To facilitate environmental resource management of intensively populated countries like the Netherlands, integrated information systems which are capable of real-time monitoring of fundamental processes in the environment, as well as providing vital hazard warnings, are required. Traditionally, sensor networks covering various geographical and temporal scales are an important source of information for this task. They allow vast amounts of relevant information to be collected with a high temporal frequency for a network of point locations that are remote, inaccessible, or lack the necessary resources to acquire such information in a different manner [1
]. For example, ground water levels in the Netherlands are monitored through a network of 4,000 semi-automated groundwater wells [2
]. Recent developments in the miniaturization of electronics and wireless communication technology will enhance the opportunities of sensor networks for real-time monitoring of the natural environment [3
]. Next to in situ
sensor networks satellite remote sensing systems are also a key source of information for many applications. Although space based sensors have a superb spatial coverage, they can frequently incur a significant data delivery latency, have a poor signal to noise ratio, and possess coarse resolutions. However, for a comprehensive monitoring system to provide timely information, a combination of in situ
and space based sensors offers a synergetic configuration [4
]. In an integrated approach, the sensor observations provide data and information; scientific models use these data and produce predictive results which are provided to end-users to assist the decision making process [5
]. Although space-based and in situ
sensor data have not been integrated in a fully self-consistent way, advanced technologies of today make it possible to pursue more integrated approaches to environmental resource assessment [4
Although information technology is an important facilitator in this process, integrated information systems are often limited by interoperability problems due to individual components which cannot easily communicate with each other [6
]. To overcome this problem, efforts at the sensor network level are required which deal with issues such as fusion of sensor data and interoperability among networks and their connections to information systems. This effort should not only pay attention to the technical facilitation but also should include organizational and standardization aspects. The concept of sensor webs as introduced by Delin in 2002 [7
] “allows for the spatio-temporal understanding of the environment through coordinated efforts between multiple numbers and types of sensing platforms, including both orbital and terrestrial and both fixed and mobile.” Compared to sensor networks, sensor webs are unique in their feature that sensors communicate with each other, share information with other sensors and are aware of their environment. Communication between the sensor web and user can be in two directions: the user receives information from the sensor web but can also send instructions to it [8
]. In an initiative called sensor web enablement (SWE), the Open Geospatial Consortium (OGC) has been developing a framework of open standards for exploiting web-connected sensors and sensor systems of all types [9
]. The available services include access to sensor measurements, retrieval of sensor metadata, controlling sensors, alerting based on sensor measurements and automatic processing of sensor measurements. Although the developed SWE concepts are being applied in a broad range of environmental domains (e.g., hydrology [9
], ecology [11
], risk management [6
]), only a limited number of studies [4
] describe the combined use of space-based and in situ
sensor sources in a sensor web based approach.
Monitoring of terrestrial plant productivity is one of the key parameters in environmental resource management as it provides information on potential food resources and sources of wood for construction, fabrication and fuel [14
]. For example, early indicators of crop health status are very valuable because management decisions can be made both by farmers at the field level but also by governments at the regional level to mitigate the economic and social impacts of yield variability. In addition, as climate and terrestrial ecosystems interact with and influence each other, vegetation productivity is also used as indicator for climate change effects [15
]. Plant productivity is calculated as Net Primary Productivity (NPP), the difference between Gross Primary Productivity (GPP) and plant autotrophic respiration (Ra), which is the net carbon fixed by vegetation through photosynthesis [16
]. At the global scale, terrestrial plant productivity is one of the most-modeled ecological parameters, with models that differ markedly in approach and complexity often yielding comparable estimates [17
]. For example, a global 8-day MODIS product (MOD17A2) is available which models GPP at a 1 km resolution using a light-use efficiency model [18
]. However, for regional applications (e.g., monitoring crop productivity) both the spatial and temporal resolution of this product is too coarse. In addition, this product has been developed for a global scale which means that several of the input parameters of the estimation model do not account for the local heterogeneity of land use and meteorological parameters [19
]. Increased availability of real-time sensor data at the local scale could increase the understanding and detection of vegetation status of heterogeneous landscapes. The added value of a sensor web based approach would be that multi-source sensor streams can be integrated in the model. Standardized modeling results can be presented to the end-user and will supply information on the spatial distribution of vegetation productivity both in the actual situation (nowcasting) and for the near future (forecasting) [22
In this study we have developed a sensor web based approach which combines earth observation and in situ sensor data to derive regular products for vegetation productivity on a regional scale level. The approach is implemented in an automated processing facility which makes the products available through a dynamic web mapping service (WMS). Within the study a prototype application has been developed which provides daily maps of vegetation productivity for regional to national scale in the Netherlands. In the results section of this paper the spatial-temporal development of GPP over the Netherlands is presented. Finally, we assess the validity of the modeling results and discuss the limitations and opportunities for further development of the presented methodology.
4. Conclusions and Outlook
In this article we have presented a sensor web based approach which combines earth observation and in situ sensor data to derive near real-time vegetation productivity products. A prototype application for monitoring GPP over the Netherlands was successfully developed and implemented within an automated processing facility. Continuous GPP maps are provided to the user through a web mapping service which not only provides functionality for spatial analysis but also includes functionality to present time-series for selected locations.
In order to achieve an added value for end-user applications using increasingly available real time earth observation data, they need to be combined with in situ sensor data and environmental models to derive higher level products relevant for environmental resource management. In this study, interoperability between sensor data streams and connection with the information system was achieved by using open-source standards for SWE and WMS. For example, meteorological data were obtained in a standardized way through a Sensor Observation Service as developed within OGC-SWE. Use of common standards is an important requirement for upscaling of the developed facility, both in terms of number of sensors and inclusion of new sensor types.
Further development of the presented approach will focus on the establishment of an integration platform for near real-time assimilation of sensor data sources into simulation models. Focus will be on integration of multi-source sensor data streams and the opportunities for remote sensing data fusion to improve spatial resolutions to relevant management units (< 30 m).