Cameron Windham from Apollo Mapping discusses the the findings of a recent paper published in Nature about a sun-sensor geometry arrangement being the cause for historically high readings of green-ness in the rainforest canopy. Windham notes that this is somewhat significant in the remote sensing field for two reasons: the first is that it undoes the supporting conclusion for the historic readings (that sunlight was the limiting factor to green-ness of the canopy rather than water availability and the lack of cloud cover facilitated more light hitting the leaves); the second is that it provokes some serious thought and hopefully discussion from the workers of the field; optical illusions may play a part in many of the readings taken for granted.
The rainforest has long provided a major source of seemingly counter-intuitive information to the remote sensing community. Workers have maintained for some time that, as evidenced by a growing body of spectral data, the relative green-ness of the canopy is more intense in dry periods than in wet times. But these findings were in direct conflict with some ground-truthed spectral studies, much to the confusion of the field. The conclusions to justify the remote measurements tended to focus on light, rather than water, as being the major limiting factor in the green-ness of the vegetation, and those conclusions have stood, until recently. A paper published in Nature this quarter has provided a new answer to the question of high dry-time green values, and the answer may surprise some: magic.
Or, rather, an optical illusion. Indeed, much as the magician would fool us into perceiving the incredible or seemingly impossible by tricking our vision, Morton et. al. propose that the differing sun-sensor geometries at play during the dry and wet seasons are responsible for the values as a result of the seasonality of the measurements. The most basic way to state the relationship is that during the dry season, the sun and the sun-synchronous sensor are on the same side of the scene, while in the wet season they are opposed. This produces a shadowing effect that under-represents the true spectral reflectance of the scene. In order to very simply illustrate what this looks like, I’d like you to imagine a bowling lane.
First, imagine looking down a bowling lane at a set of 10 pins, and imagine they are being lit by a set of lights shining in the same direction that you are looking. In this case, the sides of the pins that you see are well-lit, and the shadows the pins cast are out of sight in the background. Now, imagine a second scenario wherein the pins are instead lit from the other side, so that the pins stand between you and the lights; in this case, the faces of the pins that you see are in shadow. Now, if you replace the bowling pins for trees in the rainforest and the lights for the sun, you have a very good idea of what this paper is proposing – namely that the trees are shadowing themselves during the wet periods, which in turn decreases the reflected green values recorded by the sensor; and that sun-sensor geometry is the basic cause of this.
These findings could well have broad implications across the science of remote sensing and multi-spectral observation. For instance, these basic geometric principles hold across scale, and could therefore be influencing our observations of crops such as corn, or affecting the spectral signature of hummocky or hilly terrain. What other examples can you think of in which these geometries may be affecting our understanding of remotely-sensed data?
Morton, Douglas C., Nagol, Jyoteshwar, Carabajal, Claudia C., Rosette, Jacqueline, Palace, Michael, Cook, Bruce D., Vermote, Eric F., Harding, David J., North, Peter R. J., “Amazon forests maintain consistent canopy structure and greenness during the dry season”, Nature, 2014, vol. 506, pp. 221–224
- The New Boom: Remotely Tracking Growth of the Sugarcane Industry in São Paulo, Brazil
- The ASTER Spectral Library – a True Monument of Remote Sensing Data
- ASTER and the Spectral Recipe for Finding Gold
- Shadows and Angles: Measuring Object Heights from Satellite Imagery
- The Role of High-Resolution Imagery and GIS in Flood Relief Efforts in Colorado