Data quality control in eco-environmental monitoring

Abstract

With the development of science and technology, a number of environmental issues, such as sustainable development, climate change, environmental pollution, and land degradation become serious. Greater attention has been attached to environmental protection. The government gradually launched some eco--environmental construction projects. In 1999, China begin to carry out the project of Grain-for-Green in the west, to improve the eco-environment, and it make some good effect, but there are some questions that still can not be answered. How about the new grass or forest? Where are they? How can we do in the future? To answer these questions, the government began to monitor the eco-environment, based on remote sensing technology. Geography information can be attained timely, but the issue of uncertainty has become increasingly recognized, and this uncertainty affects the reliability of applications using the data. This article analyzed the process of eco-environment monitoring, the uncertainty of geography information, and discussed the methods of data quality control. The Spot5 span data and multi-spectral data in 2003(2002) were used, combined with land use survey data at the scale of 1:10,000, topography data at the scale of 1:10,000, and the local Grain-for-Green project map. Also the social and economic data were collected. Eco-environmental monitoring is a process which consists of several steps, such as image geometric correction, image matching, information extraction, and so on. Based on visual and automated method, land information turned to grass and forest from cultivated land was obtained by comparing the information form remote sensing data with the land survey data, and local Grain-for-Green project data, combined with field survey. According to the process, the uncertainty in the process was analyzed. Positional uncertainty, attribute uncertainty, and thematic uncertainty was obvious. Positional uncertainty mainly derived from image geometric correction, such as data resource, the number and spatial distribution of the control points are important resource of uncertainty. Attribution uncertainty mainly derived from the process of information extraction. Land classification system, artificial error was the main factor induced uncertainty. Concept defined was not clear, and it reduced thematic uncertainty. According to the resource of uncertainty, data quality control methods were put forward to improve the data quality. At first, it is more important to choose appropriate remote sensing data and other basic data. Secondly, the accuracy of digital orthophoto map should be controlled. Thirdly, it is necessary to check the result data according to relative data quality criterion to guarantee GIS data quality.