Share this page

NEUROPUBLIC’S CHALLENGE ON GAIA SENSE DATA

Posted by  Wednesday, 07-06-2016
(5 votes)

Introduction

Smart farming (or precision farming) is an environmentally-sound approach that supports farmers in increasing their production and at the same time minimizing costs by controlling the application of inputs (such as water for irrigation, fertilizers, pesticides and other agrochemicals) not only in terms of quantity but also defining different amounts in different parts of a field. Smart farming is based on the collection, management and analysis of data collected from various sources (e.g. from farmers and farm sensors, environmental measurements, remote sensing, satellite data etc.) that are combined with knowledge and allow for informed decision making regarding agricultural practices.

 

GAIA Sense is an infrastructure developed by NEUROPUBLIC, which aims at supporting the launch of a range of innovative smart farming services that will offer next generation advice to farmers. GAIA Sense consists hardware IoT devices called GAIAtrons, along with software systems which cater for the data acquisition, knowledge extraction, decision support and knowledge dissemination. GAIAtrons are telemetric environment sensing stations which are installed in the field and record atmospheric and soil parameters. The data collected from these stations are combined with data from a number of other sources and with the help of a successful blending of technological and human components, they are transformed to advice. This advice is offered through innovative smart farming services called ‘GAIA InFarm’ (GAIA intelligent farming). GAIA InFarm services offer advice and/or automation on fertilization, irrigation, pest management, traceability, hazard warnings and more.

 

Overview of a GAIAtron telemetric station

The Challenge

NEUROPUBLIC’s GAIAtrons collect and store datasets consisting of data recorded by the sensors which are integrated in the stations. However, in some rare cases these datasets may be incomplete due to various reasons, including but not limited to:

  • Environmental disasters (thunders, earthquakes, etc.)
  • Communication carrier failures
  • Human intervention (thieves, vandalism etc.)
  • Animal intervention (birds’ nests, etc)

The aforementioned parameters may cause abnormal operation of the station’s sensors leading to missing data and recording of irrational/abnormal values. In order for the process to work properly, datasets need to consist of complete and accurate values. This is a common issue identified in workflows that include sensor logging.

The expected workflow will consist of the following steps:

-        We have exported an actual dataset from our database consisting of data gathered from 5 GAIATron telemetric stations in the area of Stimagka, Corinthia, Greece between 8/3 – 5/7/2016; the different stations can be identified through the different coordinates that accompany each measurement.

-        This dataset has been processed in order to better illustrate the various problematic scenarios, mentioned above, that compromise not only its integrity but also its precision.

-        Measurements in the dataset consist of the following: (i) Date of the measurement, (ii) Longitude and (iii) Latitude of the station, (iv) Temperature and (v) Relative Humidity.

-        Participants are required to perform a normalization of the data, referring to ensuring that data only include rational values (i.e. values making sense).

-        Participants are also required to work on a solution for completing the missing values so that there is a consistency in the time series.

-        Last but not least, participants are required to produce and present a graph for each step, showing the evolution from one step to the other; more specifically, a graph representing the initial problematic data, a graph for the normalized and a graph from the filled gaps are expected as outcomes.

Participants are free to select any approach they consider as most appropriate for completing the aforementioned steps, such as selecting the most appropriate methods for normalizing data values and selecting data sources for completing the missing ones. They are also invited to take a step further and work on anything that will provide added value to the service; always with guidance from the NEUROPUBLIC mentors.

Telemetric Station Data - excel sheet enclosed

Read 2214 times Last modified on Thursday, 07 July 2016 10:52
Share »

Leave a comment