Wildfires are a common occurrence in chaparral shrublands, and post-fire patterns of biomass accumulation are important for understanding ecosystem productivity and fuel available for future fires. In this research, I examine patterns of biomass accumulation in southern California chaparral shrublands at early and late stages of post-fire recovery using a combination of detailed field work and remote sensing. Using field measurements of a site with adjacent stands of varying ages and high spatial resolution imagery, I examine patterns of species composition and associated levels of biomass to characterize long-term patterns in biomass accumulation. I also evaluate the potential for utilizing shrub growth ring widths to track annual biomass accumulation in the first decade of post-fire recovery, and test for the relationship between biomass and spatial variation in factors related to the energy and water balance. In addition, I examine the potential for extending the use of shrub growth rings to track biomass across larger spatial extents using satellite-based growth metrics. The study of stands of varying ages reveals that biomass shows substantial variation even within stands of the same age, and that species composition is different in younger stands of chaparral compared to the more mature stands. In the study of growth rings, I find that while measuring growth rings widths is a valuable method for tracking biomass accumulation in the first decade following a fire, there is no apparent relationship between biomass and factors related to the energy and water balance. Annual biomass growth, as estimated from shrub growth ring widths, shows a promising relationship with satellite-based metrics of annual growth, indicating the potential for further study of the relationship over larger spatial extents.

Aerial imagery, as a useful tool for emergency management and response, still poses challenges to its effective (Joyce, Wright, et al. 2009). The challenges include information quality, accuracy, and the timeliness of information delivery (Joyce, Wright, et al. 2009). This research utilizes and updates the Remote Sensing Communication Model (RSCM) to configure time-sensitive remote-sensing-systems based on unpiloted aerial systems (UAS) and machine-learning-based damage detection. The UAS (DJI Mavic 1 and DJI Matrice 300) configurations’ navigation accuracies are tested with repeat station imaging (RSI) and traditional imaging. Utilizing real-time-kinematic corrections, the DJI Matrice 300 flown with the RSI method at six sites representing critical infrastructure at San Diego State University is shown to have more accurate repeated navigation to camera stations (0.16m vs. 0.21m for traditional imaging) and multi-date image pairs with the RSI method are shown to have better image co-registration mean absolute error (MAE) accuracy (2.2 pixels MAE vs. 4.4 pixels MAE). The Mask R-CNN machine-learning model (He et al. 2017) evaluated on bitemporal layer-stacked images detected damage (cracks) with a better mean intersection over union (mIoU) when used with the RSI images (83.7% mIoU vs. 72.5% mIoU). A customized convolutional neural network (CNN) and recurrent neural network (RNN) were evaluated using the RSI images, with the RNN having the higher accuracy (98.4% overall accuracy vs. 96.9% overall accuracy). Using the acquisition and newly demonstrated method of analyst capacity within the RSCM, the traditional imaging method and DJI Matrice 300 configuration had the highest timeliness capacity (1.02 x 107 bits m-1 s-1 vs. 7.40 x 106 bits m-1 s-1) of acquisition, and image co-registration based on the traditional imaging method had the highest co-registration analyst capacity (1.30 x 107 bits m-1 s-1 vs. 1.16 x 107 bits m-1 s-1). Of the three machine-learning models, the CNN had the highest analyst capacity and the RNN had the next highest (9.00 x 107 bits s-1 vs. 6.68 x 107 bits s-1).

Urban land expansion can be driven by and drive population growth, but determining cause-effect relationships of urban land expansion and population growth is challenging due to the temporal resolution of decadal censuses for most countries. The relative timing of urban land expansion and population change was explored based on a case study for north Taiwan from 1990 to 2015. Data on urban land expansion were derived from a dense time series of Landsat satellite imagery, and population change from annual population registers at the district level.

A novel algorithm for identifying the starting time of urban expansion was developed and tested based on dense time series of Vegetation-Impervious-Soil (V-I-S) proportion maps derived from Landsat surface reflectance imagery. The time of urbanization estimated by logistic regression was determined from the Impervious cover time series. The identified location and estimated time for newly urbanized lands were generally accurate, with 80% of urban expansion estimated within ± 2.4 years.

The random forest (RF) classification routine was applied to Landsat image inputs for detailed urban land-use classification. Input features were extracted from a 2015 Landsat image dataset, and the top 10 most contributing features for RF classifiers were the six spectral wavebands, gray level co-occurrence matrix (GLCM) homogeneity of V-I-S, and Vegetation temporal variation. The Landsat image was tested with different input features, and the land-use maps derived from the top 10 features yielded the most accurate map. Areal extents of annual urban land uses (Residential, Employment, and Transportation corridor categories) were derived by applying a conditional statement based on the urbanization date of the urban expansion map and overlaying on the 2015 land-use map.

The relative timing and general relationship between urban land-use and population dynamics was explored at the district level. Linear regression was run to understand the general relationship between population and land use. The relative timing between population growth and land-use change was estimated with lagged correlation. Residential land use and change was most related to population and change. Population growth occurred 2.5 years later than Residential land expansion based on the median time lag.

Despite their protected status, forest reserves can be influenced by anthropogenic activities within and adjacent to reserve boundaries, resulting in environmental degradation and changes in forest cover. Long-term monitoring of environmental change within protected areas in a reliable and extensive manner is important given widespread, human-induced land-cover and land-use change. This study demonstrates the utility of optical satellite remotely sensed imagery and multi-temporal image analysis procedures for mapping and monitoring land cover and land use within cloud-prone and mountainous forest reserves and their environs in China and Ghana for the period of mid-1980s to 2018. The novel mapping and monitoring procedures yield extensive land-use dynamic information in a reliable manner by minimizing terrain-related illumination and cloud cover effects. Forest types and land-use are mapped in selected cloud-prone and mountainous forest reserves in China and Ghana to test the reliability of the optimized methods. By applying logical land-use transition rules and interpreting high spatial resolution satellite imagery, land-use changes and the anthropogenic activities associated with them are identified.

Vegetation and land-use types are mapped with moderate to high classification accuracies (64 to 94%) for study areas in China and Ghana. For Fanjingshan National Nature Reserve in China, 12 km^2 in land area is mapped as afforested bamboo and conifer lands associated with payment for ecosystem services programs, and over 25 km^2 is mapped as new development during 1995-2016. Forest area decreased by 9% for the 76 study reserves and environs in southern Ghana between 2000 and 2018. Substantial land changes associated with built development and agricultural expansion are observed in reserve environs within both study areas. Other anthropogenic activities including mining and plantation activities are identified in southern Ghana reserves, while afforestation activities associated with payment for ecosystem programs were predominant adjacent to Fanjingshan in China.

This study contributes to the land-cover and land-use mapping literature by developing and optimizing methods for extremely cloud prevalent and mountainous regions. A semi-automated mapping approach implemented on an open-access, user-friendly platform, similar to the workflow demonstrated in this study, increases the usability and transferability of such mapping techniques. Improved monitoring of other forested, mountainous, and cloud prevalent regions can benefit and inform protected area management and policy, long-term environmental change monitoring, and conservation effort assessment.

Expanding invasive herbaceous vegetation (non-native grasses and forbs or herbs) is replacing portions of native shrublands in San Diego County, California, USA through a grass-fire cycle, which contributes to an increased risk of wildfire ignition and spread as well as a changing fire regime. Despite the association between herb abundance and wildfire risk, remote sensing and image processing approaches for quantification of fractional herb cover in shrublands are not well established, nor is the association between herb fraction and proportion of ignitions. In this research, I comparatively assess the accuracy of herb cover estimation and mapping based on three spectral unmixing models applied to Landsat-derived spectral reflectance and spectral vegetation index data from multiple 2020 dates. Based on the model and methods that most accurately and reliably represent herb cover, I then reconstruct the spatial-temporal distribution of herb growth using Landsat images from 1988, 1997, and 2011 and assess the extent to which herb cover has expanded and replaced woody vegetation cover (1988 to 2020). Finally, I combine the herb cover maps with historical ignition points to evaluate the spatial association between herb fractions and locations where fires initially ignited and spread (1992 to 2020). When compared to generated reference data, results demonstrate the parsimonious spectral unmixing approach applied to a fall date estimates herb cover at the 10% accuracy level and more accurately than the more sophisticated unmixing models. Absolute change estimates derived from the earliest and most recent herb cover maps show approximately 25% of the study area exhibited an increase in herb cover > 20%, and roughly 5% experienced a decrease in herb < -20%, with the greatest concentration of change occurring in wildland-urban interface areas. Factors most strongly associated with substantial increase in herb cover include fire return interval, drought, proximity to development, and elevation. The results of evaluating historical ignitions in herbs show the largest proportion of ignitions occurred in areas with > 20% herbaceous fractional cover. Results from this study will enable improved detection of sensitive habitats by satellite for wildfire-prone communities and help identify target areas for mitigating and combating the grass-fire cycle.