Stolen Strata: Quantifying the Anthropogenic Erasure of Kashmir's Karewa Terraces and Its Threat to the Saffron Economy

SAKSHI D. MASKE 

Karewa terraces are elevated, flat-topped deposits of intermontane lake-basin age (Pleistocene/Pliocene), and they underlie one of Kashmir's most economically significant agricultural systems — the saffron-growing belt at Pampore. Everyone who follows agriculture in Kashmir already knows these terraces are vanishing under unregulated soil mining and uncontrolled urban expansion; there's ample reporting on it. What was missing was a number. So I built a systematic, multi-decadal pipeline from satellite observations to get one. From TPI and slope-threshold terrain analysis, I identify 201 karewa terraces (3,305.3 ha) across the central Kashmir Valley, and estimate the bare-ground fraction of each using season-matched Landsat and Sentinel-2 composites at four dates — 1994, 2005, 2015, and 2025. Over that period, mean bare-earth fraction (BEF) rose from 1.84% to 8.43%, and nearly all of that rise happened in the past ten years. This isn't slow, steady erosion — it's hot-spotted: 190.3 hectares of terrace surface have converted to bare earth, and 67% of that loss sits within just 12.4% of the terraces. A saffron-signature index, built on the crop's inverted phenology, flags 14 potential saffron-cultivating terraces; 43% of them sit within 1 km of an already-degraded terrace — none overlap directly, they're near degraded ground rather than on it, at least as far as the data show. At official 2024-25 yield and price figures, that at-risk segment represents roughly Rs 17.8 crore in annual production value — about 55% of the total annual production value this study traces to the area. The pattern holds even more strongly for settlements than for roads: degraded terraces sit significantly closer to roads than intact terraces (U = 1611.0, p = 0.0116), and even closer to settlements (U = 1176.0, p = 0.0001 — the strongest effect in this study). So this looks like an accessibility-driven mining pattern, not a random one. Then there's the legal question most satellite studies like this skip: is there anything to stop this excavation? No. There is presently no law protecting karewa land in Jammu & Kashmir; a private member's bill titled the 'Karewa Protection Authority' has been pending for a long time. I ran threshold-sensitivity sweeps, a resolution-matching robustness test against the Landsat/Sentinel-2 pixel-size mismatch, and effect-size and multiple-comparison corrections across all four statistical tests — none of it changes the basic picture, though a modest degree of resolution-dependence shows up in the size of the post-2015 acceleration. Combine the geomorphology, the economics, and the legal void, and this stops being a change-detection exercise. It becomes a landform loss with a price tag and a policy lever attached.