Towards Informed Use of Gridded Population Data: A User-Driven Selection Framework

Gridded population datasets play a pivotal role in a wide range of contemporary research and development, such as the distribution of aid, public health campaigns, as well as disaster risk management. However, the selection of the appropriate existing population dataset remains a non-trivial task, resulting in many practitioners choosing based on convenience or familiarity, rather than explicit use-case suitability.

In our contribution we present a user-requirement driven review of major gridded population datasets, in particular reviewing the wide array of the WorldPop suite, including their bespoke datasets, LandScan (HD), Kontur, Facebook HRSL, GPW, and GHS-Pop. We first consolidate key requirements of users in applied human-environment research and policy, based both on a literature review as well as key-informant-interviews of practitioners in the Humanitarian sector. Our synthesis reveals barriers to informed dataset choice, including scattered and inadequate documentation, limited uncertainty quantification and communication, and a lack of explicit suitability statements.

We then systematically evaluate, based on Riedler et al., 2025 (under review), how current existing datasets perform with respect to spatial granularity, temporal consistency, sensitivity to input data, the influence of settlement type on accuracy, and transparency of the product.

Based on the combination of both findings, we derive a set of generalised guiding questions for practitioners, as well as a decision tool for use-case specific dataset choice. We, furthermore, illustrate the effects of different dataset choices on down-stream applications and their potential impact on decision-making, as well as discussing alternative methods to establish population estimates and their suitability for studies and policies.

By shifting the perspective from dataset-centric descriptions to user-centred logic our review provides a foundation for operational decision-support and better understanding of gridded population products for domain agnostic users.