Remote sensing data on deforestation can be misleading

A recent paper titled “Remotely Incorrect? Accounting for Nonclassical Measurement Error in Satellite Data on Deforestation,” authored by Jennifer Alix-Garcia and Daniel L. Millimet, explores the limitations and errors associated with remote sensing methods used to study deforestation. The paper, published in the Journal of the Association of Environmental and Resource Economists, highlights the importance of understanding and addressing these errors to ensure accurate research outcomes.

Deforestation poses significant ecological challenges, including habitat loss and the risk of mass extinction. To gain insights into global land use and the threats posed by diminishing forest cover, researchers and stakeholders have increasingly relied on remote sensing data, facilitated by the numerous satellites orbiting Earth today.

While remote sensing methods offer considerable advantages, such as global coverage, the authors stress that errors in remotely sensed measurements are likely to occur. These errors can manifest at various stages of the data acquisition process. For example, the optical sensors employed by satellite systems to measure reflected energy have inherent technical limitations. Additionally, errors may arise during image pre-processing, which involves correcting distortions in natural features, and in the algorithms used to convert reflectance values into usable numerical data.

To illustrate these misclassifications, the authors analyze two different remote sensing data sources depicting forest cover in Mexico. Although these sources capture the same location at a similar time, the resulting data differ due to variations in pre-processing techniques. Differences in the measurement of forest presence between the two sources depend on factors like extreme topographical attributes (e.g., slope and elevation) and sensor characteristics.

Based on their findings, the authors propose alternative estimation methods that can account for these misclassifications. They apply these estimators to evaluate a specific scenario in Mexico—a payment for ecosystem services (PES) program that compensates landowners for preserving intact forest cover, monitored in part using remote sensing. Analyzing data from 2003 to 2015, the authors discover that once misclassifications are addressed, the PES program significantly slows down deforestation.

In conclusion, the authors emphasize the importance of interdisciplinary conversations to comprehend how remote sensing data is constructed and to avoid using simplistic statistical models that fail to consider nonclassical measurement errors. They advocate for collaboration between researchers and remote sensing scientists to gain a comprehensive understanding of data construction processes and their limitations.

The findings of this study are available in the Journal of the Association of Environmental and Resource Economists.

Source: University of Chicago

Leave a Comment