Trish Foschi
Professor Emerita of Geography
San Francisco State University


Website and database for project

Studying the Effects of Climate Variability and Change on Wetland Restoration

Efforts to restore the ecosystems of the Bay-Delta estuary and watershed could be complicated by unforeseen consequences of climate variability and change. The freshwater flow patterns upon which these ecosystems depend are generated in the headwaters of the Sacremento and San Joaquin Rivers. At the start of the project, relatively little was known about the role of vegetation in determining the watershed's hydrologic response to climate variability and change. Consequently, a 25-year 4-km Advanced Very High Resolution Radiometer (AVHRR) dataset and corresponding AVHRR-derived Normalized Difference Vegetation Index (NDVI) and Leaf Area Index (LAI) datasets were developed for California. To our knowledge, these datasets were the first long-term high-resolution AVHRR, NDVI, and LAI datasets available for California. The Knowles hydrologic model of the Sacramento-San Joaquin Watershed was altered to integrate these records of vegetation change to study the effects of climate variability and change on wetland restoration effects. The project was funded by the CALFED Ecosystem Restoration Program.


Difference image showing ice PSCs


Toward a Polar Stratospheric Cloud Climatology

Polar stratospheric clouds (PSCs) play a critical role in ozone depletion over both polar regions. When we started this project, PSCs were not well described in models of the ozone depletion process because the climatology of PSCs was largely incomplete and unknown. The most complete PSC records consisted of measurements from limb-viewing satellites that offer limited spatial and temporal coverage. In order to construct a long-term PSC climatology, we developed methods for detecting PSCs in AVHRR satellite imagery. Two approaches were examined: (1) a correlative approach relating image-derived data to verification data and (2) an interpretation of the image-derived data based on a radiative transfer model. The image-derived data—namely, density sliced channel 5 temperature data, color composites, and density sliced channel 4-5 brightness temperature difference images—provided quick views of potential PSC locations. The model determined that Type II or ice PSCs can be detected using the AVHRR thermal infrared channels. The model-based approach can be used to construct a long-term ice PSC climatology from the AVHRR archive.


PowerPoint paper (May 2003)

Using Image Mining to Map a Spectrally Variable Subject

Classifying Egeria densa, Brazilian waterweed, in color-infrared (CIR) imagery by automated methods presents a challenging problem due to a number of unfavorable conditions, including variable imaging conditions (e.g., vignetting), problems associated with water-related subjects (e.g., sun glint), and other environmental changes (e.g., exposure of Egeria at extremely low tide). Spectral response patterns for Egeria do not separate well from those of other land cover classes in CIR imagery. In addition, digital analyses also indicate that subtle changes (e.g., in Egeria canopy density or water turbidity) produce more overlapping spectral response patterns. Clearly, traditional computer-assisted multispectral classification methods are problematic under these conditions, and visual/manual image interpretation and analysis procedures are time-consuming and daunting. To address these challenges, we exploited the latest developments in data mining and investigated novel combinations of effective methods—including rule-based knowledge engineering and machine learning—to create prototype systems that detect Egeria.


Information and updates at DBW website


Monitoring Egeria densa in the Sacramento-San Jaoquin Delta

Egeria densa, an invasive waterweed, is causing navigation and reservoir-pumping problems in the Sacramento-San Joaquin Delta. This exotic submergent weed grew uncontrolled in the Delta for over 45 years. In 1997, the California Department of Boating and Waterways (DBW) started developing a control program to manage Egeria. At that time, researchers at the Romberg Tiburon Center (RTC) were hired to assess the effects of control protocols on fish and other fauna and to estimate the areal extent of Egeria. Greenhouse and in situ experiments to determine the effects of temperature, turbidity, and salinity on Egeria were conducted. The RTC team monitored the areal extent of this weed using scan-digitized color-infrared (CIR) film positives and other CIR imagery and supporting fieldwork. The RTC monitoring ended in December 2003, but the development of an automated system for detecting Egeria in digital imagery continued (see Image Mining Project above).


See Curriculum Vitae for other papers