Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Preliminary Analysis of Vegetation Distributions within Five Study Areas in the Stillaguamish River Delta, Port Susan, Washington by Sarah L. Thomas, ENVS422, 03/19/2014 ABSTRACT The purpose of this project was to conduct preliminary analysis for a research project to be conducted in the spring of 2014 under the supervision of Roger Fuller, researcher the Nature Conservancy of Washington and the Resilience Network. The primary objective of the project is to analyze the distribution of vegetation in the coastal wetlands of the Stillaguamish River Delta at Port Susan, Washington. Tidal marshlands perform many ecological services in supporting the health of coastal ecosystems, as well as the socioeconomic activities that depend them (Whelchel, 2010). This paper outlines the data collected, processed, and analyzed for preliminary analysis of vegetation distributions in five study sites within Port Susan study area. First, previous research conducted in tidal marshlands will be review. The feasibility of spring project and future analysis will conclude the report on preliminary analysis conclusions. INTRODUCTION The Stillaguamish River Delta is located within Port Susan of the Puget Sound, approximately 20 miles northwest of Everett, Washington. This preliminary analysis focuses on mapping vegetation distributions within the five coastal wetland study areas in the Stillaguamis River Delta at Port Susan. The preliminary findings will be applied to spring quarter research project with be the primary goal of modelling factors influencing wetland resilience to projected sea level rise. The importance of the research project extends beyond the local ecological and socioeconomic impacts of sea level rise, as research findings will be used for future analysis of coastal wetlands throughout the Puget Sound region. Literature covering the topics of coastal wetlands further investigates the importance of modeling the dynamics of vegetation distributions and applies techniques to be used in future analysis. Literature Review Coastal wetlands will serve a key role in ecological and human community resilience to projected sea level rise. Marshland vegetation acts as a natural defense to storm surge, as well as protects erosion by way of wave attenuation, shoreline stabilization, and floodwater attenuation (Shepard et al, 2011). Coastal wetlands also protect near shore water quality by filtering and contaminants and land-derived nutrients (Whelchel, 2010). A major concern in regards to wetland viability is limitations to wetland expansion. In many regions, including Port Susan, long standing development has taken place on near-shore land. Maintenance of the wetland functions may be limited by efforts to fortify upland developed area in order to prevent inundation (Whelchel, 2010). Many coastal wetland studies include modeling morphological features, accretion rates, and fine-scale local land cover effects on wetland viability, as well as results of restoration efforts (Whelchel 2010, Temmerman et al 2012; Kirwan et al 2008, Wamsley et al 2009). Vegetation in Resilience Temmerman et al (2012) used vegetation models to map the general effects of morphological features on vegetation growth within a simulated, idealized coastal wetland. The study mapped vegetation models within three morphological zones: the off-shore slope, the marsh platform, and marsh channels. Off shore slopes consisted of cross-shore bathymetric profiles using bathymetric models at elevations 1 to 1000 meters below sea level. The marsh platform consisted of regions with an elevation approximately 0.5 meters above sea level, which was characterized as a transition zone for mature to non-vegetated intertidal zone (Temmerman et al, 2012). Marsh channels consisted of concave and convex features within the marshland as constructed from aerial pictures from a Dutch salt marsh that was characterized as being “non-degraded”. Temmerman et al (2012) analyzed the effect of the presence of vegetation and stream patterns, and the effect of vegetation patterns on flood propagation and wave attenuation. The extent and topographic position of morphological features within a coastal wetland will affect the rate of flood propagation and wave attenuation by coastal features (Temmerman et al, 2012). In the study by Temmerman et al (2012), stream channeling wave water levels would decrease linearly without vegetation of over the marshland platform at a rate of 0.1m/km landward. In a completely vegetated marsh platform without stream channels, vegetation induced friction resulted in a linear decrease in wave level to a landward movement of 0.26 m/km. Models ran with both vegetation and the presence of stream channels demonstrated the complex, non-linear reduction of peak water levels. Generally, increasing vegetation resulted in reduction of shoreward flood movement, and reducing channel depth decreased the rate of shoreward peak water levels (Temmerman et al, 2012). The effects of vegetation on flood propagation and wave attenuation were also analyzed according to spatial patterns of vegetation. Spatial patterns of vegetation occurring over a same percentage of the wetland area with patterns of break up have different effects on flood wave attenuation (Temmerman et al, 2012). Temmerman et al (2012) noted that non-vegetated patches on marsh channels increase shoreward flood propagation at greater rates than non-vegetated patches in intermediate tidal zones with adjacent vegetation. Also noted, flood attenuation and vegetation break-up have an exponential negative correlation, where random patterns of vegetated and non-vegetated coverage will increase the potential for flood propagation (Temmerman et al, 2012). Temmerman et al (2012) also note that non-vegetated patches are less likely to occur near stream channels than in patches where vegetation is dominant. Also, a lower the marsh platform relative to sea level, as a consequence of reduced sedimentation and even erosion, is shown to have minor additional effect on shoreward flood propagation (Temmerman et al, 2012). Patterns in Disturbance Kirwan et al (2008) explored the feedbacks between vegetation and sediment transport in order to demonstrate the importance of wetland vegetation in processes of shoreline stabilization and formation. Using the assumption that biomass productivity is a good proxy for total biomass, Kirwan et al (2008) modeled gravity-driven transport of soil in respect to the local biomass of vegetation. Their model removed all vegetation from random cells, and mapped the distribution of soil sedimentation. The analysis results reported that the proportions of soil accretion increased in vegetated areas because plant roots act as a catchments for soil within a tidal marshlands (Kirwan et al, 2008). This gravity-driven transport model only applies at the vegetated banks because channel networks in tidal marshlands generally have too low of elevations to permit vegetation (Kirwan et al, 2008). This model can be appropriate when considering vegetation recovery from biotic and physical disturbances, which have been observed in less than a decade (Kirwan et al, 2008). However, the accretion rates at bed surfaces should also consider erosional processes and land use parameters that may impede sedimentary processes. Restoration Efforts A study conducted by Wamsley et al (2009) investigated the potential effects of surge as effected by coastal marsh degradation and restoration efforts. Modifications were made to a bathymetry grid and surface roughness based on future landscape predictions for physical processes, geomorphic features, water quality, and ecological succession. Bottom elevation deposition and land building restored wetlands and increased storm surge protection within the restored area relative to base conditions (Wamsley, 2009). Wamsley et al (2009) noted that projected loss of coastal wetlands may increase flood water level by .2m over 10 km, but noted that the overall potential for wave attenuation is dependent on the coastal wetland landscape and storm characteristics. Factors to consider include elevation, structures within or surrounding the wetland, and specific wetland characteristics as well as storm speed track, and intensity (Wamsley, 2009). These parameters covered by Wamsley (2009) will impact a coastal wetland performances in terms of wave attenuation, shoreline stabilization, and floodwater attenuation. METHODS This analysis utilized methods of processing data for analysis applied in the Western Washington University GIS course series. Such processes include Normalized difference in Vegetation Index (NDVI) classification and raster image differencing using raster calculator. The data used in preliminary analysis originated at two sets of remotely sensed data. The first set, used to create maps of vegetation distributions within each study area, was comprised of six aerial photography tiles in TIFF file format collected on July 25th 2013. Each tile covered a section of Port Susan using the visible and infrared portions of the electromagnetic spectrum at a spatial resolution of 0.25 meters by 0.25 meters. The tiles were mosaicked in order to create a single dataset of values for of the Stillaguamish River Delta at Port Susan. The second data set originated as LIDAR files demonstrating the bare earth and high hit series of returns from laser altimetry data collected on January 11th, 2013 using pulses at 1 meter by 1 meter spatial resolution. These datasets were used in analysis of vegetation heights within the study area of the Stillaguamish River Delta. In order to gain a general understanding of the vegetation distribution within the study areas at Port Susan, a Normalized Difference in Vegetation Index was created for the entire Stillaguamish River Delta area. Prior to conducting a NDVI analysis, the aerial photography TIFF files underwent preprocessing. First the six tiles were mosaicked using the Create Mosaic dataset tool in Raster toolset of ArcGIS software. Statistics for all bands, of each tile were calculated using the compute statistics tool in the Data Management toolset. Next, the data was exported to ENVI software and processed using classification toolset to create a Normalized Difference in Vegetation Index classification using bands 1 and 4 of the imagery to derive a scale of vegetation cover based upon the brightness and greenness values reflected by land cover. The resulting NDVI classified image was symbolized using a gradient linear 2% stretch and exported as an ASCII file for analysis in ArcGIS software and converted to ESRI raster file format using the ASCII to RASTER tool in the Data management toolset of ArcGIS. Next, the ESRI raster file was clipped to the bounds of individual study areas based on the extent of vector data of digitized study areas. Image differencing processes were used to create a map depicting vegetation heights within the study areas in the Stillaguamish River Delta at Port Susan. This process subtracted a bare earth LIDAR data from the high-hit LIDAR data for all returns of laser altimetry measurements. The following equation was entered into raster calculator in ArcGIS: (high hit raster file) – (bare earth raster file). The resulting raster file was from symbolized in a gradient stretch and clipped to each vector file of the study areas. The calculate statistics tool in ArcGIS software was utilized in order to derive the differences in vegetation heights within the study areas of the Stillaguamish River Delta at Port Susan. Figure 1. Map depicting the NDVI and image differncing values indicative of vegetation health and hirght throughout all study areas of the Stillaguamish River Delta at Port Susan. Figure 2. Map depicting the NDVI and image differncing values indicative of vegetation health and hieghts throughout Study Area 1 of the Stillaguamish River Delta at Port Susan. Figure 3. Map depicting the NDVI and image differncing values indicative of vegetation health and hieght throughout Study Area 2 of the Stillaguamish River Delta at Port Susan. Figure 4. Map depicting the NDVI and image differncing values indicative of vegetation health and hieghts throughout Study Area 3 of the Stillaguamish River Delta at Port Susan. Figure 5. Map depicting the NDVI and image differncing values indicative of vegetation health and hieghts throughout Study Area 4 of the Stillaguamish River Delta at Port Susan. Figure 6. Map depicting the NDVI and image differncing values indicative of vegetation health and hieght throughout Study Area 5 of the Stillaguamish River Delta at Port Susan. DISCUSSION Generally, vegetation distribution based on NDVI values within the study areas of the Stillaguamish River Delta at Port Susan followed the expected outcomes of previous studies. Vegetation followed the predictions of Temmerman et al (2012), where vegetation follows morphological features and vegetation densities increase in a shoreward direction. The marsh platform had the highest NDVI values and greatest image differencing values, however, non-vegetation anomalies existed within each study area. Visual examination of Study Area 5 (Figure 5), show high vegetation height values in regions indicated as no vegetation by NDVI values. These anomalies on the shoreline are most likely woody debris carried downstream on the Stillaguamish River or from incoming waves. Study Area 5 also has a distinct area of high NDVI values located within the off shore slopes of the Stillaguamish River Delta (Figure 5). This anomaly may exist due to differences in original aerial photography, where off shore data may have been collected at a different time than the on shore date thus different values based on sun angle. Vegetation distributions also followed areas of relatively recent disturbance, as expected by Kirwan et al (2008). Notably, NDVI and vegetation height values located off shore were greater near stream channels than otherwise. Visual examination of the vegetation height images demonstrates the vegetation productivity in recently disturbed areas with indication of mid height values near low heights of steams. This observation is especially clear in examining the vegetation heights at Study Area 2 (Figure 2). Notably, the vegetation heights correspond with NDVI distributions, where the highest values are adjacent to the lowest values, indicating distinct stream channel locations. Overall, the vegetation heights depicted by the image difference raster file ranged from 0 meters to 15.80 meters, but had similar mean values ranging between 0.05 and 0.07 meters (Table 1). Study Area 3 (Figure 3) showed lower spatial distribution of vegetation heights than all other study areas with heights ranging between 0 and 3.98 meters (Table 1). This statistic demonstrates conclusions proposed by Wamsley et al (2009), in regards to the viability of restoration of previously developed wetland sites. Notably, the inundated regions of the Study Area 3 had the lowest NDVI value range, but values increased in the areas disturbed during dike removal (Figure 3). Interestingly, Study Area 3 also had the largest mean and standard deviation of elevation difference values at nearly two to three times all other values (Table 1). These statistics may represent a higher elevation and proximity to human constructed features. The greatest differences between the bare-earth and high hit LIDAR data occurred at Study Area 2, with an elevation difference of 15.80 meters (Table 1). This height may be an anomaly as vegetation such vegetation heights do not generally occur so far off shore. This anomaly may have resulted from errors in creating the bare earth image, or there may be an object in water of the Stillaguamish River Delta at Port Susan. Future Analysis Future analysis of vegetation distributions within the study areas in Stillaguamish River Delta at Port Susan will incorporate field samples collected by Roger Fuller. Samples collected for vegetation include type and health parameters such as plant height and stem thickness. The preliminary analysis for distributions will be compared to field data for accuracy, and compared to the vegetation type coverage throughout the Stillaguamish River Delta. Coverage files in polygon vector file formats already exist and can be compared to the distribution derived through NDVI index mapping. Other data collected includes soil salinity and soil accretion rates. Comparison will be made between all datasets in order to create a model for the dynamics of vegetation distribution within each study site at Port Susan, and comparisons throughout other study areas in the Puget Sound. ACKNOWLEDGEMENTS I would like to acknowledge Roger Fuller, researcher for the Nature conservancy and the Resilience Network for providing the data used in this analysis, as well as guiding the studies for my Spring quarter, 2014 project. LITERATURE CITED Shepard CC, Crain CM, Beck MW. 2011. The Protective Role of Coastal Marshes: A Systematic Review and Meta-analysis. PLoS ONE 6(11): e27374. doi:10.1371/journal.pone.0027374 Whelchel, A. 2010. Coastal Resilience Long Island: Adapting natural and human Communities to Sea Level Rise and Coastal Hazards, Ecological and Socioeconomic Assessment Methods. Nature Conservancy Connecticut Chapter. February, 2010. Temmerman S, De Vries M, Bouma t. 2012. Coastal Marsh die-oof and Reduced attenuation of coastal Floods: A Model Analysis. Global and Planetary Change 92–93, pg 267–274. June 2012. Wamsley, T.V., et al.. 2009. The potential of wetlands in reducing storm surge. Ocean Engineering 7 (18) 2009. doi:10.1016/j.oceaneng.2009.07.018 Kirwan M, Murray A, Boyd W. 2008. Temporary vegetation disturbance as an explanation for permanent loss of tidal wetlands. Geophysical Research Letters, VOL. 35, L05403, doi:10.1029/2007GL032681, 2008 Ewing, K. 1986. Plant Growth and Productivity Along Complex gradients in Pacific Northwest Brackish Intertidal Marsh. Estuaries 1 (6), pg 46-62. March 1986.