Data quality, in general, is the degree of data usability in relation to a specific application purpose. Assurance of data quality is of growing importance in remote sensing, due to the increasing relevance of remote sensing data in planning and operational decision of public bodies and private firms, and the huge amount of digital services (or apps) that exploit RS data.
Different data quality dimensions exist according to the lifecycle phases of the remote sensing data: data acquisition, data storage, data pre-processing, processing and analysis and data visualization and delivery. Remote sensing data acquisition phase involves the following quality aspects: resolution, accessibility, spatial accuracy, temporal validity, accuracy and precision of the sensor calibration. Resolution is a multi-dimensional concept that includes the following dimensions: spatial resolution, temporal resolution, radiometric resolution, spectral resolution and temporal resolution. Temporal validity refers to the quality of an remote sensing data product in time, whereas spatial accuracy refers to the accuracy of the position of features relative the Earth.
Data storage includes the accessibility and completeness data quality dimensions. Accessibility includes both temporal and data accessibility. Temporal accessibility refers to the time delay between data acquisition and data delivery, whereas data accessibility refers to the availability of remote sensing data. Data completeness encompasses temporal completeness, i.e. completeness of a time series represented a phenomenon, thematic completeness, and spatial completeness which refers to the area coverage. Data preprocessing, processing and analysis phase includes consistency, completeness, temporal validity, resolution, radiometric and geometric accuracy, thematic and semantic accuracy. Thematic and sematic accuracy refers to the correctness of the remote sensing data product. The main quality dimensions of the data visualization and delivery include readability, completeness and temporal validity.
Different metrics can be used to assess the quality of the remote sensing-derived information, such as the root-mean-square error (RMSE) measuring the differences between the true and measured values of the phenomenon under investigation, confusion matrix used for assessing the classification performance, producer’s accuracy, user’s accuracy or Cohen kappa. The quality of the remote sensing data per se can be assessed using Peak Signal-to-noise Ratio (PSNR) or the Universal Image Quality Index (UIQI).
Different organizations are involved in the standardization of the image data and gridded data quality, including ISO/TC 211 ‘Geographic information/Geomatics’, Open Geospatial Consortium (OGC) or the Quality Assurance Framework for Earth Observation (QA4EO) developed by the Group on Earth Observation (GEO). These organizations are responsible for developing metadata standards that are further used by the remote sensing community to document the quality of the remote sensing data. According to the QA4EO, for example, all remote sensing data products need to be accompanied by a Quality Indicator (QI) which helps users assessing their fitness-for-use.