Posted by Tim Stahley on Mon, Nov 16, 2009 @ 09:31 AM
Tracking the news recently, it is great to see data from new satellite sensors becoming available. The USDA announced recently that it will begin using imagery from DMCii. First images from DigitalGlobe's Worldview-2 have been released. Certainly exciting times!
Technology and commercial applications typically drive sensor technology advancements to greater spatial resolution, often sacrificing spatial coverage, temporal resolution and spectral resolution. While greater spatial resolution is important and certainly a boon for applications such as Google Maps, Google Earth and other consumer oriented applications, these advancements can make it more challenging for users utilizing imagery for agricultural intelligence, environmental and resource management, land cover mapping etc. The key to many of these applications is finding the right mix of spatial, temporal and spectral resolution. With growing number of sensors available and increasing requirements for greater spatial and spectral coverage, users are increasing challenged with making trade offs in one of these areas.
For example the DMCii imagery, while providing greater combination of spatial and temporal resolution than most sensors, limits the availability of spectral bands to three, impacting the ease of crop and land cover classification, monitoring crop health and crop progress or determining attributes of such as water content or soil moisture.
Ideally every application would like to take advantage of the newest technology. So whats the solution?
More and more users are going to be required to blend imagery from multiple sources and multiple sensors for use in a single application. In order to maintain high accuracy, this blending can only be done by calibrating all the imagery used within an application to a specific level, such as top of atmosphere or surface reflectance.
Let's take a closer look at an agricultural example. Suppose I have a historical archive of P6-AWiFS or Landsat NDVI (Normalized Difference Vegetation Index) data and I wish to change to using DMCii imagery for extracting NDVI data going forward. Is my historical AWiFS or Landsat data valid for comparison with new data from DMCii imagery. The answer is no, unless both datasets have been calibrated to the same level, such as surface reflectance. Calibrating the imagery removes variations and distortion from the imagery and ensures the data is correlated to a single reference level, whether that is the top of the atmosphere or the Earth's surface.
I am an advocate of calibrating to surface reflectance, whether it is for a single scene, multiple scenes from the same sensor or multiple scenes from multiple sensors. Surface reflectance's per pixel calibration to the Earth's surface provides greater accuracy than other methodologies. However, your requirements relating to accuracy, cost and timeliness will drive your decision on which methodology best fits your needs.
Bottom line, embrace the technology advancements, but be mindful of the challenges these advancements will bring as we move forward.
If you would like to see how GDA addresses these challenges today visit our surface reflectance page.
Comments or Feedback?
Posted by Tim Stahley on Mon, Oct 19, 2009 @ 02:35 PM
Satellite imaging is prevalent in many consumer applications today, such as Google Earth, Google Maps and many in car GPS systems. Nearly everyone has seen a satellite image of their town, their favorite college football team's stadium or even their own home. I have one of Penn State's Beaver Stadium in my den. It should come as no surprise then, when one thinks of using satellite imaging in the study of agriculture crops that it is these familiar images that come to mind. Images of corn fields and soybean fields with a farmhouse or barn in the back ground. Many are surprised to learn that the practice of using satellite imaging in agriculture truely is much more science than art.
Visible light occupies a very small frequency band of the electromagnetic radiation spectrum, which ranges from long wavelength, low energy radio waves to short wavelength, high energy Gamma Rays. Satellite sensors are able to capture 
reflected light radiation in both the visible range, as demonstrated by the pictures we frequently see, and the non-visible range, which is the most valuable in the study of agriculture. By measuring the reflection and absorption of various frequencies within the electromagnetic radiation spectrum, scientist are able to extract and analyze spectral signatures that allow for the identification of specific vegatation types as well as the measurement of key plant characteristics such as chorophhyl and water content.This information taken over regular intervals allows for monitoring crops through an entire growing season and when combined with complex models and historical data provide accurate prediction methodology for crop yields, crop area and crop production.
Check back often to further explore the current and potential future impact that satellite imaging will have in the agriculture industry.