Geographia Technica, Vol 20, Issue 2, 2025, pp. 253-268
SPATIO-TEMPORAL MODELING FOR THE ANALYSIS OF HYDROLOGICAL DROUGHT AND ITS IMPACT ON RICE PRODUCTION IN THE UPPER BENGAWAN SOLO BASIN, CENTRAL JAVA, INDONESIA
Santhyami SANTHYAMI
, Jumadi JUMADI
, Kuswaji Dwi PRIYONO
, Triastuti RAHAYU
, Dewi Novita SARI
, Murnira OTHMAN
, Rudiyanto RUDI
ABSTRACT: Hydrological drought is a climate-induced disaster that directly impacts the agricultural sector, particularly rice production. This study aims to model drought in a spatial-temporal context and analyse its impact on rice production in the Upper Bengawan Solo River Basin, Central Java, Indonesia, over the period 2017–2024. The analysis was conducted using Geographic Information Systems (GIS) based on Sentinel-2A satellite imagery, annual rainfall data, and rice production records. Drought severity was quantified using the Normalised Difference Drought Index (NDDI). The results of the drought modelling were validated through correlation and regression analyses with rainfall data and the extent of drought-affected areas. Meanwhile, the impact of drought on rice production was assessed using non-parametric analysis via the LOWESS method. The findings indicate that the spatial-temporal approach is effective in identifying drought distribution and trends. Spatially, severe drought occurred in Wonogiri Regency, covering up to 1,203,014.20 hectares, while temporally, the peak occurred in 2018 with a drought area of 571,438.60 hectares. Validation tests revealed a strong positive correlation between NDDI values and drought extent (r = 0.84), and a negative correlation between NDDI and rainfall (r = -0.74), indicating that higher NDDI values correspond with wider drought-affected areas and lower rainfall. Linear regression analysis confirmed NDDI as a significant indicator for drought monitoring, with a coefficient of determination R² = 0.706, suggesting that 70.6% of the variance in drought area can be explained by NDDI, and a statistically significant p-value (p = 0.009, p < 0.05). Moreover, LOWESS analysis showed a non-linear (U-shaped) relationship between NDDI and rice production, with the highest yields at low NDDI values (2.42–2.44 million tons), declining at medium NDDI levels (~2.20 million tons), and rising again at high NDDI values (2.35 million tons). This pattern suggests that the impact of drought on rice production is not linear and is likely influenced by additional factors such as irrigation infrastructure and crop management practices. Overall, this study affirms that satellite-based spatial-temporal modelling is an effective approach for analysing hydrological drought and understanding its implications for agricultural productivity.
Keywords: Spatio-temporal modelling; Hydrological drought; Sentinel-2A satellite imagery; Rice production.