In the 1980s, remote-sensing technology and methods were proposed as a solution for environmental problems because they could continuously monitor the earth's surface. A number of satellites were launched operating in active (i.e., providing their own illumination) and passive (i.e., recording the natural radiation) modes with capabilities ranging from monitoring large spatial swaths at high temporal resolution to high-spatial-resolution imaging at low repeat cycles. The emerging technology and its potential outcomes were oversold, however: current applications in precision agriculture are limited to high-spatial-resolution satellite sensors providing coarse spectral resolution and sparsely sampled revisit times.
Key constraints for successful application of remote sensing in precision agriculture include, among others, very high spatial resolution (pixel sizes of <1m), access to visible, near-infrared, and thermal spectral bands, and use of bandwidths allowing estimation of key crop biophysical parameters such as the concentration of chlorophyll a and b, xanthophylls, carotenoids, anthocyanins, water, and dry matter, as well as leaf-area index and crop temperature. Availability of imaging at critical crop-phenological stages combined with fast turnaround times is an additional key factor.
Figure 1. High-resolution multispectral imaging of an orchard acquired with a fixed-wing unmanned aerial vehicle (UAV).
Since the combination of all of these factors cannot be met with current satellite sensors, applications of remote sensing in agriculture are limited to ‘demonstration’ studies in dedicated experimental fields using high-resolution airborne sensors, crop classification for inventory purposes, and planning studies. Nevertheless, although airborne remote sensing has proved its potential, limitations for actual implementation are driven by the cost of imaging campaigns with full-size airplanes, and the financial and technical difficulties associated with frequent image acquisition. Current methods for remote detection of plant-physiology status rely therefore almost entirely on a normalized index calculated in the visible and near-infrared spectral region and obtained by satellite sensors. This index is closely related to canopy structure but insensitive to either photosynthetic-pigment concentration or physiological process. New worldwide research objectives in crop monitoring aim at estimating photosynthetic- and nonphotosynthetic-pigment light absorption and the detection of chlorophyll fluorescence and vegetation thermal emission linked to transpiration and vegetation stomatal conductance.
Figure 2. Thermal images (expressed in °C) acquired with a helicopter UAV over a peach orchard showing water-stress spatial variability (left), and within-crown thermal differences as a function of stress (right).
Figure 3. UAV detection of crop stress using the photochemical-reflectance index (PRI) (left: well irrigated in blue, deficit irrigation in yellow and orange). On the right, PRI levels indicate the stress condition for each tree crown. sPRI: Simulated PRI for nonstress crop conditions based on radiative-transfer modeling.
These key scientific ideas are therefore operationally constrained in precision agriculture because of technical limitations. Advances in quantitative remote sensing for precision agriculture focus on coupling small hyperspectral (spanning the electromagnetic spectrum) and thermal sensors onboard unmanned aerial robots guided by autonomous navigation systems. This new paradigm will make frequent cost-effective monitoring of agricultural areas possible, enabling estimates of crop parameters currently impossible to obtain from satellite sensors. Remote-sensing detectors on unmanned aerial vehicles (UAVs) could fill this gap, providing low-cost approaches to meet the critical requirements of high spatial, spectral, and temporal resolution. The question we need to address is whether low-cost miniaturized multispectral imagers can provide results similar to those of expensive full-size heavy instruments onboard manned platforms.
We have demonstrated1 the potential to generate quantitative remote-sensing products using UAVs equipped with inexpensive thermal and narrowband multispectral-imaging sensors, acquiring imaging of <15cm resolution and 10nm full width at half maximum (see Figure 1). Surface reflectance and temperature can be obtained after atmospheric corrections on the basis of MODTRAN (MODerate resolution atmospheric TRANsmission) radiative-transfer modeling, using thermal imaging for stress detection as a function of irrigation and water availability (see Figure 2). These methods enable estimation of biophysical parameters such as the physiological index for stress detection as a function of xanthophyll-pigment absorption (see Figure 3)2 and chlorophyll content (see Figure 4), coupled with physical models. As a result, image products of leaf-area index, chlorophyll content, water-stress detection from the photochemical-reflectance index and canopy temperature, and applications such as the detection of water spills in irrigated farms (see Figure 5) can be obtained and validated successfully, demonstrating that low-cost UAV systems for precision-agriculture applications yield comparable (if not better) results than obtained by traditional manned airborne sensors.1
Figure 4. Detection of crop-tree chlorosis (Cab) from a UAV using narrow bands in the visible and red-edge spectral region.
Figure 5. Detection of water spills in irrigated orchards using high-resolution thermal imaging from a fixed-wing UAV platform.
Our future work focuses on obtaining biophysical crop parameters over hundreds of hectares using fixed-wing UAVs characterized by enhanced endurance, moderate cruise speeds, and easier operation. Taking advantage of cost reductions enabled by autopilots and imaging sensors, the vision from the 1970s of real-time irrigation scheduling and crop monitoring will soon become reality.
Pablo J. Zarco-Tejada, J. A. J. Berni, L. Suárez, E. Fereres
Laboratory for Research Methods in Quantitative Remote Sensing (QuantaLab)
Instituto de Agricultura Sostenible (IAS)
Consejo Superior de Investigaciones Científicas (CSIC)
Pablo Zarco-Tejada is director of the IAS (CSIC) and head of QuantaLab. He was previously contract faculty in remote sensing at the University of California, Davis, with main interests in applications of remote sensing for vegetation stress monitoring and precision agriculture.