Developing an integrated spatially explicit scale dependent modeling framework for wind farm suitability assessment in Iowa

Wind energy development is occurring rapidly in the United States due to the drive for energy independence and to mitigate environmental concerns. Wind is a clean, abundant, and entirely renewable source of energy and the most promising source of alternative energy. Among the top wind energy producers in the nation, the state of Iowa is experiencing tremendous growth and it’s projected to grow. However, despite the vast development, the contributing factors and spatial decision principles for optimal wind farms placement are not yet well understood. This research advanced an empirical methodology for building site suitability assessment framework for the state of Iowa. Employing the information on existing turbines locations, along with environmental factors (slope, wind power class, elevation, land cover, proximity to neighboring turbine, population density, and distance to transmission line, city, highway, railroad, airport, and river), the study analyzed the contributing factors, their relative importance and regional manifestations. This research developed a spatially explicit scale dependent modeling framework for wind farm suitability assessment based on Iowa context. The framework is based on multiscale empirical module derived from spatial lag regression and machine-learning algorithm coupled with normative component (regulations and policies). The empirical model derived from the spatial lag logistic regression and machine-learning algorithm (Maxent) identified statistically significant factors at different scales. The multiscale spatial lag logistic regression significantly improved modeling compared to standard logistic regression because it accounted for spatial autocorrelation due to the spatial clustering of turbines. Scale’s impact on factors importance were examined. At the Macroscale (statewide) model indicated a good fit to the model with Nagelkerke R square of 0.861. Slope, wind power class, elevation, and distance to transmission line, city, airport, and highway as significant factors that contribute at the Macroscale level. Mesoscale 1 model (regional level) also indicated a good fit with Nagelkerke R squared of 0.801 which identified wind power class, elevation, and distance to transmission line, city, airport, highway and population density as significant factors that impact site suitability at this scale. Mesoscale 2 model (micro-regional) with Nagelkerke R square of 0.794 identified wind power class, elevation, distance to city, river, and transmission line as predictors for site suitability. Microscale model with Nagelkerke R square of 0.784 identified elevation, distance to river and city as significant for predicting suitable site at the scale. As results illustrated, difference in scale of wind development does impact factors importance and changes their significance as well. Overall, elevation, proximity to neighboring turbine, and distance to city are the most important factors that were not impact by scale while the remaining factors displayed scale dependence. Empirical model was coupled with normative factors at a regional scale and the model accuracy of 0.88 indicates a good fit. The framework accounted for the complex technical, environmental, and social constraints to identify suitable sites in Iowa with high accuracy. Ultimately, the framework allows for improved resource characterization to maximize resource utilization. Even though the framework developed is in the context of Iowa, it can be modified for other geographic locations