N. Branisavljević, D. Prodanović, M. Arsić, Z. Simić, J. Borota
Advances in measurement equipment and data transfer enabled easy and economic automatic
monitoring of various hydro-meteorological variables. The main characteristic of such
automatic monitoring systems is that they do not rely on human activities, but only on
electronic devices. Even if those electronic devices are of highest quality and accuracy, and
properly tuned to specific problem, the reliability of measured values relyieson many other
factors and unexpected or undesired occurrences, like modification of measurement microlocation,
power supply shortages or surges, etc. The sampled and acquired data values have to
be additionally checked, validated and sometimes improved or cleared before further use. This
paper presents an innovative approach to data validation and improvement through the
framework generally applicable to all hydrological data acquisition systems. The proposed
framework can incorporate any number of validation methods and can be easily customized
according to the characteristics of every single measured variable. The framework allows for
the self-adjustment and feedback to support self-learning of used validation methods, same as
expert-controlled learning and supervision. After data validation, for low-scored data, its value
quality can be improved if redundant data exist, so framework has the data reconstruction
module. By applying different interpolation techniques or using redundant data value the new
data is created same as accompanying metadata with the reconstruction history. After data
reconstruction, the framework supports the data adjustment, the post-processing phase where
the data is adjusted for the specific needs of each user. Every validated and sometimes
improved data value is accompanied with a meta-data that holds its validation grade as a quality
indicator for further use.