite decreasing water availability.
• Identified a maladaptive trajectory in agricultural practices, indicating unsustainable water use.

Conclusions:

• The proposed framework effectively assesses cascading drought impacts in agricultural systems.
• Unsustainable water use and production patterns pose long-term risks to water and food security.
• Findings offer critical insights for adaptive agricultural and water governance in climate-vulnerable regions.