Published research

This systematic review synthesises fifty-one peer-reviewed articles to introduce a framework for measuring the quality of stakeholder collaboration in environmental governance. The SNPI links five social indicators — relationship building, power sharing, social learning, trust building, and potential for collective action — to quantitative network metrics such as density, centrality, and reciprocity. It provides the social measurement backbone for all of Dialectik's integrated assessment work.

This paper documents a transdisciplinary field study in El Salvador's coffee sector, bringing together farmers, government officials, NGOs, and researchers to co-design a climate adaptation research agenda. It maps thirty-nine stakeholder organizations across five categories and demonstrates how local knowledge and scientific methods can be integrated through stakeholder encounters. The methodology tested here provides the blueprint for The Coffee Initiative's country-level case studies.

The published basis for the final step of STAND — turning drought dynamics into a forecast of agricultural impact. Most machine-learning yield models feed on drought indices, which describe severity but not extent. This work uses the drought area itself as the input: how much of a region is under stress, aggregated over one to twelve months. Tested against nearly fifty years of rice yield records (1967–2015) across three regions of eastern India, neural network models consistently beat the polynomial-regression baseline, with prediction error as low as 48.1 kg/ha. The demonstration is Indian rice; the method is the one Dialectik applies wherever drought and a harvest meet.

Using daily evaporation and runoff simulations from a distributed hydrological model of a catchment in southeast Mexico, this study evaluates how well different meteorological, agricultural, and hydrological drought indicators — including SPI, SPEI, ETDI, SEDI, and SRI — identify drought onset and duration across time steps. The results show meteorological indicators alone miss or lag many drought events, and recommend combining them with hydrological and agricultural indicators for reliable catchment-scale drought monitoring.

A technical synthesis of the environmental data pipeline that underpins The Coffee Initiative. Building on three peer-reviewed foundations — the STRIVIng toolbox for spatiotemporal climate projections (Diaz et al., 2019), spatial drought-area modelling as input for machine-learning crop-yield prediction (Diaz et al., 2026), and the Lempa transboundary basin climate-risk index (Koshnazar et al., 2021) — the report walks through how precipitation and temperature variability, spatial drought characterisation, and integrated hazard-vulnerability-resilience analysis combine into a decision-ready risk index, with agroforestry management scenarios layered on top.

Spanish-language field report (2023) documenting the participatory method, stakeholder encounters, and co-designed adaptation research agenda that emerged from fieldwork in El Salvador's coffee sector. Released under a Creative Commons BY 4.0 licence so that Salvadoran institutions, cooperatives, and researchers can reuse the material directly. It is the practitioner-facing companion to the 2024 peer-reviewed paper in Environmental Science & Policy.

A bilingual feature setting out why drought has to be read spatially rather than as a series of index values: when an event actually begins and ends, how far it reaches, and how it moves across a territory. It walks through the spatio-temporal method behind Dialectik's drought analysis and its extension into machine-learning prediction of agricultural impact. Tracks is published by Mexico's Ministry of Foreign Affairs, where Mexican embassies propose articles highlighting the work of the diaspora.

The methodological core of treating drought as a three-dimensional object in space and time. Each event has an onset, an end, a location, a magnitude, and a spatial trajectory — characteristics a conventional index cannot express. The chapter sets out how clusters are filtered by size and how the drought-indicator threshold is optimised, which is what makes event-by-event comparison defensible.

The spatio-temporal clustering approach applied at continental scale: roughly 200 drought events detected across Europe between 1981 and 2020, using three-month Standardized Precipitation Index maps derived from ERA5 reanalysis. The largest European meteorological droughts in that period were 1996, 2003, 2002 and 2018. Evidence that the method holds up against independent expert characterisation of past events.

A concise intelligence brief prepared for the Group of Latin American and Caribbean States (GRULAC). It maps the four regulatory shifts reshaping coffee trade with Europe — EUDR, CBAM, EU–Mercosur, and EU–Central America — against projected climate risk across LAC coffee-producing zones, and argues why social fragility is the variable most often missing from compliance and risk assessments.