Agricultural Spatial Analysis and Modelling

Agricultural spatial analysis and modelling bring together geographic information systems (GIS), remote sensing, statistical techniques and machine-learning to understand and optimise farming across diverse scales. At the field level, high-resolution imagery from drones or satellites is used to map soil properties, detect pests and guide precision spraying, allowing inputs to be tailored to site-specific needs. At farm and regional scales, crop-growth and land-surface models simulate water and nutrient flows under present and future climate scenarios, informing adaptation strategies and land-use planning. Geostatistical methods such as variograms and kriging interpolate soil or pest data, while agent-based models explore how farm-level decisions aggregate to landscape patterns. Digital soil mapping and reflectance spectroscopy have transformed soil information, producing continuous maps of organic carbon, texture and pH. Coupled with big data from weather stations, farmer inventories and census records, these tools support yield forecasting, water-footprint assessments and policy design. The result is a powerful toolkit for sustainable intensification, resource conservation and risk management in the face of climate change and growing food demand.

Research from Nature Portfolio

Site-specific real-time spraying control systems have been evaluated in soybean and maize, using on-board optical sensors to activate nozzles only where foliage is present. Trials in the Brazilian Cerrado show a two-thirds reduction in pesticide volume without penalty to crop yield, even across application stages for herbicides, fungicides and insecticides. A parallel study applied unmanned aerial system (UAS) imagery to generate prescription maps for corn fields, distinguishing inter-row vegetation as weeds. By segmenting imagery with a crop-row identification algorithm and dispensing herbicide only in weed-infested grid cells, herbicide use was cut by over a quarter compared to conventional broadcast spraying. In the Middle East and North Africa, a composite drought indicator has been co-developed with national agencies. Combining remote sensing and modelled datasets of precipitation, soil moisture, evapotranspiration and vegetation greenness, this indicator is produced monthly to guide operational drought management, highlighting regional anomalies and informing policy decisions on water allocation and risk mitigation.

Research from all publishers

Global digital soil mapping at 250 m resolution has been achieved by integrating over 200 000 soil observations with more than 150 environmental covariates through ensemble machine-learning. The resulting maps cover organic carbon, bulk density, pH, cation exchange capacity and texture at seven standard depths, with cross-validated accuracies of 56–83 % across properties, marking a 60–230 % improvement over previous 1 km products. On the continental scale in Europe, visible and near-infrared reflectance spectroscopy was applied to nearly 20 000 soil samples. Calibrated models yield unbiased soil organic carbon predictions with root-mean-square errors of 4–15 g C kg⁻¹ for mineral soils, demonstrating the value of cost-effective, high-density spectroscopic sampling for national-level monitoring. In South-Western Burkina Faso, Landsat and Shuttle Radar Topography Mission data were combined with over 1100 field samples to train random forest and gradient-boosting models for six soil properties. Remote sensing predictors captured over 85 % of the variance in texture, organic carbon and nutrient content, illustrating how freely available imagery can fill soil-information gaps in data-scarce regions.

Agricultural Spatial Analysis and Modelling publication trend

The graph below shows the total number of articles in agricultural spatial analysis and modelling across all publications each year (not limited to Nature Index journals).

Technical terms

Precision spraying: site-specific application of fertilisers or agrochemicals guided by sensors or imagery to match input rates to actual crop or weed locations.

Unmanned aerial system (UAS): A remotely piloted aircraft equipped with cameras or multispectral sensors to collect high-resolution surface data for precision agriculture.

Composite drought indicator (CDI): An index combining satellite- and model-based anomalies in precipitation, soil moisture, evapotranspiration and vegetation vigour for operational drought monitoring.

Digital soil mapping: The use of statistical and geospatial techniques to predict soil attributes continuously across landscapes from discrete sample points and environmental covariates.

Visible and near-infrared reflectance spectroscopy: A rapid, non-destructive method measuring soil or vegetation reflectance between 400–2500 nm to infer chemical and physical properties such as organic carbon.

Ensemble machine learning: The application of multiple algorithms—such as random forests and gradient boosting—whose combined predictions enhance accuracy and quantify uncertainty in spatial models.

References

1. Reduction of pesticide application via real-time precision spraying. Scientific Reports (2022).
2. Towards reducing chemical usage for weed control in agriculture using UAS imagery analysis and computer vision techniques. Scientific Reports (2023).
3. Development of a composite drought indicator for operational drought monitoring in the MENA region. Scientific Reports (2024).
4. SoilGrids250m: Global gridded soil information based on machine learning. PLOS ONE (2017).
5. Prediction of Soil Organic Carbon at the European Scale by Visible and Near InfraRed Reflectance Spectroscopy. PLOS ONE (2013).
6. High Resolution Mapping of Soil Properties Using Remote Sensing Variables in South-Western Burkina Faso: A Comparison of Machine Learning and Multiple Linear Regression Models. PLOS ONE (2017).