Evaluation of agricultural drought in South Korea using socio-economic drought information

The effect of drought on agriculture is a complicated result of several factors. Several studies have been conducted to evaluate agricultural drought using facility monitoring or statistical data. However, the gap in awareness between the public monitoring of drought and local situations has not been sufficiently reduced. The objectives of this study were to analyze the correlation between a socioeconomic drought information (SEDI) based on Internet news articles and agricultural drought data and to determine whether the SEDI can provide reliable information regarding damage due to agricultural drought. The SEDI was subdivided into the following aspects: water deficit, water security and support, economic damage and impact, and environmental and sanitation impact. The correlation coefficient between SEDI and eight agricultural statistical datasets ranged from 0.74 to 0.89. In moderately dry irrigation and non-irrigation periods, the relationship between SEDI and the agricultural reservoir drought index (RDI) was evidenced by a receiver operating characteristic and an area under the curve of 0.60 and 0.68, respectively. The cross-correlation values between the SEDI and RDI ranged from -0.48 to -0.42. According to cross-correlation analysis, SEDI was able to detect drought one month earlier than the RDI was able to do. Thus, our proposed method can contribute to the development of an impact-based early warning system for agricultural drought management.

Automated summary

This study aims to analyze the correlation between socioeconomic drought information (SEDI) based on Internet news articles and agricultural drought data, and to determine whether SEDI can provide reliable information regarding damage due to agricultural drought. The results show that SEDI has a significant correlation with agricultural statistical datasets, and it can detect drought one month earlier than the agricultural reservoir drought index (RDI). The findings suggest that our proposed method can contribute to the development of an impact-based early warning system for agricultural drought management.