Research project SR/00/302 (Research action SR)
CONTEXT AND OBJECTIVES
Given the expected increase in extreme events due to climate change, more drought events can be expected in the future. These events have often devastating impacts on society and the environment. Therefore, adequate monitoring of these events is of utmost importance within disaster management. Remote sensing can provide important information, though does not allow for a complete assessment of droughts as (1) only measurements of the surface are obtained and (2) the spatial and temporal resolutions are often too coarse. Combining remote sensing with land surface models is generally opted for, and is already in place in many drought monitoring systems. However, several parts of these systems can be improved with respect to (1) the use of multiple sources of remote sensing data, (2) the modelling approach used and (3) the updating of models based on remotely sensed observations. If any of these components can be improved, a more precise monitoring and modelling can be expected, and therefore enhanced predictions of droughts can be made.
The objective of HYDRAS+ is to work on these three domains and to demonstrate the benefits of the joint assimilation of several remote sensing sources in land surface models. It furthermore aims at assessing whether conceptual models can be used instead of complex and computation-expensive land surface models. If such models can be used, a faster computation of droughts at very large scale becomes possible.
HYDRAS+ does not foresee to develop an alternative drought monitoring system, but aims at developing methodologies that can improve many of the currently existing systems. Any improvement in the currently available systems will have important positive consequences with respect to disaster management as it will allow for an improved management of resources, reducing the number of casualties.
METHODOLOGY
HYDRAS+ will focus on three main aspects: (1) remote sensing imagery, (2) land surface modelling and (3) data assimilation. The methodologies used in each of these aspects involve:
a. Downscaling, bias correction and uncertainty characterization of the remote sensing data
Different sensors provide data at different resolutions. In order to use these data, two options are possible: assimilation at the scale of the observations, or assimilation at the scale used by the model. In this project, the second approach will be investigated in detail, where a new framework, based on copulas, will be developed and optimized for scaling and/or merging data at the model resolution. Major attention will be paid to bias-correction, as assimilation requires bias-free estimates of model states, and to the assessment of the uncertainty in downscaled products, as the latter information is of crucial importance in data assimilation systems.
b. Hydrologic model integration
The hydrologic model integration involves the integration of a radiative transfer model and a land surface model, being both a physically-based model and a conceptual model. Such coupled system is necessary for assimilating level-1 products (backscatter and brightness temperatures).
c. Data assimilation integration
Different data assimilation algorithms will be implemented, based on the Kalman Filter. Major attention will go to the implementation of a dual state-parameter estimation technique that should allow to update model parameters every time observations become available. It is hypothesised that such approach is necessary when simplified conceptual models are used instead of physically-based land surface models.
d. Data assimilation experiments
The model framework is going to be used for a set of dedicated assimilation experiments, which should demonstrate the benefit of this approach for water managers, and organizations dedicated to climate research and/or reporting such as ECMWF, JRC, or IPCC. The assimilation experiments will enable to provide recommendations on which algorithms of current early-warning systems may potentially be improved by implementing the developed algorithms within the project.
e. Analysis of extreme events and development of an early-warning system
The developed framework that optimally makes use of remotely sensed observations will be assessed with respect to currently available drought monitoring/forecasting systems in order to make recommendations for improving these operational systems. Therefore, an existing drought monitoring/forecasting system will be used as benchmark and methodologies developed in the framework of HYDRAS+ will be implemented in this system in order to validate the merit of the proposed methodologies.
EXPECTED SCIENTIFIC RESULTS
Several scientific results are expected with respect to:
- Downscaling remote sensing data to the model resolution
- Conceptual modelling of the mass and energy balance at large scale
- Data assimilation of a suite of remote sensing observations
- Merging of data from different sensors
- Dual state-parameter estimation
- Recommendations with respect to the use of remote sensing data in land surface models
- Recommendations with respect to improving operational drought monitoring/forecast systems
EXPECTED PRODUCTS AND SERVICES
The main products that can be expected are new methodologies and insights in the application of remotely sensed observations in land surface models in order to improve drought monitoring. The products that will initially be delivered will be contributions to workshops/conferences/symposia and publications in peer-reviewed journals. However, end users (mainly organizations that provide drought forecasts and space agencies) will be informed on the advances made through this project, and how these may benefit their services.
