Integrating RS&GIS for Precision Agriculture Solutions

This section emphasises monitoring pistachio trees, preparing case studies for training, and acquiring fieldwork data, with an emphasis on skill development.

In the next section, the challenges within the pistachio industry, mission planning, the definition of pest and weed problems in pistachio orchards, data acquisition, including the use of remote sensing and drone data, as well as digital tools for monitoring, detecting, classifying pistachios, creating management zones, and pest detection, will be discussed.

Pest management: Agonoscena pistaciae (Pistachio Psyllid)

Significance: The pistachio psyllid, Agonoscena pistaciae, is identified as a "serious pest" causing "significant foliar damage due to its sap-sucking nymphs that secrete wax and honeydew," ultimately impairing photosynthesis and reducing nut yield and quality. It is one of the most economically significant pests in pistachio orchards.

Detection Challenges: Measuring psyllid density remotely requires detecting "subtle changes in leaf reflectance or chlorophyll content caused by psyllium infestation" using high-resolution cameras or RGB/multispectral sensors on drones or movable platforms.

• Multi-stage Detection Strategy: MZ Creation: First, MZs are created using Sentinel-2 data to identify "weak areas in the orchard" where pest issues are likely.
• Leaf-level Detection (Nymph/Egg Stage): "In each MZ, a chassis cage will be constructed with cameras for digital control of the pest on leaves (compound leaves) from different directions... for the identification (detection and classification) of the pest." This involves counting individuals on leaf surfaces.
• Adult Stage Monitoring: A method will be developed to "monitor and determine the population change of the adult stage of the pest in the orchard to inform further action by farm managers (such as chemical control)."

Sensor Considerations: Practical considerations for sensors, including autofocus, close-range focus, wired/wireless connectivity, ruggedness, accuracy, affordability, and digital communication for remote data access. Note challenges with motion blur and stray light. Sample images of psyllids on leaves are provided, noting the ability to identify pests visually and with bounding box overlays from Matlab-developed code.

AI Integration: "Edge cameras with AI - AI-driven LoRaWAN pests detector using RGB AI Sensor via App" are proposed for fast image analysis and automated pest detection.

Pest Models and Spray Timing: "Pest models are customised to individual orchards, allowing for fine-tuned sprays for maximum impact and potentially reducing the number of treatments." These models, combined with climate information (e.g., degree-day modelling), provide "accurate and reliable risk levels sooner, making spray scheduling and logistics management easier."

As a result, this section showcases the precision agriculture solutions for fruit trees using IoT sensors, machine learning, and big data analytics. Key case studies include insect pest management through movable platforms and sensors, …, plant stress monitoring to optimise irrigation. The VETfarm partners emphasise hands-off setup, installation and maintenance of sensor solutions for pistachio growers.

The pistachio growers in the established VETfarm network need to measure the density of psyllium or similar pests and diseases in the pistachio orchards using remote sensing technology to achieve accurate measurements.

Measuring Psyllium Density in Pista Fields Using Remote Sensing

Psyllium, a pest, can significantly impact pistachio yields. Can monitoring psyllium density via remote sensing be achieved with high-resolution cameras mounted on a stick or RGB/multispectral sensors on a drone? We need to detect subtle changes in leaf reflectance or chlorophyll content caused by psyllium infestation.

A literature review on the accuracy achieved based on the resolution of the sensors and the method used, if any, is presented below.

A review on the biology, ecology, and management of Agonoscena pistaciae

It has been determined that there are more than 100 harmful insect species in pistachio fields in Iran, and 20 of them cause economic damage and produce a 50% loss of product (Davatchi, 1958).

The VETfarm team in Armenia reported in 2025 that there are some harmful insect species in the study area in the Armavir region, and Agonoscena pistaciae is one of them.

Agonoscena pistaciae Burckhardt & Lauterer (Hemiptera: Psyllidae), commonly referred to as the pistachio psyllid, is a serious pest of Pistacia vera L. (Sapindalis: Anacardiaceae) throughout the Middle East and Mediterranean basin. This pest was first reported by Keriokhin (1946) from both domesticated pistachio (P. vera) and wild pistachio (Pistacia mutica) in Iran (Takalloozadeh, 2008). Known for its rapid population growth and multiple generations per year, this pest causes significant foliar damage due to its sap-sucking nymphs that secrete wax and honeydew. These excretions contribute to the development of black sooty mold, thereby impairing photosynthesis and reducing nut yield and quality. Because of its economic importance, sustainable management approaches are necessary, and these must combine a deep understanding of the insect’s biology, routine field monitoring, and the application of modern technologies such as remote sensing.

Requirement analysis for skills development

In this section, a practical infrastructure and sensor design and platforms for quality data acquisition on Psyllidae will be discussed. This will be used to determine the population development of Agonoscena pistaciae through field studies, sampling of the adult stage, the nymph and egg stage of the pest through remote sensing techniques.

First, Management Zones (MZ) will be created in a pistachio orchard using the Sentinel-2 satellite data, using the WI tools discussed above. This part of the course deals with the techniques and applications of generating MZ. MZs are defined as sub-units of farm fields with a relatively homogeneous combination of yield-limiting factors. Each zone can be managed with a different but specific single-rate management practice to maximise the efficiency of farm inputs in the context of precision agriculture. Based on the pistachio farm management inquiries discussed above, phenological information and climate information (drought) are vital factors in MZ delineation. In this context, MZ applications are:

• Identify MZ with low expected yield (weak areas in the orchard) in confirmation with the farm manager. Detect patches of productive and unproductive trees and their variations.
• identify sample sick and low-fertile trees in weak MZ together with the farmer manager and local experts (Expert knowledge),
• How accurately can the trees be compared with high and low-load (yield) (fertile and low-fertile trees) using Mobile Labs?

In this part, we generate MZ based on the open-source Sentinel-2 satellite data for variable-rate applications.

As discussed above, the generated MZ will help identify where representative yield/pest samples should be taken. A general guideline is to divide fields of interest into three segments, with at least three samples taken from each segment. This results in a total of 9 samples (3 x 3), unless the differences between the regions or zones are not statistically significant.

Second, in each MZ, a chassis cage will be constructed with cameras for digital control of the pest on leaves (compound leaves) from different directions under the chassis during the pistachio growth stage. This stage is for the identification (detection and classification) of the pest. Additionally, the lifespan of Agonoscena pistaciae at different temperatures and growth stages will be considered. In this stage, the pest counting (observed significant differences in terms of colour and size) and their distribution will be done in each chassis cage. The counts of the egg and nymph stages of the pest will be made by counting the individuals on the lower and upper surfaces of the compound leaves.

Third, a method will be developed to monitor and determine the population change of the adult stage of the pest in the orchard to inform further action by farm managers (such as chemical control), as well as post-action monitoring and impact assessment.

Feasibility of use in the field, manual phase:

How to reach the leaves?

• Most likely to be affected at the top, we can start with a manually controlled telescopic "selfie" type stick or pole.
• Wear goggles as a monitor for guiding the 'camera'?
• Movable portal system with all your sensors, including RGB camera with (low/high) resolution, focused at a short distance, position, look angle, plant ID, date + time stamp. There is also a problem with stray light that can be overcome. The low resolution may be an advantage for cameras or camera boards with GPU edge computing. These could do fast image analysis. The required spatial references, GPS, magnetic field, and pressure can be added as HAT boards. We can make a preliminary table with pros and cons. The weight factors are up for discussion.
• The cost-benefit for the pest detections, based on the need of the UNDER SUN company and network!
• The mechanical phase could be implemented as a self-balancing mobile selfie stick.
• The positioning problem relative to moving leaves could be solved by taking video bursts followed by algorithmic selection of suitable data [experimental].
• 3D plant models
• Edge cameras with AI - AI-driven LoRaWAN pests detector using RGB AI Sensor via App

Requirements and considerations:

Using different sensors and devices, we want to monitor real-time data on soil, pistachio, and pest, and environmental indicators, meteorological information, and more, providing accurate references for pistachio production and environmental monitoring. In this way, the VETfarm team can help support the pistachio industry achieve practical and affordable solutions.

Practical considerations for leaf and canopy level sensors, along with the requirements for easy installation on a movable platform system, data collection, and sharing the data:

• Pan & Tilt (PTZ),
• Autofocus, and
• Effective focus at very close range (less than 5 inches / ~12 cm),
• Wired or wireless
• Rugged close-up pest monitoring
• Accuracy, reliability, ease of use, and installation convenience.
• Affordable costs
• Easy way to power sensors
• Digital communication and data transmission (remote access to data on the cloud), and management (data loggers!).
• Can we connect these sensors to a network to create an IoT sensor network?
• Training and technical support for the installation, maintenance, and operation of these sensors for ANAU and UNDER SUN.
• What is the warranty for these sensors?
• Waterproof and weatherproof features
• Triggering it from a smartphone as an alternative to drones.

Preliminary results

Mobil Labs in Orchards: Enhance traditional farm management by incorporating cutting-edge innovative technologies and mechanisation. Introduce concepts like measurement and control, AI, precision agroforestry practices, sensor-based monitoring, and data analytics to optimise input and address challenges such as pest infestation and water scarcity. In this course, we will explore the design and setup of mobile labs. The labs include portable sensors that can be used on a farm to connect with the local community. Using present-day technology, such as combining Remote Sensing, IoT, AI, and geoinformation, it is possible to monitor the growth and development of plants with economic and environmental significance.

Equipment Procurement Strategy: Conduct a comprehensive market analysis to identify cost-effective sensor selection and integration systems, including soil and plant sensors, platform and pole-mounted sensors and cameras for monitoring pests and diseases. Additionally, establish a laboratory for testing and developing customised technological solutions. Implement Mobile Labs for testing and experimentation, using IoT-enabled systems to detect stress factors such as insect infestations, diseases, and water deficiencies. This approach aims to monitor early signs of disease outbreaks, facilitating proactive and targeted interventions.

Sample psyllid images:

Below is a figure displaying seven RGB images taken in the field for the pest detection challenge.

← Previous Next →

Figure 1: The size, colour, and distribution of Agonoscena pistaciae and their droppings on leaves as a result of feeding adults and nymphs(acquired visual images using a cell phone).

Figure 2: Overlay bounding boxes after running a Matlab-developed code over seven raw RGB images, giving an overview of RGB + bounding boxes. See image seven above for easier discussion.

It is up to the users to decide whether this version is sufficient, given the other challenges in data acquisition.

From an academic/teaching point of view, we suggest:

• For skill training, this could be good enough, quick/short development time, no tuning of methods and parameters after initial human intelligence-based development.
• academic/course materials on knowledge-based image analysis/publication. We could add a search for an optimum confusion matrix result. Optimum is to be defined by a user.

Practically:

• The present manually acquired data has to be stored in a GIS-type data structure with position, look angle, plant ID, date + time stamp. Keywords to retrieve raw and processed data.
• Data acquisition with any camera (RGB sensor, iPhone, borehole inspection, etc.) has an unsolved problem of keeping leaves still relative to the camera to avoid motion blur (investigate tolerances).

Start with a manual + "selfie stick" type of extension to sample high leaves. How does the pest population density increase with Z or with outward growing "wave"? If available, experiment with, for example, a cherry picker or a small crane. Drones give extra turbulent leaf movement!

Quality data acquisition of leaf damage due to possible virus infection, etc., should be the first step for the data analysis part to make sense.

What is the priority and the next priority?

Still to do: configure the sensors, such as an RGB camera or an endoscope camera, for portable/field use.

Review the data acquisition process and the resulting images.

Showcase typical acquired images per sensor and challenges.

For example, the below figure, when displayed at 100%, shows an iPhone picture that clearly shows a problem with the autofocus and a preference for a considerable object distance. But if we fix the focus, it will provide 4x better linear resolution than the borehole/endoscope camera.

The borehole camera has a much lower resolution, but the focus at short distances is better.

There is also a problem with stray light that may be overcome.

The low resolution may be an advantage for cameras or camera boards with GPU edge computing. These could do fast image analysis.

The required spatial references, GPS, magnetic field, and pressure can be added as HAT boards.

We can make a preliminary table with pros and cons. The weight factors are up for discussion.

These results will guide us to plan for an optimum setup and chassis or moveable portal.

Mobil Labs and Precision Farming Platform (PFP)

Design and setup 'Precision Farming Platform (PFP)’

Concept: "Mobile Labs" and a "Precision Farming Platform (PFP)" are introduced as practical, portable systems for on-farm data collection and experimentation. These "incorporate cutting-edge innovative technologies and mechanisation" including "measurement and control, AI, precision agroforestry practices, sensor-based monitoring, and data analytics."

Design and Functionality: A multi-functional chassis, developed by AgriWatch and adaptable for pistachio trees, serves as the PFP. It can integrate various sensors (RGB, multispectral, depth cameras, infrared radiometers) for detailed crop monitoring, including pest detection.

Advantages: The PFP enables "standardised sampling" (e.g., 3m x 3m grid) within MZs, allowing for manageable data collection and comparison of health metrics across the farm. It also facilitates the evaluation of "impact of farm management practices, specifically pest management and pesticide application, on plant growth and health under controlled conditions."

Assessing the health and conditions of fruit trees is crucial for maintaining plant/fruit quality and yield. However, measuring health across the entire farm and considering all relevant factors presents a significant workload using ground sampling, especially when considering the manual inspection involved (for checking the pest infestation). To streamline this process, a standard sampling measure of, e.g., 3 meters by 3 meters, has been proposed by AgriWatch using the delineated MZ by WI tools using Sentinel-2 multitemporal data and can be improved using drone images. This standardised sampling allows for practical and manageable (considering labour requirements) sample data collection, facilitating the comparison of health metrics across the farm without the exhaustive effort required for a comprehensive analysis of all sections.

A movable sensor platform was developed in the Netherlands at AgriWatch for this sampling, within blueberry plants, to demonstrate the necessary hardware resources and practicality. This chassis cage with a movable sensor platform is referred to as the 'Precision Farming Platform (PFP)’. This platform can easily be adapted for pistachio trees in the UNDER SUN farm.

Multi-Functional Chassis:

The multi-functional chassis design can be adapted for various applications in monitoring fruit trees and pistachios. The design platform aims to provide cost-effective, data-driven solutions for sustainable pistachio farming, leveraging cloud-based tools, IoT sensor networks, and mobile lab setups for real-time monitoring and decision support.

← Previous Next →

Anti-bird net: Cover for fruit trees to protect from birds

Updated chassis frame with wheels, used in conjunction with a depth camera. Data requires MATLAB for reading and processing.

In conclusion, the ongoing assessment of pistachio health and conditions and the testing of chemical application methods are critical steps toward improving farm management. The use of standardised sampling, combined with IoT remote sensing and imagery analysis and targeted pest identification, offers a practical and cost-effective approach to understanding and addressing the challenges faced on the farm. Continued monitoring and adaptation of these practices will be key to enhancing yield and ensuring the long-term sustainability of the pistachio orchard.

Additionally, the impact of farm management practices, specifically pest management and pesticide application, on plant growth and health can be evaluated using IoT sensor data on a movable platform under controlled conditions.