Landslide Inventory Maps (LIMs) are the simplest tool to report the spatial distribution of landslides in a territory. They can be prepared using different techniques and base data (e.g. remote sensing images), each bringing intrinsic limitations and potential sources of mapping errors, hence affecting the overall accuracy and reliability.
LIMs are a precious source of information for any subsequent analyses in landslide research (e.g., land management and planning, model training and validation, susceptibility, hazard, and risk assessment, among others). A common operational assumption carried out when using such data is to consider them as “correct”, which results in transferring/propagating the mapping error(s) to the subsequent products.
Recent research works have defined the quality of LIMs as the result of three factors: geographic accuracy, thematic accuracy, and completeness/statistical representativeness. Geographic accuracy refers to the location, size, and shape of each landslide reported in the LIM. Thematic accuracy refers to the consistency of attributes assigned to each landslide in the LIM (e.g. classification, degree of activity, age/date of occurrence, among others). Completeness refers to the ratio of landslides reported in the LIM and the “ground truth”. Since the ground truth is hardly available, more recently the concept of statistical representativeness has been preferred, i.e. assuring that the statistical distribution of landslides reported in the LIMs is a statistically representative sample of the actual distribution of landslides in an area. Each of these aspects is currently under-explored in terms of evaluation/quantification/metrics, propagation, and handling/management in derivative maps.
Within this general framework, this session welcomes contributions specially focused on (but not necessarily limited to) the following topics:
• Definition of metrics (numeric, heuristic, morphometric, etc.) for the evaluation of mapping accuracy, errors, and uncertainty;
• Statistical modelling of mapping errors;
• LIMs quality assessment methods;
• Impact of error propagation in maps obtained from LIMs, including training of machine learning and/or AI-based detection algorithms, susceptibility models, hazard and risk assessment;
• defining links between LIMs quality and use limitations.
In contributions, all methods for the preparation of landslide inventories are welcome, from manual to semi- and fully automated.
LIMs are a precious source of information for any subsequent analyses in landslide research (e.g., land management and planning, model training and validation, susceptibility, hazard, and risk assessment, among others). A common operational assumption carried out when using such data is to consider them as “correct”, which results in transferring/propagating the mapping error(s) to the subsequent products.
Recent research works have defined the quality of LIMs as the result of three factors: geographic accuracy, thematic accuracy, and completeness/statistical representativeness. Geographic accuracy refers to the location, size, and shape of each landslide reported in the LIM. Thematic accuracy refers to the consistency of attributes assigned to each landslide in the LIM (e.g. classification, degree of activity, age/date of occurrence, among others). Completeness refers to the ratio of landslides reported in the LIM and the “ground truth”. Since the ground truth is hardly available, more recently the concept of statistical representativeness has been preferred, i.e. assuring that the statistical distribution of landslides reported in the LIMs is a statistically representative sample of the actual distribution of landslides in an area. Each of these aspects is currently under-explored in terms of evaluation/quantification/metrics, propagation, and handling/management in derivative maps.
Within this general framework, this session welcomes contributions specially focused on (but not necessarily limited to) the following topics:
• Definition of metrics (numeric, heuristic, morphometric, etc.) for the evaluation of mapping accuracy, errors, and uncertainty;
• Statistical modelling of mapping errors;
• LIMs quality assessment methods;
• Impact of error propagation in maps obtained from LIMs, including training of machine learning and/or AI-based detection algorithms, susceptibility models, hazard and risk assessment;
• defining links between LIMs quality and use limitations.
In contributions, all methods for the preparation of landslide inventories are welcome, from manual to semi- and fully automated.