WO2025068727A1

WO2025068727A1 - Development of a cost-optimised soil sampling method to support precision farming

Info

Publication number: WO2025068727A1
Application number: PCT/HU2023/000019
Authority: WO
Inventors: Vince LÁNG
Original assignee: Discovery Center Nonprofit Ltd
Current assignee: Discovery Center Nonprofit Ltd
Priority date: 2023-09-29
Filing date: 2023-09-29
Publication date: 2025-04-03
Anticipated expiration: 2026-03-29

Abstract

A soil sampling method to support precision farming comprises the steps of: Building a terrain model, done by remote sensing, by collecting logged data to RTK navigation machines, or from publicly available sources; Producing derived data of a topography model, such as slope angle, topographic wetness index, channel network, relative slope position, elevation; Analyse the data by way of principal component analysis and then transform the data into principal components to reduce dimensionality and decorrelation between variables; Randomly sample the database fifty times using the Latin Hypercube method to select the sampling points, which are clustered and developing the final sampling design in form of a table; Examining the sample points, and then using the random forest algorithm to extend the sample points to the entire table. Thus, a sampling algorithm based on a terrain model and derived data with the right representativeness by taking the smallest sample size is provided.

Description

t Pateat description

1) of the invention

Development of a cost-optimised soil sampling method to support precisian farming

2) Name of the technical area:

Natural science - agricultural robnee

3) A description ofthe state of foe art known to foe applicant, including a description of solutions - and, if possible, sources - which are dose to the invention and which will help to understand foe invention and to asms its patentability:

Smart design of a

grotechnological elements is a key clement of precision fanning. A key element of foe hash: database andthe management zone systems derived from is foe knowledge of the soil, its physical, chemidal and bioiogical properties and its variability within the field. At present, real roil mapping by means of profile pit anaiysation and the combined interpretation of natural geography , geology and geomorphology data represents only a small share of the precision market, mainly due to its high cost and the need for skilled human rerources. For this reason, the industry currently uses mainly indirect solutions, Current mapping methods on the market - yield maps, vegetation indices, soil moisture maps derived from soil geophysical properties and conductivity - only reflect the seasonal effect and the current soil moisture status, but are not necessarily able to explain foe causes of heterogeneity' and foe mode of intervention. Numerous examples show that without a real knowledge of our soils, even the most sophisticated tools are not useful.

By identifying areas with different yield potentials and mapping them, areas of high heterogeneity can be mapped with a relatively small sample size. In many cases, topography models are not included in methods for delineating areas of different yield potentials, even though topography is one of the primary factors in roil development.

4) indicate the problem to be solved theinvention:

The aim of the project is to develop a sampling algorithm based on a wain model and derived data that gives the right representativeness by taking the smallest sample size

S) Solving the problem: hi the process, the areas were mapped with the different methods used In the current precision services for the agricultural fields under study, with high point density. The relevance of the data in terms of yield potential was analysed and foe most important variables were selected as foe basis for foe Latin Hypercube-based sampling point assignment

1 b) a detailed description of one or more embodiments of the invention justifying the scope of protection :

L a terrain model, which can remote sensing, by collecting logged data from RTK navigation machines, or from publicly available sources.

2. In the second step, the derived data of the topography model are produced, such as slope angle_, topographic wetness index, channel network, relative slope position, elevation

3. The data are subjected to principal component analysis and then transformed into principal components to reduce dimensionality and decorrelation between variables.

4. The database is randomly sampled fifty times using the Latte Hypercube method to select the sampling points, which are clustered and the final sampling design is developed

S. after examining the sample points, the random forest algorithm is used to extend the sample points to the entire table

2 Kfo/t/ retting Yield mapping was perforrnied through a John Deere harvester m the years of 2017- 2020.

EC jfcsmsren- A contact EC scanner was used in Sis study, tint scanning was performed with a

EAfSeassm 2 Non performed during different time periods, fee same branded eearmw was used, but the scanning was performed by different operators. Swath width was 30 meters for both surveys.

A DJI Phantom 4 m was used to create a detailed DEM model of the area. The

Srrtsfi tie images ESA Sentinel 2. images were used «> derive NDVI and MSB! indexes. NDVI indexes were derived from several periods, and were averaged after of each

Intensive soil Surrey An intensive soil survey was performed, with soil profile descriptions in every Vi hectare to a depth of I meter. For the profile descriptions fee World Reference Base for Soil R«&ouroes (WRB.2015) was used. Calcic, Gamble, Mollie, Ochric_# Argfc horizons were identified and described, along with coiluvk soil material and surface soil horizon Mansell colour, which wax translated into RGB triplets for furtiier analysis.

Besides lhe field description iaboratOft analysis was performed for the following properties; pfr (KC1), Organic tiMU$er, Plsstfol^, CaCO? ermtent, Soil Etetirical

DWripfetoflf «t*w method* uwd tn theetu^t

10 most influential properties

♦ Slope

♦ TW1

♦ Potassium

♦ Topsoil colour

» Colluvic material

♦ Calcic horizon

♦ CaCO3 con tent

♦ Channel Network

♦ Relative slope position

♦ pH (KCI)

7.) An exphnatioo of how the invention is susceptible of industrial application, if is not apparent from the nature or description of the invention:

As described in the previous points, the applicabtlily in agriculture is supported,

S) Description of the beneficial effects of the invention in relation to the state of the art a) an economical estimation procedure based on technical parameters collected as by- products, free of charge, b) Indirectly, it can also help to estimate other soil parameters e) can be produced as a fraction of the current mapping costs d) has a higher data content than cmam nmp^ng methods

Patent claim(s):

1. Development of a cost-optimised soil sampling method to support precision farming

3

Claims

Patent claim(s):

1 , Development of a cost-optimised soil sampling method to support precision farming

4