Optimizing Land Cover Classification Accuracy for Change Detection: A Combined Pixel-based and Object-based Approach in a Mountainous Area in Mexico

Applied Geography

Applied Geography, Volume 34, May 2012

Jesus Aguirre-Gutiérrez, Arie C. Seijmonsbergen, and Joost F. Duivenvoorden


  • The land cover classification accuracy is optimized with a combined approach.
  • The optimized classification is the input in a change detection analysis.
  • This method produces higher classification and change detection accuracies.
  • This approach can potentially be applied to other mountainous regions.

“Inventories of past and present land cover changes form the basis of future conservation and landscape management strategies. Modern classification techniques can be applied to more efficiently extract information from traditional remote-sensing sources. Landsat ETM+ images of a mountainous area in Mexico form the input for a combined object-based and pixel-based land cover classification. The land cover categories with the highest individual classification accuracies determined based on these two methods are extracted and merged into combined land cover classifications. In total, seven common land cover categories were recognized and merged into single combined best-classification layers. A comparison of the overall classification accuracies for 1999 and 2006 of the pixel-based (0.74 and 0.81), object-based (0.77 and 0.71) and combined (0.88 and 0.87) classifications shows that the combination method produces the best results. These combined classifications then form the input for a change detection analysis between the two dates by applying post-classification, object-based change analysis using image differencing. It is concluded that the combined classification method together with the object-based change detection analysis leads to an improved classification accuracy and land cover change detection. This approach has the potential to be applied to land cover change analyses in similar mountainous areas using medium-resolution imagery.”

Sensor Web Services for Early Flood Warnings Based on Soil Moisture Profiles

XXII ISPRS Congress, 25 August to 01 September 2012, Melbourne, Australia

Thomas Brinkhoff and Stephan Jansen

“Recent disastrous floodings have demonstrated the demand for early flood warnings. This need will be enforced by the expected climate change that probably leads to more torrential rain. In general, flood forecasts concentrate on large rivers considering water gauges. However, there exist a large number of smaller drainage areas with a size of 100 to 1000 square kilometers. In such areas, disastrous drainages may happen after heavy rain that cannot reasonably be observed by gauges because the water level changes very fast without having enough time for warnings.

WEBBOS web client depicting the sensors of a sensor network

WEBBOS web client depicting the sensors of a sensor network.

“Thus, the creation of early flood warnings for small drainage areas is still an unsolved challenge. One promising approach for solving this problem is the observation of the soil humidity because that measure decides how much of the precipitation will drain off. WEBBOS is a joint project of two universities funded by the German Federal Ministry of Education and Research. It has the objective to build a web-based sensor system for early flood warnings by measuring soil moisture profiles. Within this project, special sensors measuring the soil humidity in different depths have been developed and successfully tested. These sensors were integrated into sensor nodes that form local sensor networks. Such a sensor network should be located in an area with a high variety of soil humidity, which typically happens near to small rivers on sloping terrains.

“Like in other early-warning and disaster management systems, it is important to incorporate the sensor measurements into a geospatial information service that allows managing sensor data. This encloses the visualization of the sensor measurements and the control of the soil humidity sensors as well as the detection and processing of alarm events. Interoperability is an important issue for such systems. Therefore, the Open Geospatial Consortium (OGC) started the Sensor Web Enablement (SWE) initiative 5 years ago and specified several services and data models in the meantime. The 52° North Sensor Web framework is the most prominent implementation of SWE specifications.

“WEBBOS implemented its information system using those specifications and the 52° North SWE framework. Many standard requirements can be solved by following such an approach without additional efforts. However, often there are important demands that need more sophisticated solutions. Such cases and their solutions for the WEBBOS project will be discussed in the following:

“The support of early flood warnings based on soil moisture profiles requires temporal 3D measurements and 3D models of soil humidity for each sensor network: Only the relation and difference between the humidity in different depths and their development over the time allow detecting relevant events. The temporal aspect is sufficiently covered by the current SWE specifications. However, these specifications and their implementation are mainly 2D-oriented. The latter is also caused by the underlying spatial database systems which are restricted to simple feature geometries. For solving this challenge, a layered sensor model has been developed and applied to the SWE implementation. This enables us to cope with these restrictions.

“An important issue for such systems is the detection of potentially dangerous situations. This requires the description of spatio-temporal events and their detection. For this purpose, the OCG currently proposed (as discussion papers) the Sensor Event Service (SES) and the Event Pattern Markup Language (EML). The WEBBOS project demonstrated that these specifications can be successfully applied for the evaluation of soil moisture profile measurements.

“Standard visualization tools are often not sufficient for an adequate visualization of sensor data. However, the use of SWE services restricts and formalizes the syntax and semantics of the sensor measurements and of sensor metadata. Therefore, open toolkits are well-suited for developing flexible solutions that can be easily adapted to the requirements of special applications. In case of the WEBBOS project, we followed this approach by extending the popular JavaScript map library “OpenLayers”. We introduced subclasses for sensors and sensor data that can be flexibly used in standard vector layers of OpenLayers. The web client allows retrieving sensor data and metadata by the Sensor Observation Service (SOS) as well as controlling and evaluating the SES.

“Overall, the WEBBOS project demonstrated the strengths and deficiencies of the current SWE services and provided feasible solutions for sensor networks measuring soil moisture on the top of standard geospatial software packages. ”