The stakes are enormous. Date palms cover approximately one million hectares worldwide, with more than 100 million trees, around 60 percent of them in the Arab world. The red palm weevil, a Category one pest according to the FAO in the Middle East and North Africa, threatens more than 40 palm species and has been spreading relentlessly since it was first documented in the Gulf region in the mid-1980s and identified in Spain on Phoenix canariensis in 1995. In the Gulf region alone, annual losses from eradicating severely infested palms have been estimated at between 5.18 and 25.92 million US dollars, depending on whether infestation levels stand at 1 or 5 percent. Once larvae burrow into the trunk and feed on its interior, an infested palm will die unless the pest is detected and treated at an early stage.

The core problem with existing detection methods is that they see only the surface. Visual inspection remains the standard practice, but a qualified inspector can examine only about 200 to 300 palms per day depending on field conditions, palm density, and sanitation practices. Thermal imaging, which exploits the higher surface temperature of an infested palm under uniform climatic and solar conditions, and multispectral or hyperspectral imaging, which reads subtle biochemical stress signatures in plant reflectance, are promising for large-scale monitoring. Yet all of these approaches are limited to surface-level analysis, depend heavily on environmental conditions, and, most critically, provide no information about where the pest is hiding inside the trunk. That localization matters enormously: knowing the internal position of the weevil allows growers to optimize treatment, whether by tuning the frequency of microwave heating to the required penetration depth or by adjusting insecticide dosage and timing for chemical applications.

Microwave imaging offers a fundamentally different approach because it interrogates the interior rather than the surface. The technique is well established in medical diagnostics, particularly breast cancer screening, where it exploits the dielectric contrast between healthy and malignant tissue. It has also proven useful for non-destructive food inspection, from imaging entire watermelons to correlate microwave images with sugar content and internal anomalies, to screening eggs by albumen volume, to detecting millimeter-sized plastic or glass fragments inside jars of cream on a production line. The underlying physics is the same in every case: low-power, non-ionizing electromagnetic waves, with energy levels comparable to mobile phones and Wi-Fi routers, are transmitted into the object, and the backscattered signals are collected and processed to reconstruct an image of the internal structure. Variations in permittivity produce characteristic scattering patterns, and computer processing of this data enables detection and localization of embedded targets.

In the case of the red palm weevil, the dielectric contrast between the insect and the surrounding palm tissue is substantial, which is precisely what makes microwave imaging viable. Earlier microwave-based pest detection studies used antennas and complementary split-ring resonator sensors that emit waves penetrating deeply into palm trunks, demonstrating that the presence of the weevil alters the dielectric properties of the palm tissue and produces measurable shifts in the scattering response. But those studies were confined to numerical simulations and showed only whether the pest was present or absent, based on frequency shifts. The new work takes a decisive step forward by experimentally validating not just detection but accurate spatial localization of weevil pupae at different depths and positions inside a real palm trunk, bridging the gap between theory and practice.

To do this, the team, led by Fatima zahrae El arroud of the Microwave Energy Sensing group at Mohammed VI Polytechnic University in Benguerir, Morocco, with collaborators at the University of Leicester, the University of Glasgow, and the AgroBioSciences Program at UM6P, built a biologically realistic phantom replicating the pupal stage of the weevil in both geometry and dielectric properties. The pupa was represented as a cylinder measuring 2 centimeters in diameter and 4 centimeters in height, following morphological data for the species, while the palm trunk model used compact dimensions of 20 centimeters in diameter and 10 centimeters in height to manage the intensive memory requirements of 3D finite element simulations in HFSS. The phantom was synthesized as an oil-in-gelatin dispersion using distilled water, gelatin, NaCl, dishwashing detergent, oil, and food coloring, following a stepwise protocol in which the mixture was heated to 80 degrees Celsius, cooled to 35 degrees before adding salt, and brought to 28 degrees before oil was stirred in until the mixture became white and homogeneous. It was then cast into a 3D-printed cylindrical mold and left to solidify overnight.

Dielectric validation was performed with the open-ended coaxial probe technique using a Keysight 85070E dielectric probe kit connected to a Keysight N5235B PNA network analyzer, measuring permittivity across the 0.5 to 6 gigahertz range. The phantom exhibited a dielectric constant ranging from 40 down to 33.5 and electrical conductivity between 0.5 and 3.2 siemens per meter, while the real palm trunk, obtained from the university experimental farm with a moisture content of 138 percent, showed dielectric constant values from 23 to 17 and conductivity between 0.6 and 2.67 siemens per meter. Both materials displayed the typical dispersive behavior of biological tissues, with dielectric constant decreasing as frequency rises due to dipolar relaxation phenomena. The close agreement between the measured phantom properties and the in vivo dielectric characteristics of red palm weevil pupae and palm trunks reported in the literature confirms that the phantom is a valid substitute for real infestations in controlled imaging studies.

The imaging system itself uses broadband Vivaldi antennas operating in the 1.6 to 5.9 gigahertz band, a compromise between penetration depth and spatial resolution for palm tissue, with a maximum gain of 8.5 decibels. In the numerical model, ten identical Vivaldi antennas were uniformly distributed around the trunk, positioned 30 millimeters from its surface. Experimentally, the team adopted a clever low-cost alternative to conventional switch matrices: two fabricated antennas, one transmitter and one receiver, each mounted on a Nema 17 stepper motor driven by TB6560 drivers and coordinated by an Arduino Uno microcontroller. The transmitting antenna rotated around the trunk in fixed 36-degree steps, corresponding to ten transmitter positions, while at each transmitter position the receiving antenna swept through nine discrete positions, effectively mimicking a circular multistatic array. Scattering parameters were measured with a PicoVNA 108 vector network analyzer over 401 discrete frequency points with 16 averages to improve the signal-to-noise ratio, and images were reconstructed using delay-and-sum and delay-multiply-and-sum algorithms, with the latter offering improved localization accuracy.

The results were striking on both simulation and experimental fronts. In electromagnetic simulations, the technique successfully identified pupae at different depths within the trunk and detected pupae of reduced dimensions representative of earlier developmental stages, suggesting potential for catching infestations invisible to conventional inspection. In more complex scenarios involving multiple pupae at advanced developmental stages, the system resolved and localized multiple targets simultaneously. Experimentally, the DMAS algorithm clearly distinguished an uninfested trunk from one containing large pupae at two different locations, and when two pupae were embedded at different positions, the algorithm detected both. Comparing the estimated pest position with the ground-truth location yielded a localization accuracy of approximately 93 percent, indicating that the method can reliably estimate pest position with high spatial precision.

The researchers emphasize that the proposed system is not yet field-ready; all measurements were performed on cut, static palm trunk sections under controlled laboratory conditions at about 22 degrees Celsius, and the pupal stage was selected because it is immobile, allowing reliable measurements with the rotating setup. Future work will move to young palm trees grown in large pots to better mimic field scenarios with natural moisture content and living tissue, extend detection to mobile life stages such as larvae and adults using a fast-switching system that acquires scattering data within seconds, explore indirect infestation indicators such as tunnels, feeding galleries, frass deposits, and boreholes, and integrate machine-learning-assisted reconstruction algorithms alongside metamaterial-based or reconfigurable antenna designs to enhance resolution, penetration depth, and sensitivity. If those steps succeed, microwave imaging could become a deployable, non-destructive diagnostic tool for early pest detection, providing high-resolution information about the number, location, and spatial distribution of weevils inside palm trunks and supporting more effective integrated pest management strategies for sustainable date palm cultivation.

Subject of Research: Non destructive detection and localization of internal pests in agricultural hosts using microwave imaging with application to red palm weevil

Article Title: Non destructive detection and localization of internal pests in agricultural hosts using microwave imaging with application to red palm weevil

Article References: El arroud, F. Z., Zubair, M., Abbasi, Q. H., El Fakhouri, K., Zaarour, Y., Ramdani, C., El Bouhssini, M., & Griguer, H. (2026). Non destructive detection and localization of internal pests in agricultural hosts using microwave imaging with application to red palm weevil. Frontiers in Sensors, 7, Article 1741826. https://doi.org/10.3389/fsens.2026.1741826

Image Credits: AI Generated

DOI: 10.3389/fsens.2026.1741826

Keywords: destructive, detection, localization, internal, pests, agricultural, hosts, microwave, imaging, application, palm, weevil