Adalberto Ospino-Castro

Wind resource assessment increasingly depends on integrating heterogeneous reanalysis data, turbine information, geospatial operations, and persistent computational services. This article presents WindAnalysis, a containerized multi-service framework for coastal wind-resource workflows in the Colombian Caribbean. Its contribution lies in the architectural and operational integration of scheduled ERA5 and MERRA-2 ingestion, persistent PostgreSQL 15.17 storage, typed FastAPI 0.141.1 services, an authenticated React 19.1.0 geospatial interface, persistent experiment records, turbine power curves, and inverse-distance-weighted interpolation within a single inspectable workflow. The evaluated database state comprised 32,633,302 ERA5 and 3,941,280 MERRA-2 records. Controlled benchmarks showed a 3962-fold acceleration for planner-based row counts and a 19.4-fold acceleration for 1% block-sampled mean queries, with 0.389% mean absolute relative error. Internal leave-one-grid-point-out validation of the deployed p = 2, 50 km IDW configuration yielded a 10 m wind-speed RMSE of 0.366 m/s. The La Guajira demonstration, based on direct ERA5 wind at 100 m and a Vestas V82 turbine, produced a gross AEP of 8.901 GWh/year and a gross capacity factor of 61.58%; a nonparametric bootstrap of daily mean power yielded a 95% interval of 8.05–9.77 GWh/year. The evaluation also includes 2839 archived automated tests and release-level UI-to-runner parity. WindAnalysis therefore provides a reproducible integration layer for regional wind-resource studies, while distinguishing software verification and internal consistency from external physical validation.

Reconstructing missing values in meteorological time series is essential for reliable renewable-energy assessment, but on-device imputation must respect the power, memory and latency budget of edge hardware. We benchmark 46 PyPOTS imputation models—four naïve baselines, TRMF and 41 neural networks—on two tropical coastal stations (UCC and CUC, Colombian Caribbean) resampled to 15 min and subjected to a PyGrinder missingness scenario, running each model on an NVIDIA Jetson Orin Nano (25 W) and measuring MAE, RMSE, power, memory and time. A hybrid Entropy–CRITIC weighting scheme with TOPSIS and MARCOS ranked the alternatives. The best neural networks cut the mean-imputer error by 40–45% (MAE) and 28–40% (RMSE), yet resource usage spanned two orders of magnitude, uncorrelated with accuracy (|ρ|≤0.15). Because both methods placed 64–76% of the weight on resource criteria, naïve baselines and TRMF topped the station-averaged and UCC rankings, TiDE was the most rank-stable neural network, and SCINet led at CUC under the untuned configuration. TOPSIS and SPOTIS agreed closely (ρ≥0.98); MARCOS agreed moderately with both (ρ≈0.68–0.93). Station-to-station agreement was low for TOPSIS (ρ=0.52)—a confound rather than a station effect: the same masking configuration produced an asymmetric gap structure at the shorter, previously incomplete UCC record and a uniform one at the longer, complete CUC record, so station-level conclusions are not directly comparable. An automated per-model hyperparameter-search protocol, which reorders the CUC top, is provided in an appendix. These results offer a reproducible, hardware-validated framework for edge-constrained imputation model selection.

Wind resource assessment increasingly depends on integrating heterogeneous reanalysis, turbine data, geospatial operations, and persistent computational services. This article presents WindAnalysis, a containerized web framework for coastal wind studies in the Colombian Caribbean. Its novelty is architectural and operational: rather than introducing new wind models, the framework integrates scheduled ERA5 and MERRA-2 ingestion, persistent PostgreSQL storage, typed FastAPI services, an authenticated React geospatial interface, experiment retention, turbine-level power curves, and inverse-distance-weighted (IDW) interpolation within a single reproducible workflow. The evaluated database state comprised 32,633,302 ERA5 and 3,941,280 MERRA-2 records. Controlled benchmarks showed a 3962-fold acceleration for planner-based row counts and a 19.4-fold acceleration for 1% block-sampled mean queries, with 0.39% mean absolute relative error. Internal leave-one-grid-point-out IDW cross-validation yielded a 10 m wind-speed RMSE of 0.366 m/s for the deployed p = 2, 50 km configuration. The documented La Guajira case, using direct ERA5 wind at 100 m, produced a gross AEP of 8.901 GWh/year and a gross capacity factor of 61.58% for a Vestas V82 turbine. The evaluation further includes 2839 archived automated tests and release-level UI-to-runner parity. WindAnalysis therefore provides an inspectable, repeatable integration layer for regional wind-resource assessment while keeping computational reproducibility distinct from external physical validation.

Offshore wind development in the Colombian Caribbean requires characterization of tropical cyclone exposure to support climate-resilient planning, yet objective regime classifications for this basin remain limited. This study develops an unsupervised Self-Organizing Map (SOM) framework using the NOAA HURDAT2 best-track database for 1970–2023, comprising 668 six-hourly observations from 103 storms and nine physically interpretable features describing position, intensity, kinematics, and seasonality. A 4 × 4 SOM combined with hierarchical Ward clustering identified four cyclone exposure regimes: R0, late-season slow north-eastward disturbances near San Andrés (11.5%); R1, fast westward mature tropical storms across the central Caribbean (62.0%); R2, slow north-westward disturbances off Central America (21.9%); and R3, rare major hurricanes crossing the central Caribbean (4.6%, seven storms; historical mean recurrence interval 7.7 years). R3 exhibited mean maximum sustained winds of 67.8 m/s and a minimum central pressure of 938.5 hPa, with peak winds reaching 79.7 m/s. The results provide a complementary historical hazard characterization layer for offshore wind planning, supporting risk-informed spatial and operational assessment. However, the identified regimes do not represent wind-resource conditions or probabilistic design return periods and should therefore be integrated with site-specific wind, wave, and structural assessments.

High-resolution atmospheric modeling over topographically complex coastal zones requires multi-source spatial data fusion to bridge the scale gap between satellite-assimilated global reanalysis and localized surface boundary layer dynamics. This study presents a spatial data fusion and post-processing framework that couples convection-permitting Weather Research and Forecasting (WRF) simulations (3 km grid spacing) with multi-model machine learning calibration and a geographic information system multi-criteria decision analysis (GIS-MCDA) across continuous computational grids (24,871 cells). Focusing on a high-resolution seasonal case study during the peak-intensity regime of the Caribbean Low-Level Jet (CLLJ) and Northeast Trade Winds along the Colombian Caribbean coast, ten candidate predictive algorithms (a baseline linear model, five ensemble/boosting architectures, Support Vector Regression, and three autoregressive benchmarks) were evaluated across multiple validation schemes. Support Vector Regression (SVR) with a radial basis function (RBF) kernel achieved superior residual calibration (R²=0.9523, RMSE=0.4601 m/s, RMSSE=0.2444) under day-grouped cross-validation (GroupKFold), while strict chronological out-of-sample testing revealed the critical sensitivity of kernel models to harmonic temporal encodings. A direction-preserving vector scaling algorithm transferred scalar corrections to horizontal wind vectors, which were subsequently scaled vertically to 80 m hub height using WRF-derived prognostic shear and roughness parameters. The calibrated fields were integrated into a non-compensatory Boolean GIS-MCDA model incorporating thermodynamic wind power density, topographic slope limits, environmental/indigenous reserves, and logistical infrastructure proximity. The spatial fusion framework identified an 8.9% macro-corridor of high feasibility (2,214 cells; ≈19,926km²) concentrated in northern La Guajira, while substantially reducing the overestimation artifacts of uncalibrated mesoscale simulations in the sheltered leeward wake of the Sierra Nevada de Santa Marta. The methodology provides a reproducible data fusion approach for screening-level renewable resource assessment in data-sparse tropical coastal environments.

The development of offshore wind energy in tropical cyclone-prone regions requires analytical frameworks that capture non-stationary climate dynamics. This study presents a multi-scale spectral approach to characterize Atlantic tropical cyclone variability and assess implications for offshore wind resilience in the Caribbean Basin. The methodology integrates Fast Fourier Transform (FFT) and Continuous Wavelet Transform (CWT) to resolve temporal variability in sea surface temperature, cyclone frequency, and intensity, complemented by two-dimensional kernel density estimation (KDE) and non-stationarity analysis. Using NOAA and National Hurricane Center datasets, results identify dominant periodicities at annual and ENSO (2–7 year) scales, a post-1995 spectral energy shift associated with the positive AMO phase, and a thermodynamically consistent energy corridor along 12–16° N. A statistically significant change point in 1987 (Pettitt test, p < 0.05) is detected, although spatial displacement is not significant. An integrated Wind Risk Index highlights the central-western Caribbean as a high-exposure zone overlapping offshore wind development areas. Exceedance analysis shows that 39.8% of observations surpass 25 m/s, 6.0% exceed 50 m/s, and 1.3% approach 70 m/s, indicating relevant design considerations. These findings support the need for non-stationary, multi-scale approaches in offshore wind risk assessment under tropical cyclone influence.

The development of offshore wind energy in tropical cyclone-prone regions requires analytical frameworks that capture non-stationary climate dynamics. This study presents a multi-scale spectral approach to characterize Atlantic tropical cyclone variability and assess implications for offshore wind resilience in the Caribbean Basin. The methodology integrates Fast Fourier Transform (FFT) and Continuous Wavelet Transform (CWT) to resolve temporal variability in sea surface temperature, cyclone frequency, and intensity, complemented by two-dimensional kernel density estimation (KDE) and non-stationarity analysis. Using NOAA and National Hurricane Center datasets, results identify dominant periodicities at annual and ENSO (2–7 year) scales, a post-1995 spectral energy shift associated with the positive AMO phase, and a thermodynamically consistent energy corridor along 12-16°N. A statistically significant change point in 1987 (Pettitt test, p < 0.05) is detected, although spatial displacement is not significant. An integrated Wind Risk Index highlights the central-western Caribbean as a high-exposure zone overlapping offshore wind development areas. Exceedance analysis shows that 39.8% of observations surpass 25 m/s, 6.0% exceed 50 m/s, and 1.3% approach 70 m/s, indicating relevant design considerations. These findings support the need for non-stationary, multi-scale approaches in offshore wind risk assessment under tropical cyclone influence.