Digital cameras, including those used for aerial RS purposes, utilize arrays of detectors that each sense a particular range of wavelengths of incident electromagnetic energy and translate this into an electrical current. Such an array of detectors is called a focal plane array (FPA), which, typically, will contain millions or tens of millions of detectors arranged in a rectangular grid. It is the electrical output from each of these millions of detectors that is translated into the individual pixel values that make up an image. The issue of concern, here, is that, for any given range in light wavelengths, the spectral response of different cameras is, simply put, different. As a result, data products--even simple products such as normalized difference vegetation index (NDVI)--derived from different cameras are often not comparable.
Orthorectification is the process whereby a collection of digital aerial images is stitched together into a single composite image wherein the geographic location of each pixel is precisely known. The positioning estimation is done, for each pixel, in both the horizontal and vertical. This positioning is done to remove distortions in the imagery due to topographic characteristics of the photographed area; it is also needed because much of the original aerial imagery was obtained at non-nadir/oblique angles. As such, common byproducts of the orthorectification process are digital elevation models (DEMs) and point clouds of the geographic area represented in the full image mosaic. The orthorectification process is done using the fundamental principles of photogrammetry and, for much of the currently available software, pattern recognition algorithms. The major concern with the orthorectification process is the accuracy of the estimate of the geographic location of the pixels in the resulting mosaic image.
Radiometric calibration is the process by which the individual digital value of each spectral band represented in each pixel captured by the focal plane array is translated into a meaningful, scientific measurement. It is the process by which digital photographs are turned into scientific data that can be used in models and various analytics to generate information products of potentially high value to precision agriculture. In general, the process of radiometric calibration maps the digital pixel values into radiance, reflectance, or brightness temperatures. High quality radiometric calibration becomes more difficult as the geographic extent of the RS target increases. This is the case for all spectral bands, but infrared/thermal wavelengths are particularly challenging. There are few sensor packages available today for use on small UAVs that include a thermal camera because of the difficulty of achieving a high quality radiometric calibration, because of the relatively higher cost of the infrared sensor technology, and because of the greater difficulty in accurately orthorectifying the imagery. However, high-quality measurement of energy and water fluxes between the land surface and the atmosphere using RS technologies all require these surface temperature measurements. Therefore, if operators want to use UAV RS technology to better manage irrigation water, it will be necessary to estimate evapotranspiration rates and, if possible, soil moisture; in using small UAVs to accomplish this RS task, the currently best available sensors are infrared/thermal sensors for evapotranspiration estimation, and short-wave infrared sensors for soil moisture estimation. RS of both of these phenomena will also require simultaneous acquisition of other RS data, e.g., red and near infrared.
The nominal spatial resolution of imagery acquired from deployment of a UAV can be estimated by calculating the geographic area covered in the field of view of the camera and dividing by the number of individual sensor elements embedded in the camera’s focal plane array. This calculation yields the nominal area on the surface of the earth that is represented by the digital value captured in a single pixel; the square root of this number is the nominal spatial resolution of the image. Taken at face value, then, a focal plane array with more individual sensor elements should produce higher spatial resolution than a focal plane array with fewer individual sensor elements. Unfortunately, things are much more complicated than this. In particular, the process of light diffraction and imperfections in the quality and/or alignment of the lens, optical filter, and other optical components of the camera can significantly reduce the real spatial resolution of the camera. This is rarely discussed, or even acknowledged, for RS equipment used on UAVs.
Recent papers supporting the need for agricultural aerial remote sensing standards