Abstract ID: 558

Strategy for an Automatic Micrometeorological Data Cleaning

One of the challenges of a sensors network structure applied to micrometeorological stations in long term experiment is to maintain a methodological protocol for control data quality consistently and to use the appropriate mechanisms of correction for scenario of collections and data sources that generates anomalies, and support in decision-making process for maintenance, reducing the response time to failures. This work aims to determine technique for treatments of micrometeorological data and to understand the needs and feasibility of functional requirements for a possible system. Initially, it was studied the methods parsing, data transformation, integrity constraint enforcement, duplicate elimination and statistical methods, and existing tools AJAX, FraQL, Potter’s Wheel, ARKTOS and IntelliClean, for treatments of data. A Web platform was prototyped to implement the functionality of the user interface with diagnoses features for displaying the results of the tasks of cleaning and the pre-treatment process, such as merge and re-sequencing of data. The specific layers of application of the system and the tasks for transaction in databases are supported by (object) relational database system. The study showed that we lack tools that encompass functionalities to deal with all different types of anomalies. Therefore, an efficient system able to clean micrometeorological data must consider the methods and tools functionalities, together with computational environment that represents the various techniques in a single system. The implementation process affects the data quality showing that the system can respond positively and demonstrate that the methods and techniques adopted are recommended and that the applications fit the requirements revealing important aspects of the diagnostic functions and other features of the system.

Session:  Feedbacks to Climate - Land cover, surface hydrology, and atmospheric feedbacks.

Presentation Type:  Poster