Digital mapping of agrochemical properties on a detailed scale using UAVS

DOI: 10.35595/2414-9179-2025-2-31-457-472

View or download the article (Rus)

About the Authors

Aleksey N. Chashchin

FSBEI HE “Perm State Agro-Technological University named after academician D.N. Pryanishnikov”,
23, Petropavlovskaya str., Perm, 614990, Russia,
E-mail: chascshin@mail.ru

Natalya M. Mudrykh

FSBEI HE “Perm State Agro-Technological University named after academician D.N. Pryanishnikov”,
23, Petropavlovskaya str., Perm, 614990, Russia,
E-mail: nata020880@hotmail.com

Iraida A. Samofalova

FSBEI HE “Perm State Agro-Technological University named after academician D.N. Pryanishnikov”,
23, Petropavlovskaya str., Perm, 614990, Russia,
E-mail: samofalovairaida@mail.ru

Abstract

The effectiveness of using aerial photography from an unmanned aerial vehicle (UAV)
data to improve the accuracy of agrochemical mapping has been evaluated. The practical signi-
ficance of integrating topographic and spectral predictors obtained from UAVs in digital soil
mapping models has been determined. The research was conducted in the Non-Chernozem zone on the territory of the Perm Municipal District of the Perm Territory. The object of research is a site at the educational and scientific experimental field of the Perm SATU, Perm. Soil sampling was carried out before sowing the crop from a depth of 0–20 cm on a fixed grid of 100 × 200 m. A set of point agrochemical data (spot samples and data from the arable layer in sections) was collected for four indicators (humus, mineral nitrogen, phosphorus, potassium) and their spatial modeling was performed using the “ordinary kriging” method. Using the UAV, highly detailed data was obtained, on the basis of which topographic and spectral predictors were calculated. The method of correlation analysis (Scatterplot in SAGA GIS) revealed the most informative predictors with a statistically significant relationship with the agrochemical properties of soils. The identified informative predictors were used to map agrochemical indicators as input variables to the machine learning model (Random Forest method). The obtained property maps were compared with the results of the initial interpolation using RMSE. Additionally, it was investigated whether the improved forecasting accuracy affects the practical agronomic significance, using the example of calculating soil demand for phosphorous fertilizers. A comparison of the results of machine learning and geostatistical methods (kriging) allows us to conclude that it is advisable to use UAVs in precision farming systems, with the possibility of optimizing the distribution of mineral fertilizers.

Keywords

agrochemical properties of soils, digital soil cartography, aerial photography from UAVs, predictors, machine learning, kriging

References

  1. Belenkov A.I., Mazirov M.A., Voronov M.A. Results of Scientific and Practical Development of Precision Farming in Field Experiment. Plodorodie (Fertility), 2025. No. 1 (142). P. 31–41 (in Russian). DOI: 10.25680/S19948603.2025.142.08.
  2. Bryzhko I.V., Shabalina T.V. GIS-Based Support for Precision Farming on the Example of the Tyumen Region. InterCarto. InterGIS. Proceedings of International Conference, 2021. V. 27. Part 4. P. 66–81 (in Russian). DOI: 10.35595/2414-9179-2021-4-27-66-81.
  3. Chashchin A.N., Gilyov V.Yu. Soil Mapping Based on UAV Aerial Photography. AgroEcoInfo, 2024. No. 5. DOI: 10.51419/202145502. Web resource: http://agroecoinfo.ru/STATYI/2024/5/st_502.pdf (accessed 10.04.2025) (in Russian).
  4. Digital soil cartography. Moscow: RUDN University, 2017. 156 p. (in Russian).
  5. Dokuchaev P.M. Construction of a Digital Soil Map and Carbon Cartogram Using Digital Soil Mapping Methods (On the Example of the Vyatka-Kamskaya Province of Sod-Podzolic Soils of the Southern Taiga). The dissertation for the PhD in biological sciences. Moscow, 2017. 206 p. (in Russian).
  6. Gafurov A.M. Possible Use of Unmanned Aerial Vehicle for Soil Erosion Assessment. Uchenye Zapiski Kazanskogo Universiteta. Seriya: Estestvennye Nauki (Proceedings of Kazan University. Series: Natural Sciences), 2017. V. 159. No. 4. P. 654–667 (in Russian).
  7. Han H., Suh J. Spatial Prediction of Soil Contaminants Using a Hybrid Random Forest—Ordinary Kriging Model. Applied Sciences, 2024. V. 14. No. 4. P. 1666. DOI: 10.3390/app14041666.
  8. Hunt E.R., Cavigelli M., Daughtry C.S.T., Mcmurtrey J.E., Walthall C.L. Evaluation of Digital Photography from Model Aircraft for Remote Sensing of Crop Biomass and Nitrogen Status. Precision Agriculture, 2005. V. 6. P. 359–378. DOI: 10.1007/s11119-005-2324-5.
  9. Hunt E.R., Doraiswamy P.C., McMurtrey J.E., Daughtry C.S.T., Perry E.M., Akhmedov B.A Visible Band Index for Remote Sensing Leaf Chlorophyll Content at the Canopy Scale. International Journal of Applied Earth Observation and Geoinformation, 2013. V. 21. P. 103–112. DOI: 10.1016/j.jag.2012.07.020.
  10. Kashtanov A.N., Vernyuk Yu.I., Savin I.Yu., Shchepotyev V.V., Dokukin P.A., Sharychev D.V., Li K.A. Mapping of Rill Erosion of Arable Soils Based on Unmanned Aerial Vehicles Survey. Eurasian Soil Science, 2018. V. 51. No. 4. P. 479–484. DOI: 10.7868/S0032180X18040111.
  11. Kutsaeva O.A. Creation of Management Zones for Differentiated Application of Mineral Fertilizers with the Use of Geostatistics Tools. Bulletin of the Belarussian State Agricultural Academy, 2020. No. 2. P. 176–181.
  12. Minaev N.V. Digital Model of Soil-Landscape Connections of the Vladimirskoe Opolye. The dissertation for the PhD in biological sciences. Moscow, 2020. 149 p. (in Russian).
  13. Mudrykh N.M., Samofalova I.A., Chashchin A.N. Forecasting Soil Erosional Losses Using the RUSLE Model. AgroEcoInfo, 2020. No. 4. Web resource: http://agroecoinfo.narod.ru/journal/STATYI/2020/4/st_430.pdf (accessed 10.04.2025) (in Russian).
  14. Pivchenko D.V., Meshalkina Yu.L., Yaroslavtsev A.M., Tikhonova M.V., Vizirskaya M.M., Vasenev I.I. Comparative Analysis of Vegetation Indices for Agroecological Monitoring of Winter Wheat in the Moscow Region. AgroEcoInfo, 2019. No. 3. Web resource: http://agroecoinfo.narod.ru/journal/STATYI/2019/3/st_324.doc (accessed 10.04.2025) (in Russian).
  15. Pouladi N., Møller A.B., Tabatabai S., Greve M.H. Mapping Soil Organic Matter Contents at Field Level with Cubist, Random Forest and Kriging. Geoderma, 2019. V. 342. P. 85–92. DOI: 10.1016/j.geoderma.2019.02.019.
  16. Savin I.Yu., Vernyuk Yu.I., Faraslis I. The Possible Use of Pilotless Aircrafts for Operative Monitoring of Soil Productivity. Dokuchaev Soil Bulletin, 2015. V. 80. P. 95–105 (in Russian). DOI: 10.19047/0136-1694-2015-80-95-105.
  17. Sekulić A., Kilibarda M., Heuvelink G.B.M., Nikolić M., Bajat B. Random Forest Spatial Interpolation. Remote Sensing, 2020. V. 12. No. 10. P. 1687. DOI: 10.3390/rs12101687.
  18. Susantoro T.M., Wikantika K., Saepuloh A., Harsolumakso A.H. Selection of Vegetation Indices for Mapping the Sugarcane Condition Around the Oil and Gas Field of North West Java Basin, Indonesia. IOP Conference Series: Earth and Environmental Science, 2018. No. 149. P. 012001. DOI: 10.1088/1755-1315/149/1/012001.

For citation: Chashchin A.N., Mudrykh N.M., Samofalova I.A. Digital mapping of agrochemical properties on a detailed scale using UAVS. InterCarto. InterGIS. Moscow: MSU, Faculty of Geography, 2025. V. 31. Part 2. P. 457–472. DOI: 10.35595/2414-9179-2025-2-31-457-472 (in Russian)