Introduction Accurate land area estimates form the foundation of reliable agricultural statistics. They provide critical inputs to estimating crop production, calculating yield, and deriving indicators that inform key agricultural policy decisions. These statistics in turn help national governments be more efficient in providing subsidies, designing irrigation programs, and deploying essential extension services. Errors in these estimates introduce biases in production statistics and may introduce inefficient allocations of resources. The Asian Development Bank (ADB) has committed $40 billion through 2030 to strengthen food and nutrition security in Asia and the Pacific. This comprehensive program funds food systems transformation across the region. The initiative helps generate diverse and nutritious food, create jobs, reduce environmental impacts, and promote resilient agricultural supply chains. Within this broader agenda, reliable land area statistics play a pivotal role, allowing member economies to channel resources toward the areas of greatest need and impact. Traditional field surveys face significant challenges in collecting land area statistics, especially in the Pacific Islands, where environmental conditions, physical distance, and natural barriers can make data collection cost-prohibitive, and where cost-effective technologies such as Unmanned Aerial Vehicles (UAV) may offer a solution. In 2022, the ADB, in collaboration with the Cook Islands Ministry of Agriculture, conducted a post-enumeration survey on the island of Rarotonga to assess the coverage of agricultural households and validate agricultural area estimates reported in the 2021 Cook Islands Census of Agriculture. Validation involved comparing subjective area measurements reported by farmers against objective measurements obtained through handheld GPS devices. The analysis, detailed in the study Plotting from Above: Enhancing Agricultural Mapping in Asia and the Pacific (ADB, 2022), revealed that farmers generally underestimated their agricultural land areas by approximately 13% relative to objective GPS measurements, with larger parcels (>2,000 m²) more likely underestimated and smaller parcels slightly overestimated. In a follow-up activity conducted in 2023, the ADB team expanded this approach by employing unmanned aerial vehicles (UAVs) to measure agricultural land areas, aiming to reduce resource intensity while maintaining accuracy. UAVs capture high-resolution imagery that enables subject matter experts to delineate parcel boundaries and perform precise agricultural area measurements directly from the images. UAV technology is particularly suitable for the Pacific context, where frequent cloud cover, fragmented land parcels, and difficult terrain require flexible, localized, and detailed data collection methods. Beyond agricultural area validation, the team also explored additional UAV applications for agricultural statistics, including crop counting and plant health monitoring. By developing multiple use cases for UAVs, this activity aimed to increase the cost-effectiveness of drone operations and produce key agricultural statistics. UAVs offer a powerful complement to satellite imagery and traditional data sources by providing high-resolution, timely, and flexible imagery that enables precise mapping of agricultural parcels, crop counting, and plant health monitoring – especially in smallholder and fragmented farming systems. Their ability to fill critical data gaps, validate self-reported or GPS-based estimates, and support sample-based monitoring makes them particularly valuable in contexts like the Pacific, where cloud cover and terrain pose challenges for satellite-based approaches. However, fully realizing the potential of UAVs requires addressing operational constraints, including limited flight coverage, high initial and recurring costs, regulatory hurdles, and gaps in local technical capacity. Building in-country expertise, aligning with aviation protocols, and adopting scalable, sample-based approaches are essential to integrating UAVs sustainably into agricultural statistics systems. Spatial Resolution and Sensor Capabilities of Satellite and UAV Imagery Satellite and UAV imagery offer complementary capabilities for agricultural data collection, each suited to specific spatial, temporal, and operational needs. Think of them as different brush sizes for an artist—satellite imagery provides broad strokes for national-scale monitoring, while UAVs deliver fine detail at the local level. Just as a broad brush outlines the general structure, fine brushes add clarity and precision where it matters most. Compared to public satellite image sources, UAV imagery delivers higher spatial resolution, typically achieving ground sampling distances (GSD) between 1 and 5 centimeters, depending on flight altitude and sensor type (Figure 1 and Table 1). A GSD of 3 centimeters means each pixel represents a 3 cm × 3 cm area on the ground. Smaller GSD values capture finer spatial detail, allowing analysts to identify individual plants, row spacing, field boundaries, and canopy structure. Figure 2 demonstrates this resolution, showing a mixed cropping area at approximately 3 cm GSD, alongside a close-up where individual leaves are visible. Figure 1: Visual Comparison of Spatial Resolution Note: Visual comparison of spatial resolution for a 500 m² area using imagery from Landsat 8 (left), Sentinel-2 (center), and UAV (right).Source: Landsat 8 and Sentinel-2 imagery retrieved from the Google Earth Engine catalog, corresponding to acquisitions from August 2023. Landsat 8 imagery is courtesy of the United States Geological Survey; Sentinel-2 imagery is courtesy of the European Space Agency via the Copernicus SciHub. UAV imagery was collected during an ADB field exercise conducted in the same month. Table 1: Comparison of Satellite and UAV Imagery for Agricultural Remote Sensing Satellite Imagery UAV Imagery Spatial resolution (Ground size per pixel) Commercial: ~30 cmPublic: 10 m (Sentinel-2 RGB) to 1 km (MODIS) Dependent on flight altitude and camera specs. With a 20 MP RGB camera (24 mm focal length): ~1 cm at 30 m altitude; 3–5 cm at 100 m altitude Spectral resolution (Number and type of wavelengths captured) Multispectral, SAR (radar) Typically, RGB; multispectral, thermal, and LiDAR possible with additional sensors Temporal resolution (Frequency of image capture over the same location) Sentinel-2: 5 daysLandsat: 16 daysMODIS: daily Sentinel-2: 5 daysLandsat: 16 daysMODIS: daily Coverage Area (Total area captured per image or flight) Large area coverage, defined by satellite swath (e.g., Sentinel-2: 290 km) Limited coverage per flight, constrained by battery life, altitude, and line-of-sight regulations Cloud Cover (Effect of clouds and weather conditions on data acquisition quality) Major limitations, especially in tropical zones Flies below clouds but sensitive to weather conditions like rain and wind Cost (Financial investment) Public sources: free Commercial sources: variable pricing Higher initial investment; requires skilled operators. Economically viable typically for areas of 5–20 hectares Data size (Volume of data generated) Moderate-to-high; typically, smaller per unit area compared to UAV imagery Very large; significant computing resources required for processing Suggested use cases in agricultural statistics National/ regional agriculture and crop area mapping, crop monitoring, drought and water stress, land use and cover mapping, disaster impact Parcel boundary delineation, ground truthing for crop area mapping with satellite imagery, canopy heigh estimation, plant counting Advantages Broad coverage, with regular scheduled monitoring. Historical data is available. Very high detail, flexible when needed, low marginal costs beyond initial investment Limitations Coarser detail, quality depends on atmospheric and cloud cover, and costs for higher resolution commercial imagery are significant. Area of coverage is limited, data processing requirements, and regulatory restrictions Source: Compiled by ADB data collection team. Rarotonga, Cook Islands, 2023. Figure 2: Zoom Progression Illustrating Spatial Resolution in UAV Imagery Notes: Left: Aerial view of a mixed cropping area at approximately 3 cm GSD, with a red marker indicating the zoom location. Right: Close-up of the same location, showing individual leaves of a single plant.Source: UAV imagery collected during the ADB field exercise in Rarotonga, Cook Islands, August 2023. UAVs carry various sensors, including RGB, multispectral, thermal, and LiDAR, for agricultural analysis (Figure 3). RGB sensors capture true-color images for mapping and interpretation. Multispectral sensors calculate vegetation indices (e.g., NDVI, NDWI) to assess plant health and moisture levels. Thermal sensors detect temperature differences, helping identify livestock, soil moisture, and crop stress. LiDAR sensors generate detailed 3D data on elevation, canopy structure, biomass, and drainage patterns. Figure 3: Comparison of UAV-mounted Image Sensors Source: RGB and multispectral imagery captured during the ADB field exercise in Rarotonga, Cook Islands, August 2023. Thermal image retrieved from UAS Vision (2019). LiDAR image retrieved from WINNEWS (2024). Practical Applications of UAVs in Agricultural Mapping Use Case 1: Area estimation Obtaining accurate and unbiased agricultural area estimates has long posed a challenge in agricultural statistics. Self-reported figures from farmers often lack reliability, while objective ground-based methods, such as taping, compass measurements, or GPS surveys, require significant labor and resources (ADB, 2024). One alternative involves delineating field boundaries using georeferenced imagery, such as satellite or UAV data. Some studies have used Google Maps (ADB, 2018) or satellite imagery (ADB, 2024) to delineate fields. However, limited spatial resolution and outdated images often reduce accuracy, especially in Pacific Island Countries, where small plots may span only a few pixels. UAV imagery offers recent, high-resolution data that enables clear identification and delineation of individual parcels. Figure 4: Comparison of Parcel Area Measurements: Garmin GPS, UAV, and ESRI Basemap Digitization via Survey Solutions Note: Measurements from Garmin GPS and Survey Solutions were collected during the post-enumeration survey for the 2022 Cook Islands Agriculture Census in late 2022. Parcel boundaries drawn on UAV imagery were completed in August 2023. Source: Asian Development Bank visualization using QGIS. 2025. Manila. To delineate agricultural parcels on UAV imagery, a subject matter expert, such as an agricultural extension officer or trained enumerator, visually interprets the image. Using GIS software, the expert manually traces field boundaries on the UAV-derived image. At a GSD of 3 to 5 cm, parcel and plot edges, roads, vegetation types, land use, and sometimes crop types are clearly visible (Figure 4). While UAV-based delineation provides high spatial accuracy, it has two key limitations. First, it does not automatically link field boundaries to specific farmers, which limits its use in household-level surveys unless paired with additional data collection. Second, acquiring UAV imagery at scale remains operationally demanding and resource-intensive. As an alternative, this method can be applied on a sample basis to validate or calibrate area estimates from other sources, such as self-reports or GPS measurements. Use Case 2: Counting crops Accurate crop counts play a vital role in agricultural statistics, especially for estimating planted areas in smallholder farms and mixed or scattered planting systems. These counts help derive indicators such as tree density, yield potential, and planted area estimates. They also provide policy-relevant insights for plantation monitoring, productivity assessments, and post-disaster loss estimation. With high-quality labeled reference data, machine learning methods can train models to recognize spatial patterns of specific crop types at a given resolution. Figure 5 illustrates this by applying a convolutional neural network (CNN) object detection model—YOLOv8—to high-resolution UAV imagery to identify and count individual coconut treetops (Ultralytics, 2025). The original image (left) shows a UAV view of a coastal coconut plantation, while the processed output (right) displays the automatic detection of 23 treetops, each marked with a confidence score. Figure 5: Detection of Individual Treetops from UAV Imagery Using YOLOv8 Note: Image captured by ADB in August 2023.Source: Asian Development Bank visualization using Ultralytics YOLOv8. 2025. Manila. Training machine learning models to identify crop types requires high-resolution UAV imagery and well-labeled data. For coconut trees, accurate detection of treetops depends on training data that outline individual crowns under varying conditions, such as changes in lighting and growth stages. UAV imagery provides the spatial detail needed to distinguish tree crowns from surrounding vegetation and ground cover. While the process may result in some misclassifications, this proof of concept demonstrates how combining UAV imagery with machine learning can support the generation of agricultural statistics. Use Case 3: Monitoring plant health UAVs equipped with multispectral sensors efficiently generate high-resolution crop health maps. By capturing reflectance data in the red and near-infrared (NIR) spectral bands, vegetation indices such as NDVI can be calculated as proxies for plant health and indicators of crop stress. Figure 6 presents an NDVI map from a citrus orchard in the Cook Islands, captured in August 2023. The color gradient represents vegetation health based on NDVI values: green indicates high NDVI values, corresponding to relatively healthy vegetation; yellow and orange suggest moderate health; and red shows low NDVI values, potentially indicating bare soil or crop stress due to pests, disease, water deficiency, or nutrient deficiency (conditions that may require farmer intervention). Figure 6: NDVI Map from UAV Imagery for Crop Health Assessment Note: Image captured by ADB in August 2023 and processed in OpenDroneMap.Source: Asian Development Bank visualization using OpenDroneMap. 2025. Manila. The spatial resolution of the imagery enables detailed within-parcel analysis, identifying crop health variations across planting rows and plots. This information provides direct feedback to farmers to target the application of fertilizers and pesticides throughout the cropping seasons. Operational Challenges in UAV-Based Agricultural Statistics Limited scalability of UAV operations. Consumer-grade rotary UAVs are constrained by battery capacity and line-of-sight regulations, which require operators to maintain visual contact with the drone. These limitations reduce the area that can be covered in a single flight. In larger fields or under strong winds, field teams must carry multiple batteries and allocate additional time to complete coverage, adding to the logistical burden and operational costs. High upfront and operational costs. Although consumer-grade drones have become more affordable over time, cost remains a barrier for many NSOs. Beyond purchasing UAV hardware, agencies must also budget for accessories (e.g., batteries, sensors, ground control equipment) and operational expenses such as software licenses, drone operator training, insurance, and maintenance. Regulatory and legal barriers. UAV operations are subject to national aviation and privacy laws, which differ significantly. Common requirements include operator licenses, altitude limits, visual line-of-sight rules, and restrictions near airports or government facilities. In some cases, flight permissions from local authorities or adherence to region-specific protocols may also be necessary. Community acceptance and privacy concerns. Even where UAV operations are legally permitted, it is important to inform communities about flight paths and coverage areas. Communication materials and engagement with local leaders may be needed to ensure transparency and obtain community consent. Gaps in technical capacity and expertise. Operating UAVs and effectively using the imagery for agricultural statistics requires skilled personnel. Licensed operators must handle the UAVs, while trained analysts must process and interpret the imagery to produce meaningful agricultural data. Strategies for Effective UAV Integration Build local analytical capacity. Develop local capacity to analyze geospatial data using accessible tools and data sources. Start with small-scale applications and use free, open-source software to build analytical skills. Leverage publicly available satellite imagery to understand its capabilities and limitations compared to higher-resolution UAV imagery. Match UAV use to operational scale. When designing UAV use cases for agricultural statistics, consider the operational scale and cost-effectiveness. UAVs are best suited for applications requiring high spatial detail over small areas—ideal for sample-based data collection, validation exercises, or monitoring specific zones. Full coverage mapping across large or dispersed areas remains impractical due to flight time, battery limits, and processing demands. Use UAVs for sample-based validation. Instead of mapping entire areas, focus on statistically selected units. These observations can validate key agricultural variables from surveys or administrative reports. For example, UAV-based measurements or crop counts can be compared with farmer-reported data to assess bias. Observed crop types or signs of poor health can help validate crop varieties or indicate lower yields. Align with local regulations and protocols. Understand national UAV regulations and establish clear operating protocols. This may involve coordinating with civil aviation authorities and relevant government agencies. Develop field safety procedures and implement community engagement protocols to address privacy and security concerns. Document and share methodologies. Continue developing and refining UAV methodologies for agricultural statistics. As UAV technologies and analytical techniques evolve, documenting use cases supports broader adoption, encourages innovation, and improves data quality. Ask the Experts Anthony Burgard Consultant, Data Division, Economic Research and Development Impact Department, Asian Development Bank Anthony is an agricultural statistics consultant with the ADB, offering more than16 years of expertise in the design and implementation of agricultural censuses and surveys across the Asia-Pacific region. His work leverages cost-effective technologies like digital data collection using tablets and drones, geographic information systems, and remote sensing. Anthony holds degrees in Economics from the University of California, Berkeley, and Chulalongkorn University. Follow Anthony Burgard on Anna Christine Durante Consultant, Data Division, Economic Research and Development Impact Department, Asian Development Bank Anna Christine has been involved in agricultural statistics projects for nearly 13 years, specializing in innovative methodologies to support policymaking in the agriculture sector. She holds a degree in Agricultural Engineering from the University of the Philippines, and a master’s degree in Environment and Natural Resources Management from the same institution. Prior to joining ADB, she worked on agricultural policy analysis and development assistance projects at the Philippine Department of Agriculture. Takaaki Masaki Senior Economist (Data Science), Data Division, Economic Research and Development Impact Department, Asian Development Bank Takaaki Masaki’s work focuses on leveraging GIS, big data, and machine learning to enhance development statistics, support spatial targeting, and evaluate the distributional impact of policy interventions. Prior to joining ADB, he served as a Senior Economist in the World Bank’s Poverty and Equity Global Practice and previously worked at AidData. His research has been published in journals such as the Journal of Development Economics, World Bank Economic Review, World Development, and the Journal of Politics. He holds a PhD in Political Science from Cornell University. Asian Development Bank (ADB) The Asian Development Bank is a leading multilateral development bank supporting sustainable, inclusive, and resilient growth across Asia and the Pacific. Working with its members and partners to solve complex challenges together, ADB harnesses innovative financial tools and strategic partnerships to transform lives, build quality infrastructure, and safeguard our planet. Founded in 1966, ADB is owned by 69 members—49 from the region. Follow Asian Development Bank (ADB) on Leave your question or comment in the section below: View the discussion thread.