Launch of SoilPass, an interactive knowledge portal providing information on the hygienic status of Czech agricultural soils

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00027049%3A_____%2F24%3AN0000087" target="_blank" >RIV/00027049:_____/24:N0000087 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/00027049:_____/24:N0000086

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Launch of SoilPass, an interactive knowledge portal providing information on the hygienic status of Czech agricultural soils

Popis výsledku v původním jazyce

The presentation and practical demonstration will introduce the main output of the project SS03010364, which is the interactive knowledge portal „SoilPAss = „Soil Pollution Assessment“ (https://soilpass.vumop.cz/). The programmed user interface presents a procedure to extract the required and understandable information on the expected soil hygiene status from the source data available in the Czech Republic - i.e. it simulates an expert practice and uses existing data sources to provide an understandable output, without free availability of individual data of each data keeper. At the same time, the interactive portal takes an innovative approach to providing full information about the results of the predictive spatial model, including the treatment of the uncertainty of the results themselves, which is not common practice in the publication of map results. A downloadable meta-information record is provided for each predictive chemical compound/element content, containing comprehensive information on the input data, the process of building and evaluating the predictive model, allowing the map-building process to be reproduced. A quantile random forest algorithm from the family of machine learning methods was used in the map creation process, which has the advantage of being able to estimate the quantiles of the modelled variable, thus obtaining the distribution parameters of the probable value of the target variable. Therefore, the model allows for a local assessment of the model‘s uncertainty usingthe width of the prediction interval, which complements the overall model accuracy quantified from the differences between predicted and actual measured values in the dataset subtracted from the primary sample before training the model. The principle of machine learning is to extract („learn“) valuable information from the training data and apply the learned patterns to spatial predictive modelling outside the feature space of training data. In the case of (geo) chemical predictive mapping, these patterns are derived from the relationship between soil elemental/substance contents and other descriptive environmental variables relating to soil-geological, climatic, geomorphological conditions or parameterising sources and inputs of contaminants to the environment. The above algorithm has been used to train predictive models not only for estimating concentrations of individual elements/compounds in soil, but also probabilistic indicator models for exceeding relevant safety thresholds for soil contamination. The developed interactive interface for accessing digital maps contributes significantly to improving the current knowledge of the hygienic status of soils, as nocomparable information tool has been developed in the Czech Republic.

Název v anglickém jazyce

Launch of SoilPass, an interactive knowledge portal providing information on the hygienic status of Czech agricultural soils

Popis výsledku anglicky

The presentation and practical demonstration will introduce the main output of the project SS03010364, which is the interactive knowledge portal „SoilPAss = „Soil Pollution Assessment“ (https://soilpass.vumop.cz/). The programmed user interface presents a procedure to extract the required and understandable information on the expected soil hygiene status from the source data available in the Czech Republic - i.e. it simulates an expert practice and uses existing data sources to provide an understandable output, without free availability of individual data of each data keeper. At the same time, the interactive portal takes an innovative approach to providing full information about the results of the predictive spatial model, including the treatment of the uncertainty of the results themselves, which is not common practice in the publication of map results. A downloadable meta-information record is provided for each predictive chemical compound/element content, containing comprehensive information on the input data, the process of building and evaluating the predictive model, allowing the map-building process to be reproduced. A quantile random forest algorithm from the family of machine learning methods was used in the map creation process, which has the advantage of being able to estimate the quantiles of the modelled variable, thus obtaining the distribution parameters of the probable value of the target variable. Therefore, the model allows for a local assessment of the model‘s uncertainty usingthe width of the prediction interval, which complements the overall model accuracy quantified from the differences between predicted and actual measured values in the dataset subtracted from the primary sample before training the model. The principle of machine learning is to extract („learn“) valuable information from the training data and apply the learned patterns to spatial predictive modelling outside the feature space of training data. In the case of (geo) chemical predictive mapping, these patterns are derived from the relationship between soil elemental/substance contents and other descriptive environmental variables relating to soil-geological, climatic, geomorphological conditions or parameterising sources and inputs of contaminants to the environment. The above algorithm has been used to train predictive models not only for estimating concentrations of individual elements/compounds in soil, but also probabilistic indicator models for exceeding relevant safety thresholds for soil contamination. The developed interactive interface for accessing digital maps contributes significantly to improving the current knowledge of the hygienic status of soils, as nocomparable information tool has been developed in the Czech Republic.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

40104 - Soil science

Projekt

<a href="/cs/project/SS03010364" target="_blank" >SS03010364: Systém na podporu rozhodování při hodnocení kvality půdy z hlediska obsahu rizikových látek v zemědělských půdách České republiky</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)<br>I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů