Markov Models of Inertia and Dynamism on Two Contiguous Ohio Landscapes

Abstract

This study examined landcover change during 1940–1988 in two contiguous landscapes of approximately 12,400 hectares in central Ohio, USA: an inertial till plain area and a more dynamic moraine area. Agriculture dominated both landscapes throughout, however, the cover of natural vegetation and urban‐suburban development increased over time on the moraine while natural vegetation decreased and agriculture increased on the till plain. Markov process models for landcover change were constructed for three 14–17‐year time intervals and for the entire forty‐eight‐year study period. Probabilities of self‐replacement for agricultural lands (0.91–0.97) were higher and those for forests (0.54–0.65) lower than those reported for other landscapes. Predictions of landcover percentages for fifty years in the future were made using Markov process models derived from actual cover changes in 1940–1957, 1971–1988, and 1940–1988. All three models produced similar predictions for the inertial till plain area. In contrast, the three models' predictions differed considerably for the moraine area. Only the 1971–1988 transition matrix was sensitive to increases in urbanization and reforestation which began on the moraine in the 1970s. These results indicate that reliance on lengthy study intervals and net rates of change can obscure spatial and temporal patterns of landcover change that are caused by natural and socioeconomic factors operating on shorter time scales.