Predicting the extent of massive wildfires once ignited is essential to reduce the subsequent socioeconomic losses and environmental damage, but challenging because of the complexity of fire behaviour. Existing physics-based models are limited in predicting large or long-duration wildfire events. Here, we develop a deep-learning-based predictive model, Fire-Image-DenseNet (FIDN), that uses spatial features derived from both near real-time and reanalysis data on the environmental and meteorological drivers of wildfire. We trained and tested this model using more than 300 individual wildfires that occurred between 2012 and 2019 in the western US. In contrast to existing models, the performance of FIDN does not degrade with fire size or duration. Furthermore, it predicts final burnt area accurately even in very heterogeneous landscapes in terms of fuel density and flammability. The FIDN model showed higher accuracy, with a mean squared error (MSE) about 82% and 67% lower than those of the predictive models based on cellular automata (CA) and the minimum travel time (MTT) approaches, respectively. Its structural similarity index measure (SSIM) averages 97%, outperforming the CA and FlamMap MTT models by 6% and 2%, respectively. Additionally, FIDN is approximately three orders of magnitude faster than both CA and MTT models. The enhanced computational efficiency and accuracy advancements offer vital insights for strategic planning and resource allocation for firefighting operations.

User-generated content (UGC) is a relatively young field of research; however, it has been proven useful in disciplines such as hospitality and tourism, to elicit public opinions of place usage. In landscape architecture and urban planning, UGC has been used to understand people's emotions and movement in a space, while other areas and additional functions are yet to be discovered. This paper explores the capability of UGC in revealing city-scale park management problems and the applicability of social media as a future tool in bridging visitor feedback to city parks and recreation department staff. This research analyzed the spatial characteristics and patterns of Google Maps review quantity, rating score, and review comments. The results of this pilot study indicate the spatial and structural features of the Chicago parks and demonstrate distribution problems, financial investment priority concerns, park usage characteristics, and user preferences of the park attributes. Findings affirm that user-generated online reviews can be used as an alternative and self-reporting data source to effectively assess the natural performance and users' experience of city parks and can potentially serve as an evaluative tool for public park management.

The large and catastrophic wildfires have been increasing across the globe in the recent decade, highlighting the importance of simulating and forecasting fire dynamics in near real-time. This is extremely challenging due to the complexities of physical models and geographical features. Running physics-based simulations for large wildfire events in near real-time are computationally expensive, if not infeasible. In this work, we develop and test a novel data-model integration scheme for fire progression forecasting, that combines Reduced-order modelling, recurrent neural networks (Long-Short-Term Memory), data assimilation, and error covariance tuning. The Reduced-order modelling and the machine learning surrogate model ensure the efficiency of the proposed approach while the data assimilation enables the system to adjust the simulation with observations. We applied this algorithm to simulate and forecast three recent large wildfire events in California from 2017 to 2020. The deep-learning-based surrogate model runs around 1000 times faster than the Cellular Automata simulation which is used to generate training data-sets. The daily fire perimeters derived from satellite observation are used as observation data in Latent Assimilation to adjust the fire forecasting in near real-time. An error covariance tuning algorithm is also performed in the reduced space to estimate prior simulation and observation errors. The evolution of the averaged relative root mean square error (R-RMSE) shows that data assimilation and covariance tuning reduce the RMSE by about 50% and considerably improves the forecasting accuracy. As a first attempt at a reduced order wildfire spread forecasting, our exploratory work showed the potential of data-driven machine learning models to speed up fire forecasting for various applications.

Human encroachment into wildlands has resulted in a rapid increase in wildland–urban interface (WUI) expansion, exposing more buildings and population to wildfire risks. More frequent mapping of structures and WUIs at a finer spatial resolution is needed for WUI characterization and hazard assessment. However, most approaches rely on high-resolution commercial satellite data with a particular focus on urban areas. We developed a deep learning framework tailored for building footprint detection in the transitional wildland–urban areas. We leveraged meter scale aerial imageries publicly available from the National Agriculture Imagery Program (NAIP) every 2 years. Our approach integrated Mobile-UNet and generative adversarial network. The deep learning models trained over three counties in California performed well in detecting building footprints across diverse landscapes, with an F1 score of 0.62, 0.67, and 0.75 in the interface WUI, intermix WUI, and rural regions, respectively. The bi-annual mapping captured both housing expansion and wildfire-caused building damages. The 30 m WUI maps generated from these finer footprints showed more granularity than the existing census tract-based maps and captured the transition of WUI dynamics well. More frequent updates of building footprint and improved WUI mapping will improve our understanding of WUI dynamics and provide guidance for adaptive strategies on community planning and wildfire hazard reduction.

The severity of wildfire burns in interior lands of western US ecosystems has been increasing. However, less is known about its coastal mountain ecosystems, especially under extreme weather conditions, raising concerns about the vulnerability of these populated areas to catastrophic fires. Here we examine the fine-scale association between burn severity and a suite of environmental drivers including explicit fuel information, weather, climate, and topography, for diverse ecosystems in California's northern coastal mountains. Burn severity was quantified using Relative difference Normalized Burn Ratio from Landsat multispectral imagery during 1984-2017. We found a significant increasing trend in burned areas and severity. During low-precipitation years, areas that burned had much lower fuel moisture and higher climatic water deficit than in wetter years, and the percentage of high-severity areas doubled, especially during the most recent 2012-2016 drought. The random forest (RF) machine learning model achieved overall accuracy of 79% in classifying categories of burn severity. Aspect, slope, fuel type and availability, and temperature were the most important drivers, based on both classification and regression RF models. We further examined the importance of drivers under four climatic conditions: dry vs. wet years, and during two extended drought periods (the 2012-2016 warmer drought vs. the 1987-1992 drought). During warm and dry years, the spatial variability of burn severity was a mixed effect of slope, long-term minimum temperature, fuel amount, and fuel moisture. In contrast, climatic water deficit and short-term weather became dominant factors for fires during wetter years. These results suggest that relative importance of drivers for burn severity in the broader domain of California's northern coastal mountains varied with weather scenarios, especially when exacerbated by warm and extended drought. Our findings highlight the importance of targeting areas with high burn severity risk for fire adaptation and mitigation strategies in a changing climate and intensifying extremes.

Climate change mitigation policies have usually considered forest-based actions as cheap and fast options to reduce CO2 concentration in the atmosphere and slow down global warming. Most economic analyses, however, have ignored the effects of these actions on land surface albedo and the resulting effect on energy balance and temperature. This study estimates the marginal cost of forest mitigation associated with both carbon sequestration and albedo change, by introducing regional and forest-specific albedo information in a global dynamic forestry model. Our analysis indicates that traditional forest sequestration policies have underestimated the costs of climate mitigation, driving forest-based actions in regions where subsequent changes in albedo are significant. To reduce this inefficiency, this paper proposes a novel approach where both carbon sequestration and albedo effect are incorporated into pricing. Our results suggest that, under the same carbon price path, the integrative policy provides greater net global mitigation in absolute terms and per hectare of forest, and thus it is more efficient and less intrusive than the traditional policy.

Building type information is crucial to many urban studies, including fine-resolution population estimation, urban planning, and management. Although scientists have developed many methods to extract buildings via remote sensing data, only a limited number of them focus on further classification of the extracted results. This paper presents a novel building type classification scheme based on the integration of building height information from LiDAR, textural, spectral, and geometric information from high-resolution remote sensing images, and super-object information from the integrated dataset. Building height information is firstly extracted from LiDAR point clouds using a progressive morphological filter and then combined with high-resolution images for object-oriented segmentation. Multi-resolution segmentation of the combined image is performed to collect super-object information, which provides more information for classification in the next step. Finally, the segmentation results, as well as their super-object information, are inputted into the random forest classifier to obtain building type classification results. The classification scheme proposed in this study is tested through applications in two urban village areas, a type of slum-like land use characterized by dense buildings of different types, heights, and sizes, in Guangzhou, China. Segment level classification of the study area and validation area reached accuracies of 80.02% and 76.85%, respectively, while the building-level results reached accuracies of 98.15% and 87.50%, respectively. The results indicate that the proposed building type classification scheme has great potential for application in areas with multiple building types and complex backgrounds. This study also proves that both building height information and super-object information play important roles in building type classification. More accurate results could be obtained by incorporating building height information and super-object information and using the random forest classifier.