Remotely Incorrect? Accounting for Nonclassical Measurement Error in Satellite Data on Deforestation

Abstract

Research relying on remotely sensed data on land use and deforestation has exploded in recent years. While satellite-based measures have clear advantages in terms of coverage, the presence of measurement error within these products is often overlooked. Here, we detail the econometric implications of these errors when analyzing the determinants of binary measures of deforestation or forest cover. We then discuss estimators that exploit knowledge of the remote-sensing process to obtain consistent estimates. Finally, we assess our estimators via simulation and an impact evaluation of a conservation program in Mexico. We find that both geography and characteristics of the raw data can lead to systematic underreporting of deforestation. However, accounting for these sources of error, which are common across many satellite-based metrics, can limit the bias from misclassification.