Data science in water resources management

IDAEA is an environmental science institute dedicated to future challenges linked to climate change and water scarcity.

The Department of Geosciences investigates the hydraulic, chemical, thermal and mechanical processes associated with hydrogeology. Participates in the development of numerical and mathematical models and modelling techniques for complex processes. In this sense, it has developed multiple innovative applications for efficient data management.

These include an application to create geological cuts from sounding data; the use of artificial intelligence to predict the levels of rivers, aquifers, and similar physical environments; a program for the calculation of hydraulic parameters from hydrogeological data combined with the integration of neural networks; in addition to other geospatial applications based on the use of artificial intelligence, big data and the advanced processing and processing of big data.

These methodologies arise from the need to properly manage the amount of data intake implied by the current digitalization of the sector, a growing trend in recent years. To this problem must be added the need for an efficient and immediate management that can be achieved with these open-access applications.

Responsible entity

The Institute for Environmental Assessment and Water Research (IDAEA) is an institute of environmental sciences of the Spanish National Research Council (CSIC) founded in 2008 in Barcelona and dedicated to the study of the human footprint in the biosphere.

https://www.idaea.csic.es
IDAEA-CSIC’s work focuses on two of the environmental challenges of our time: the preservation of water quality and availability and air quality, guided by the principle that our scientific understanding of current threats to global ecosystems is best addressed from a holistic perspective.

The Institute stands out in the analysis of organic and inorganic pollutants and their impact on ecosystems, the study, modelling and management of water resources, the development of algorithms in different scientific fields and the study of inhalable particles and toxic gases.

Detailed explanation

A responsible and efficient management of water resources requires knowledge to face the possible conditions and problems related to water.

It is important to integrate conceptual knowledge of water bodies. For this it is needed; i) know the system (conceptual model), usually with fieldwork and georeferencing using geographic information systems (GIS); ii) a monitoring network that provides information on the status of the resource and validates the conceptual model.

Large volumes of data must be collected and stored efficiently in order to be consulted.

In order to relate the conceptual models with the treatment of the data generated by the monitoring, a series of platforms and applications appear to transfer data, integrate them into database systems in order to perform a correct analysis and interpretation and make a more efficient management.

Creation of geological sections: “Geopropy Tool”

It allows the creation of three-dimensional geological cuts using an open source in Python. Profiles are generated from polling data.

Water Level Prediction Tool

This tool using machine learning is able to show the relationships between variables in a certain time series within a hydrological model. In some hydrological models it is achieved from real-time monitoring, after an upstream condition (rainfall episodes, excess extractions …) predict the short-term behavior downstream.

Hydraulic parameter calculation tools

With granulometry, the permeability of the terrain can be determined from a series of analytical formulas. Meddiante artificial intelligence (neural networks) it is possible to identify which formula describes the reality in each case in a more detailed way. Hydraulic parameters are predicted from “cheap” data efficiently with the complementation of empirical formulas and neural networks

Hydrochemical indicators

The use of machine learning is used to find the chemical components that indicate a certain group of very specific compounds or high analytical cost.

Geospatial models

Geospatial maps are generated not based on physical processes. An example is its application in the Andes, a determination of the isotopic composition of rainwater throughout the territory was determined continuously in order to concisely evaluate the recharge of the aquifer without direct measures of it, in a complex management situation.

Meteorological data provide data on volumes and rainfall. The isotopic data are collected from different analytics and campaigns carried out in the territory that are usually not obtained systematically and / or continuously. Two models are generated that are integrated into one. The first represents the meteorological distribution that overlaps the timing and value of the analytics. The integrated model allows calculations of the isotopy of the waters that are recharging the aquifers.

Managing large volumes of data

To deal with a large volume of data, special database management platforms and data calculation and processing hardware are needed.

IDAEA has applied this methodology in radar interferometry to measure ground deformations. The images that are generated contain millions of pixels, generating hundreds of images. To deal with this volume of data, the data is filtered and then quantified and interpreted. To be able to perform this task requires a more capable system than Excel or a database and hardware with greater computing capacity.