Supporting organic and conventional farming through geospatial methods

The practical problem addressed

Three information gaps in field management were identified:

1. Crop condition was assessed mostly by visual observation and experience; satellite monitoring was rarely used.
2. Soil fertility data were outdated or insufficient; fertilizers were often applied at a uniform rate across the field.
3. A locally validated methodology was lacking to justify irrigation timing, yield forecasting and management zones.

Farmers asked for simple, mobile-friendly maps and decision support. These needs shaped the project workflow: diagnostics → monitoring → prediction → web dissemination.

Key objectives

Complementary roles of the partners

TIIAME National Research University

• GIS and remote sensing
• Field, soil and yield measurements
• Sentinel‑2 and machine learning
• Conventional irrigated farming

University of Nyíregyháza

• Organic farming expertise
• Organic fertilizer and triticale trials
• Training and international conferences
• Joint publications and exchange of experience

Study areas

2025: sample segments of 10.07 ha / 13.8 ha · 2026: 39.8 ha / 8 fields, ≈69 ha

Scientific workflow

• All measurements at georeferenced points
• Field–lab–Sentinel‑2 linkage through a common ID
• Quality control: duplicates, blanks, calibration, QC
• Model evaluation with grouped cross-validation
• Turning results into maps and web products

Soil nutrient supply classes, 2025

Available phosphorus fell into the “very low” or “low” class at many points and was identified as the main limiting factor. Potassium was generally high to very high, humus mostly low to medium, and nitrogen varied with irrigation and agronomic background.

2025 — spatial distribution of nitrogen and phosphorus

Spatial variability of nitrogen depends on irrigation and fertilization history; phosphorus deficiency in both areas calls for zonal, differentiated fertilization.

2025 — potassium and humus

Potassium reserves are sufficient in many zones; to preserve humus, organic matter, residue return and minimum tillage are recommended.

2026 — Middle Chirchiq soil maps (nitrogen, phosphorus, potassium, humus)

2026 — Boka area soil maps (nitrogen, phosphorus, potassium, humus)

Five neighbouring farms were covered together with the main field; results enabled management zones to be delineated at a wider scale.

Soil properties, yield and three-year comparison

Results show that a universal recommendation based on a single parameter is insufficient: fertilizer, moisture, texture, irrigation and phenology must be interpreted together.

Important negative result. Soil moisture under dense canopy could not be reliably estimated from Sentinel‑2 optical indices; in 2026 all cross-validated models gave R²_CV < 0. For irrigation decisions, Sentinel‑1 radar, permanent sensors and synchronous field measurements are recommended.

Time series of NDVI, EVI, GNDVI, NDRE, SAVI and NDMI supported spatio-temporal monitoring of growth, heading, grain filling and ripening.

Spatial variation of NDVI

NDVI differences over time and space reflected variations in irrigation technology, cultivar, agronomic background and canopy density.

Field vs Sentinel‑2 NDVI: scale mismatch

R² ≤ 0.03three campaigns

1.8–3.2×field NDVI higher

• Leaf level: centimetre scale
• Sentinel‑2: 100 m² mixed pixel
• Soil background, shadow, ditches and weeds
• Date and GNSS errors as additional factors

Field spectra and satellite values cannot be equated directly: band equivalents and a scale-appropriate protocol are required.

2026 cereal spectral library: 23 narrow-band indices — vegetation, chlorophyll, red-edge, pigments, photosynthetic efficiency and water status.

Red-edge, PCA and Sentinel‑2 band equivalents

PC1 + PC2 explained 88.6–94.1 % of total variance; the O₂‑A artefact near 760 nm was separately controlled in red-edge calculations.

179samples

16–106range, c/ha

R² 0.71grouped CV

10.8RMSE, c/ha

7.7MAE, c/ha

≈19%rRMSE

Model: seven mid-May Sentinel‑2 bands + five indices → ExtraTrees regression → cross-validation grouped by field and season.

Most important spectral features

SWIR1 reflected canopy water status; green and NDRE reflected chlorophyll/nitrogen status; NIR reflected biomass and leaf area. Highest importance: SWIR1 (0.40), Green (0.34), NIR (0.28), NDRE (0.21), SWIR2 (0.13).

Spatial wheat yield prediction maps

Methodological caveat. The predicted range is narrower than the measured yield range. The maps are therefore presented as relative productivity and management zones, not absolute yield values. To become a practical forecasting tool, the model must be tested on independent 2026 samples and retrained with three years of data.

Practical recommendations and implementation directions

1Monograph

7Textbooks / guides

14Scopus articles

3Foreign journals

10International conferences

31National journals

10National conferences

2Intellectual property

Annual dynamics: 12 → 20 → 44 scientific outputs (2023 → 2025).

DOI-verified publications

1. Leveraging Vegetation Indices and Random Forest for Soil Nutrient Monitoring in Winter Wheat

   Journal of the Indian Society of Remote Sensing, 2025

   DOI: 10.1007/s12524-025-02198-9 ↗

2. The effect of fermented poultry manure on the yield and quality parameters of triticale cultivated organic way on acidic sandy soil

   E3S Web of Conferences 590, 01009, 2024

   DOI: 10.1051/e3sconf/202459001009 ↗

3. Temporal Evolution of Climate-Related Land Changes in Zarafshan Valley: MODIS and Google Earth Engine Perspective

   AIP Conference Proceedings 3256, 040016, 2025

   DOI: 10.1063/5.0266743 ↗

4. Search for the lost Baghi Chinar garden in the Samarkand region and creation of a Digital Terrain Model

   E3S Web of Conferences 590, 05002, 2024

   DOI: 10.1051/e3sconf/202459005002 ↗

5. Sustainable development of Turkmenistan in the context of climate change

   E3S Web of Conferences 590, 05007, 2024

   DOI: 10.1051/e3sconf/202459005007 ↗

6. Land use and land management in Kazakhstan

   Problems of AgriMarket, No. 3, 2024

   DOI: 10.46666/2024-3.2708-9991.14 ↗

7. Theoretical and Methodological Basis of Mapping the Ecological State of Irrigated Lands in the Conditions of Climate Change: Bibliometric Analysis

   ISPRS Archives, 2025

   DOI: 10.5194/isprs-archives-XLVIII-5-W3-2025-63-2025 ↗

2026 publications

• Multi-temporal monitoring of vegetation dynamics using Sentinel‑2 and Google Earth Engine — Proceedings of SPIE, 2026.
• Preparation and integration of spatial data in a GIS environment: the case of Boka and Middle Chirchiq — 2026.
• Spatial variability of soil agrochemical indicators and Trimble Juno 3B / ArcGIS Survey123 — Nukus, 2026.
• Soil fertility and fertilization efficiency using remote sensing and field analyses — 2026.
• Yield components of triticale with granulated poultry manure — Nyíregyháza international conference, 2026.
• Remote Sensing for Precision Agriculture in Arid and Semi-Arid Regions: A Case Study of Uzbekistan — Springer Nature book chapter, in preparation.