It is well known that ground truth data is an essential component of applying remote sensing data. At its core, remote sensing is the collection and analysis of data from distance, through the use of satellite imagery or other platforms. This can be leveraged to provide rich information that goes far beyond the visual interpretation of pictures. However, this approach also presents a significant challenge: how do we interpret readings from space as bio-physical parameters on the ground?
Herein lies the importance that ground truth may play – this set of in situ observations or measurements is frequently used to calibrate remote sensing data. Last year, one of our analysts headed to Kalimantan as part of the client’s ground truthing exercise for our oil palm health model. A major takeaway from this exercise is that oil palm “health” is a complex array of symptoms and observations, with many different methods and metrics used to assess the health status of the crop.
While we continue to improve accuracy with input from such exercises, optical or radar data still offers precise, consistent, and rapid methods of assessing conditions as proxies to health. It’s worth noting that on-site measurements are often fraught with subjectivities and uncertainties, with different methods often yielding different “true” values. Although ground assessments are effective at a local level, for our typical large-scale analysis of agricultural and environmental processes, remote sensing may be a better tool to understand global and regional phenomena.
In-situ measurements are considered critical tools for remote sensing practitioners. By combining remote sensing data with on-the-ground observations, there is general understanding that models can be improved to be more accurate and reliable. At the same time, there are some who argue that we should abandon the term “ground truth” as it has ontological issues in defining what we believe is the “true” measurement of a physical phenomenon. As remote sensing methods and capabilities are further refined, we may find ourselves reconsidering whether on-site measurements are the best and truest way to understand the conditions on the ground.
Comments