Geospatial and Statistical Analysis of Hill Agriculture in the Kollihills, Tamil Nadu, India A Temporal and Spatial Study (2001–2025)

Abstract : The Kollihills (Kolli Malai) are a hill range in Namakkal district, Tamil Nadu, India. They are among the most important tribal agricultural landscapes in the southern Eastern Ghats, both ecologically and culturally. The hills are mainly inhabited by the Malayali tribal community. They cover about 28,293 hectares, with elevations from 1,000 to 1,300 metres above sea level. More than half the area is used for cultivation. This study presents a detailed geospatial and statistical analysis of hill agriculture in the Kollihills from 2001 to 2025. It uses multi-year satellite remote sensing data (Landsat 5 TM, 7 ETM+, 8 OLI, 9 OLI-2), MODIS NDVI time series, Indian Census data (2001, 2011, 2021), and crop statistics from the Tamil Nadu Department of Agriculture. A five-phase methodological framework was used. It included data acquisition and pre-processing. Supervised land use and land cover classification was performed using Random Forest and Support Vector Machine algorithms. Spatial analysis involved Moran's I, Getis-Ord Gi* hotspot analysis, and kernel density estimation. Statistical testing was also part of the methodology. Four research hypotheses were formulated and tested. H1: Food crop area declined significantly over the study period. H2: Cash crop expansion shows statistically significant positive spatial autocorrelation. H3: NDVI values have a significant positive correlation with altitude gradient. H4: NDVI-based agricultural vigour varies significantly across elevation zones. Results confirm all four hypotheses at a p < 0.05 significance level. Food crop area (millets, paddy, ragi) declined by approximately 22.3% between 2001 and 2025, while cash and horticultural crops (tapioca, pineapple, guava, banana, coffee, pepper) expanded by approximately 40.7% over the same period. NDVI analysis reveals a significant positive correlation with elevation (r = 0.58, p < 0.001) and a significant NDVI divergence between forest (increasing trend, Mann-Kendall τ = +0.52) and agricultural land (declining trend, τ = -0.38). Forest cover declined from 12,454 ha to approximately 10,893 ha. One-way ANOVA confirms significant NDVI differences across three elevation zones (F = 12.4, p < 0.001). These findings carry critical implications for food security, tribal livelihood sustainability, biodiversity conservation, and agricultural policy in the hill ecosystems of South India.