<![CDATA[Classical land-rent theory predicts that as returns to farming rise, cultivation climbs onto costlier terrain, and Indonesian upland policy allocates conservation and monitoring budgets largely on that premise. Whether the premise still holds in a commercial highland economy whose binding constraints lie on the input and price side rather than on land availability is an empirical question. This study tests it for the Dieng highlands of Central Java, a leading potato-producing area, using a directional prediction: if the margin is still advancing, land-surface disturbance should correlate positively with elevation and slope, whereas a zero or negative correlation would place recent adjustment inside the land already in use. A balanced panel of 5,470 grid cells of 150 m was constructed for 24 villages in Kejajar and Batur, observed annually from 2013 to 2024 through Harmonized Landsat Sentinel-2 composites. Greenness, canopy moisture, and surface exposure were combined into a relative grid-level surface-disturbance score, and its association with terrain was estimated using Spearman correlations with intervals from 2,000 resamples of 83 spatial blocks. The predicted positive association does not appear. Surface disturbance was instead negatively associated with elevation and slope, at -0.130 and -0.122 annually and -0.111 and -0.115 under seasonal matching, with the two rankings agreeing at 0.940. All ten specifications returned negative signs, although two of ten intervals for each terrain attribute crossed zero, so the robust finding is the absence of a positive gradient. The pattern is consistent with, but does not demonstrate, adjustment within land already in production. Altitude-based targeting should therefore be checked against a ranked, field-verified diagnostic before it guides scarce budgets]]>
