Vegetation Mapping
During my time working on various projects in the National Forests of Southern California and at the California State Parks Northern Service Center, I gained extensive experience in spatial data management for GIS data libraries, imagery acquisition, and vegetation mapping. I developed, implemented, and maintained spatial data standards to ensure the accuracy and consistency of GIS data libraries. Additionally, I utilized a range of techniques for producing maps, from traditional methods such as hand delineating polygons from aerial photos to advanced remote sensing techniques.
My expertise includes image classification, canopy reflectance modeling, spectral mixture analysis, and the integration of remote sensing with terrain modeling. I have also employed Object-Based Image Analysis (OBIA) using the Image Processing Workbench (IPW) and the object-oriented data model in SPRING GIS to enhance my mapping capabilities.
In addition to data analysis, I have actively collected floristic field data to establish environmental terrain rules, which were then incorporated into GIS models. I have also gathered structural tree data to calibrate a geometric canopy model that estimates canopy crown size class and density accurately.
One of my research interests lies in the integration of remote sensing, GIS, and climate variables with statistical analysis techniques such as CART (Classification And Regression Tree) and Random Forest. For my master's thesis, I utilized CART analysis to predict riparian vegetation in the Descanso District of the Cleveland National Forest. In my dissertation, I expanded on this approach and employed an ensemble of multivariate statistical classifiers, including Random Forest, Generalized Additive Model, and CART, to predict the potential distribution of Great Basin bristlecone pine (Pinus longaeva) across the Great Basin region encompassing California, Nevada, and Utah.
Below, you will find a sample of my vegetation mapping projects, which showcase my proficiency in this field.
Using multivariate methods to predict the distribution of Great Basin bristlecone pine forests
Accurate maps are necessary to make informed decisions on ecology species and habitat. This is especially true with a fragmented species such as Great Basin bristlecone pine (GBBP), which occurs on ‘islands’ of high elevation in the Basin and Range peaks of California, Nevada, and Utah. Due to the inaccessibility of many of the sites that this species occurs, information on their location and abundance is incomplete, and thus is needed. Understanding the distribution of this species is required to evaluate their potential response to disturbances such as fire and climate change. I modeled the distribution of GBBP using widely available topographic and spectral variables calculated from a geographic information system (GIS). I tested several multivariate statistical models to produce a GIS layer (map) that provides a foundation to examine large scale changes to GBBP.
Partial dependence plots for selected predictor variables (6 most important from MDA) for random forest (RF) predictions of the presences of GBBP. Partial dependence is the dependence of the probability of presence on one predictor variable.
Species distribution map of Great Basin bristlecone pine (Pinus longaeva) from Random Forest model (RF) predicted into the Great Basin Floristic Province, part of the Mojave Desert (south) and in the Henry Mountains (east) where a known stand occurs. Purple polygon line weights were increased for visualization, resulting in areas on map appearing greater than on the ground
Big Basin Redwoods, Año Nuevo, and Butano State Parks
Combining image segmentation using the object-oriented data model in the open source SPRING GIS, remote sensing image classification, and intensive field survey mapping, vegetation communities were mapped for all of the four parks in the Big Basin area. (Big Basin Redwoods, Año Nuevo State park, Butano State Park, and Año Nuevo Reserve. Vegetation communities were classified into to the Manual of California Vegetation alliances . GIS layers were used for resource inventory and in the Big Basin State Park General Plan.
Emerald Bay State Park Vegetation Map
Fort Ross State Historic Park
Created from integrating Object-Based Image Analysis (OBIA) and ground based vegetation sampling.
Grover Hot Springs State Park Botanical Survey
Intensive wetland vegetation surveys completed by UC Davis botanists, were digitized and integrated with existing vegetation maps of upland communities to produce a multi-scale vegetation map and inventory.
Wetland Survey prepared by: Ellen Dean
