Katia Michelle Villarnobo Gonzalez, Hugo Carlos-Martínez, Jorge Omar López Martínez; Artículo académico
1. Introduction
Reliable estimates of above-ground biomass density (AGBD) are essential for forest carbon accounting, REDD+, and conservation policies. Although Mexico's National Forest Inventory (INFyS) provides the most reliable field-based AGBD data, its limited spatial coverage and high acquisition costs constrain large-scale monitoring, particularly in tropical forests. Earth Observation (EO) products, such as ESA CCI Biomass and GEDI-derived canopy height, offer continuous spatial coverage but are affected by non-stationary relationships with AGBD and sensor saturation. Consequently, integrating EO and field data requires models that explicitly account for spatial dependence and uncertainty, making spatially varying coefficient (SVC) geostatistical models a suitable approach.
2. Research Gap
Models fitted over large, heterogeneous domains often borrow strength across broad ecological gradients and tend to shrink local predictions toward the national mean. While statistically efficient at the national scale, this can lead to regional miscalibration, particularly where Earth Observation (EO) covariates lose sensitivity. Both ESA Climate Change Initiative Biomass (CCI Biomass), which becomes weakly informative beyond roughly 150–200 Mg ha$^{-1}$, and GEDI-derived canopy height, which saturates near the upper end of tall tropical canopies, degrade precisely in dense tropical forests. Quintana Roo, with semi-evergreen tropical forest, secondary vegetation, coastal gradients, karstic substrates, and heterogeneous land-use histories, exemplifies this condition. A state-scale application therefore requires an independent modeling framework that jointly addresses response scale, data quality, and local covariate behavior, explicitly identifying where each EO source drives predictions.
3. Methodology
We develop such a reformulation for Quintana Roo within the SVC–SPDE framework. On a transformed response, the model is written in the standard additive form
$$ g{Y(\mathbf{s})}=[\alpha+\tilde{\alpha}(\mathbf{s})]+[\beta+\tilde{\beta}(\mathbf{s})]x_1(\mathbf{s})+[\eta+\tilde{\eta}(\mathbf{s})]x_2(\mathbf{s})+\epsilon(\mathbf{s}), $$
where $\tilde{\alpha},\tilde{\beta},\tilde{\eta}$ are mean-zero Matérn Gaussian fields approximated by the SPDE over a triangular mesh. The database combines INFyS plot AGBD, CCI Biomass, and GLAD-GEDI canopy height, harmonized around 2019, with covariates extracted as footprint-weighted means per plot. Because spatial covariance is inferred from the data, a single high-leverage record can distort local coefficient surfaces, so we applied a quality-control protocol—coordinate and unit verification, removal of duplicates and implausible values, and restriction to mainland forest. The SPDE mesh was confined to Quintana Roo rather than a national grid, improving computational efficiency.
The choice of $g$ matters once predictions return to the Mg ha$^{-1}$ scale, motivating our departure from the square-root response used nationally. Writing $\mu(\mathbf{s})$ for the linear predictor, the square-root link inverts by squaring, whereas the logarithmic link inverts by exponentiation:
$$ g=\sqrt{\cdot}: Y=\mu^2=\dots+2,\beta(\mathbf{s})\eta(\mathbf{s}),x_1x_2+\dots, \qquad g=\log: Y=e^{\alpha(\mathbf{s})},e^{\beta(\mathbf{s})x_1},e^{\eta(\mathbf{s})x_2}. $$
Squaring introduces covariate interaction terms, so the marginal effect of each covariate depends on the other and on the intercept, making EO effects entangled. The logarithm instead factorizes biomass into independent multiplicative terms, matching its power-law structure and preserving per-covariate effects. Zero-valued records were dropped, as $\log 0$ is undefined; since $\exp{E[\log Y]}\neq E[Y]$, predictions were back-transformed from posterior predictive samples. Accuracy was assessed with $R^2$, RMSE, MAE, and bias, calibration with leave-one-out Probability Integral Transform, and prediction structure through covariate-contribution surfaces $C_{\mathrm{CCI}}(\mathbf{s})=\beta(\mathbf{s})x_1(\mathbf{s})$ and $C_{\mathrm{GEDI}}(\mathbf{s})=\eta(\mathbf{s})x_2(\mathbf{s})$, with relative dominance $D_{\mathrm{CCI}}=|C_{\mathrm{CCI}}|/(|C_{\mathrm{CCI}}|+|C_{\mathrm{GEDI}}|)$.
4. Findings
The logarithmic formulation substantially outperformed the square-root one, returning an $R^2$ of 0.68 against 0.38, RMSE of 26.7 against 46.9 Mg ha$^{-1}$, MAE of 21.9 against 33.2 Mg ha$^{-1}$, and bias of 4.2 against 16.8 Mg ha$^{-1}$—a 78% gain in explained variance, a 43% reduction in RMSE, and a near four-fold reduction in relative bias, from 27.6% to 7.0%. These gains reflect the positive, right-skewed, heteroscedastic nature of biomass: aligning the linear predictor with multiplicative mechanisms prevents spatially varying coefficients from absorbing distortions arising from an unsuitable response scale. Applied locally, the PIT diagnostic confirms better-calibrated behaviour once variance is stabilised and anomalous records are removed.
Because the log scale keeps covariate effects separable, the contribution surfaces admit a clean reading that the squared scale does not. A coefficient map alone is misleading—a steep local slope contributes little where its covariate is near zero—so weighting each slope by the value its covariate takes reveals which EO source governs prediction at a given point. Mapping dominance over Quintana Roo exposes a spatial division of labor: zones where CCI Biomass carries the signal, where canopy height does, and where neither does and the latent field absorbs the residual baseline. This matters because similar canopy heights can correspond to different biomass densities depending on composition and degradation. Hunka et al. (2024) map spatially varying coefficients but stop there; carrying the analysis through to contribution and dominance turns a surface of parameters into an operational map of per-sensor reliability—and is coherent only under a response scale that does not entangle the covariates
5. Contributions
This study proposes an independent log-scale Spatially Varying Coefficient–Stochastic Partial Differential Equation (SVC–SPDE) formulation for tropical biomass that integrates quality control, a state-restricted mesh, and posterior-sample back-transformation into a unified workflow, improving performance in high-biomass regimes where Earth Observation (EO) covariates saturate. Its main contribution is the spatial decomposition of covariate effects using a logarithmic response that preserves separability on the natural scale, enabling identification of where each EO source governs prediction. The framework also supports posterior predictive aggregation over administrative and conservation units with explicit uncertainty, providing a transferable approach for biomass estimation.
Fabricio Cruz Lopez, Sergio Ivvan Valdez Peña; Artículo académico
Introduction. Traditional geospatial analysis of crime incidence is usually based on kernel density models and physical contiguity analysis to explain the displacement of violence, under the assumption of the so-called "neighborhood effect". However, this approach is insufficient to capture the nature of complex systemic phenomena where violence does not necessarily spread through proximity, but rather through nodes of functional connectivity and flows of human mobility. Recent literature on the geography of crime has begun to suggest that spatial models based solely on proximity omit the underlying dynamics that interconnect non-contiguous territories. This study proposes a comprehensive methodology based on network analysis to evaluate the geospatial dynamics of femicide in the 46 municipalities of the state of Guanajuato, Mexico. This approach allows us to transcend the static view of the territory and visualize the state as an interconnected system of dynamic risks, offering a perspective that integrates temporal synchronization variables on the spatial structure.
Methodology. The research starts from the construction of a topological network where the nodes represent the municipalities and the edges symbolize the intensity of the Hurdle temporal correlation (r) between their respective historical series of femicides. To filter out statistical noise and focus on strong relationships, a dynamic threshold of statistical synchrony (|r| > 0.4) was applied to municipalities with more than 3 feminicides. Subsequently, the Louvain community detection algorithm was implemented, a heuristic method designed to optimize network modularity and segment the territory into clusters based on synchronous criminal behaviors, regardless of their geographical location. To facilitate the exploration of this data, a dynamic geovisualization tool was developed (implemented using advanced web visualization libraries). This platform allows for real-time interaction: users can adjust correlation thresholds, visualize the weight of nodes (based on the cumulative total of cases), and examine the network topology through an intuitive interface. This functionality allows a transition from traditional static analysis to high-level visual analytics, facilitating the identification of patterns hidden from the naked eye. Results. An interesting disconnect is revealed between territorial proximity and criminal dynamics, questioning the hegemony of the spatial contagion model. Three distinct criminal ecologies were identified through network analysis: 1) The Isolation of Industrial Epicenters: This geospatial anomaly is maintained in the absolute epicenters of violence: León, Celaya and Irapuato. These municipalities, which concentrate a high volume of accumulated cases, operate under an apparent endemic temporary isolation. Topological analysis (output-input degree of the nodes) shows that these centers, despite being focal points of greater nominal incidence, maintain weak or negative correlations with their immediate periphery. This suggests that femicidal violence in these centers responds to structural, demographic and internal security factors specific to the urban core, characterized by a constant saturation that does not always fluctuate in sync with the rest of the state. 2) The "Strategic Periphery" and the Territorial Conflict: A critical finding is the identification of an extensive cluster that connects municipalities that, historically, are not part of the conventional industrial corridor, but are located on the borders of the state. This group includes Comonfort, Apaseo el Grande, Cortazar, Guanajuato and San Miguel de Allende, exhibiting robust internal connectivity. This network extends towards Tarimoro, San José Iturbide and Salvatierra. The configuration of this cluster is revealing: being located on the state peripheries, these municipalities have been, for years, the scene of an intense territorial dispute between criminal organizations (cartel war). The synchronicity in criminal behavior in this bloc is not accidental, but appears to be a byproduct of territorial control dynamics, where instability derived from high-impact conflict escalates to levels of gender violence, creating an ecosystem of shared insecurity in these border corridors. 3) Functional and Proximity Correlations: Finally, the model captures correlations that, while following contiguity logics, reinforce the validity of the network model by reflecting transport dynamics and population flow. The Salamanca-Yuriria and Acámbaro-Pénjamo axes stand out as "sensitive" or intuitive correlations. These connections demonstrate how administrative and regional mobility dynamics influence the spread of violence, allowing us to observe that the risk of femicide is not always random, but is channeled through established communication axes. Taken together, these findings demonstrate that femicide risk operates under a hybrid network logic: influenced by contiguity in functional communication axes, but dominated by a geopolitics of violence in border areas where conflict between criminal actors alters the municipal social fabric. Conclusions. The structuring of public prevention policies should not be limited to containment in geographically adjacent areas nor should it focus exclusively on the municipalities with the highest absolute volume of cases. Geospatial analysis, enhanced by analysis through graph theory algorithms, as well as interactive visualization, allows the discovery of non-contiguous corridors of violence (temporal clusters with r > 0.4). This evidence compels us to redefine intervention strategies towards an asymmetric and anticipatory model, where prevention is coordinated simultaneously in municipalities that, although distant on the map, share an identical criminal rate. This approach allows for the optimization of resources and action on the nodes of the network that function as precursors to violence in the state territory.
Cristina Eduardo; Artículo académico
This study focuses on the Ecological Restoration Zone (ERZ) within the area of influence of the Endhó Dam, encompassing the municipalities of Atitalaquia, Atotonilco de Tula, Tepeji del Río de Ocampo, Tepetitlán, Tezontepec de Aldama, Tlahuelilpan, Tlaxcoapan, and Tula de Allende in the state of Hidalgo, Mexico. This region is of particular concern due to the impacts of climate change and environmental pollution on its vulnerable ecosystems. The Mezquital Valley, where the ERZ is located, faces significant environmental challenges that have been intensified by the historical use of wastewater from the Mexico City Metropolitan Area for agricultural irrigation. This practice has contributed to high levels of water contamination and the presence of a wide range of environmental pollutants.
The primary objective of this research is to analyze climate trends by combining historical and future interannual climate information for the ERZ. Specifically, the study examines trends in mean temperature and precipitation for the period 1958–2024 and their projections for 2025–2100 under different climate change scenarios using two Global Climate Models (GCMs), GFDL-ESM4 and MIROC6. The goal is to generate scientific information that supports environmental management and land-use planning.
During the first phase, the TerraClimate dataset was used to obtain spatiotemporal precipitation and temperature data for the 1958–2024 period. This dataset, generated through the interpolation of observations and reanalysis data, provides a spatial resolution of approximately 4 km × 4 km, enabling a detailed regional assessment of climate trends.
Future climate projections were obtained from two General Circulation Models (GCMs) included in Phase 6 of the Coupled Model Intercomparison Project (CMIP6): GFDL-ESM4 and MIROC6. These models were selected because they provide data for the four Shared Socioeconomic Pathway (SSP) scenarios evaluated in this study: SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5, which represent different radiative forcing trajectories and socioeconomic challenges related to climate change mitigation and adaptation.
Climate trends were evaluated using the nonparametric Mann–Kendall test to identify trend direction and the Theil–Sen slope estimator to quantify the magnitude of change. Statistical significance was assessed using p-values at a 95% confidence level (α = 0.05).
Historical TerraClimate data and future GCM projections were integrated through a downscaling process. Because the original interannual GCM outputs had coarse spatial resolutions (100 km × 100 km for GFDL-ESM4 and approximately 150 km × 150 km for MIROC6), downscaling enabled the construction of a continuous annual time series with a 4 km × 4 km spatial resolution.
Results for the historical period (1958–2024) indicate an increase in annual mean temperature of 1.57°C, rising from 15.13°C in 1958 to 16.70°C in 2024. This warming trend is statistically significant, with an estimated annual increase of 0.022°C according to the Theil–Sen slope estimator. Although precipitation did not exhibit a statistically significant trend (p-value > 0.05), total annual precipitation decreased by 12.65 mm over the study period, with several recent consecutive years experiencing precipitation deficits relative to the historical average.
Spatial analysis reveals substantial heterogeneity in climate conditions across the ERZ. Historical mean annual temperatures range from 12.4–14.0°C in the southwestern portion of the region to 16.5–17.8°C in the northeast. Annual precipitation varies from 329–426 mm in the northern municipalities of Tezontepec de Aldama and Tlahuelilpan to 640–818 mm in the southern municipality of Tepeji del Río. This spatial variability reflects the influence of topography on regional climate patterns.
Future climate projections indicate that both models project declining precipitation through 2100 under all four SSP scenarios. The GFDL-ESM4 model projects the most severe reductions, reaching as much as 20.43% under the SSP5-8.5 scenario, whereas MIROC6 projects more moderate decreases of approximately 13%. For mean temperature, MIROC6 projects greater warming by 2100, ranging from 2.20°C under SSP1-2.6 to 3.95°C under SSP5-8.5. In contrast, GFDL-ESM4 projects more gradual increases, ranging from 1.95°C to 2.86°C across the same scenarios.
These findings have important implications for environmental management. The projected increase in temperature and decrease in precipitation, particularly under the most extreme climate scenarios, could intensify water stress, reduce water availability for agriculture, and alter regional ecosystems.
The integration of historical observations with future climate projections provides a robust basis for understanding the region’s climate trajectory and its potential impacts on ecological restoration and natural resource management. The study highlights the need to implement adaptation and mitigation strategies that account for the vulnerability of both ecosystems and local communities to projected climate change.
Overall, this research contributes to the scientific understanding of climate change in Mexico’s semi-arid and temperate regions by providing valuable information to support environmental policy and sustainable development decision-making in the Mezquital Valley, particularly within the Ecological Restoration Zone.