| Summary: | The earth environment is undergoing drastic changes at a global scale, and many spectacular climatic accidents have recently drawn much attention to the ecology and monitoring of the environment.
The SensorScope developed in EPFL, Switzerland is a new generation of portable and inexpensive measurement system based on a wireless sensor network with built-in capacity to produce high temporal and spatial density measures. The system has been successfully used in quite a number of deployments and has gathered hundreds of megabytes of environmental data from a large range of research domains and aims at providing the fundamental tools that will help in taking care of the earth.
However, a price to pay for using simple sensing devices is that the gathered measurements are not as reliable and accurate compared to traditional expensive and heavy weather stations. Consequently, it is desirable and sometimes necessary to preprocess and improve the quality of the raw data from the sensor networks before presenting them to environmental researchers. In particular, the problem addressed in this project is the detection of outliers and interpolation of the missing values. Due to various reasons, some sensors may possibly fail and it is highly possible that these failed sensors are not capable of reporting this situation to the central weather station. It is therefore vital to detect the data outliers and determine the missing values in the sensor measurement data in order to improve the robustness of the overall system.
|
|---|