Future Developments

Dynamic data ingestion

Although LHAT can theoretically take dynamic inputs, certain datasets may need expert judgement or statistically derived thresholds to determine the contribution of a variable (eg. rainfall) to the susceptibility of landslides occurring. Determining thresholds may be challenging as one region may have different thresholds compared to others. However, this is important to address in order to support Early Warning Systems and enable rapid action for eg. providing humanitarian aid. Other types of dynamic data to consider also include InSAR data, which can help detect deformations on the surface. By detecting significant deformations, this will contribute to the risk assessment.

Model refinement

Currently, model parameterisation is performed automatically using GridSearch by providing the best combination of model parameters that generate the highest accuracy. A caveat of this process is that this may lead to overfitting. Future developments will include looking into an alternative factor to rank the performance of the various model parameters.

Secondly, model refinement can also be performed by providing data on the return period of landslides. However, this is a challenging activity as many publicly available datasets either scarce, incomplete or insufficiently representative of landslide occurrences in such areas. Although it remains a priority to generate return periods of landslide occurrences, this is largely hindered by lack of sufficient data.

Do you have a suggestion on potential data sources that can help us? Feel free to contact one of us (giorgio.santinelli@deltares.nl or robyn.gwee@deltares.nl)

Making an executable

To improve accessibility to LHAT, an executable and/or an API (Application Programming Interface) can be made to enable inter-operability between other tools and improve access by non-python users.

Impact Assessments: Damage functions and coupling with impact assessment tools

The output of LHAT is a machine-readable format, which can be read by other Deltares impact assessment tools such as RA2CE, Delft-FIAT Accelerator and Ri2de. Next logical step is to generate impact assessments from these tools. To empower the analysis, damage functions of landslide hazards can be made as an estimate of potential economic damage.