Abstract: Satellite-based image classification facilitates low-cost measurement of the Earth's surface composition. However, image classification techniques can lead to misleading conclusions about transition processes (e.g., deforestation, urbanization, and industrialization). We propose a correction for transition rate estimates based on the econometric measurement error literature to extract the signal (truth) from its noisy measurement (satellite-based classifications). No ground-level truth data is required to implement the correction. Our proposed correction produces consistent estimates of transition rates, confirmed by Monte Carlo simulations and panel validation data. In contrast, transition rates without correction for misclassifications are severely biased.
Keywords: Measurement Error, Remote-Sensing Data, Land Cover, Hidden Markov Model
JEL Classification: C13; Q15; R14