Ten years of land cover change on the California coastdetected using landsat satellite image analysis: part 1—Marinand San Francisco counties

Christopher Potter

Received: 7 January 2013 /Revised: 22 April 2013 /Accepted: 24 April 2013# Springer Science+Business Media Dordrecht (outside the USA) 2013

Abstract Landsat satellite imagery was analyzed to generatea detailed record of 10 years of vegetation disturbance andregrowth for Pacific coastal areas of Marin and San FranciscoCounties. The Landsat Ecosystem Disturbance AdaptiveProcessing System (LEDAPS) methodology, a transformationof Tasseled-Cap data space, was applied to detected changesin perennial coastal shrubland, woodland, and forest coverfrom 1999 to 2009. Results showed several principal pointsof interest, within which extensive contiguous areas of similarLEDAPS vegetation change (either disturbed or restored)were detected. Regrowth areas were delineated as burnedforest areas in the Point Reyes National Seashore (PRNS)from the 1995 Vision Fire. LEDAPS-detected disturbancepatterns on Inverness Ridge, PRNS in areas observed withdieback of tanoak and bay laurel trees was consistent withdefoliation by sudden oak death (Phytophthora ramorum).LEDAPS regrowth pixels were detected over much of thepredominantly grassland/herbaceous cover of the OlemaValley ranchland near PRNS. Extensive restoration ofperennial vegetation cover on Crissy Field, Baker Beachand Lobos Creek dunes in San Francisco was identified.Based on these examples, the LEDAPS methodology willbe capable of fulfilling much of the need for continual,low-cost monitoring of emerging changes to coastalecosystems.

Keywords Landsat . Coastal vegetation . Disturbance .

Regrowth . Restoration

Introduction

Northern California’s coastal ecosystems include sandybeaches, steep bluffs, rocky headlands, evergreen forests,coastal scrublands, grazed rangelands, native grasslands,intertidal zones, wetlands, lagoons, and other diverse shore-line types. Coastal development has been accompanied bymajor investments in public infrastructure, including roads,airports, and harbors. A network of state and federal parksrun along the shoreline to support a popular recreationalsector (Kildow and Colgan 2005).

Degradation due to coastal erosion can have many neg-ative effects on coastal developments and wildlife habitats(CBEAS 2010). Dams and debris basins, channelized riversand streams, hardened shorelines, and vegetated areasconverted to impervious surfaces may further alter sedimentprovided to the shoreline. Improved inventories of coastalecosystem change are needed provide a sound geographicbasis for decision making on habitat protection in the face offuture climate and projected sea level changes of nearly 1 ftrise by mid-century, and 3 or more feet (1-m) by the end ofthe century (Rahmstorf 2007; Hanak and Moreno 2012).Frequent and more powerful waves during winter stormsmay accelerate coastal erosion and shoreline retreat.

In this study, Landsat satellite imagery was analyzedfrom the years 1999 and 2009 to provide a detailed recordof 10 years of vegetation disturbance and recovery forPacific shoreline areas of Marin and San Francisco counties.The analysis of Landsat imagery over time to map allpotential changes in land cover along the coastal zone at aregional scale is a unique approach to intended detect andcategorize multiple agents of change simultaneously, includ-ing coastal erosion, wildfire, invasive species, and human-induced restoration or development.

C. Potter (*)NASA Ames Research Center, Mail Stop 232-21,Moffett Field, CA, USAe-mail: chris.potter@nasa.gov

J Coast ConservDOI 10.1007/s11852-013-0255-2

Regional geography and study area

The northern coast of Marin County includes the PointReyes Peninsula and National Seashore (PRNS) andTomales Bay, located about 35 miles north of the GoldenGate Bridge and the city of San Francisco. The southerncoast of Marin County includes Muir Beach, Stinson Beach,and the 75,000-acre (30,000 ha) Golden Gate NationalRecreation Area (GGNRA). The San Francisco Countyshoreline includes a portion of the GGNRA, the Presidioof San Francisco, Ocean Beach, and Fort Funston.

Methods

The LEDAPS Disturbance Index (DI) was described byMasek et al. (2008) as a transformation of the LandsatTasseled-Cap data space (Crist and Cicone 1984; Huang etal. 2002; Kauth and Thomas 1976), specifically designed forsensitivity to perennial shrubland, woodland, and forestcover change. The Tasseled-Cap brightness, greenness, andwetness indices are standard transformations of the originalLandsat spectral bands, effectively capturing the threemajor axes of spectral variation across the solar reflectivespectrum. As demonstrated by Healey et al. (2005), the DIis a simple and effective means of tracking vegetationdisturbance across ecosystems dominated by perennialcover types. Unlike simple visible/near-infrared indices(e.g. the Normalized Difference Vegetation Index—NDVI),the LEDAPS DI incorporates Tasseled-Cap wetness, therebyincluding information from the shortwave infrared wave-length, which was shown to be valuable for assessing changesin perennial vegetation structure (Cohen and Goward 2004).

The LEDAPS DI quantifies the normalized spectraldistance of any given pixel from a nominal “densewoody vegetation” class to a “bare soil” class. The DIis calculated using the Tasseled-Cap (brightness–greenness–wetness) indices for Landsat images (Kauth and Thomas1976):

DI ¼ B0− G0 þW 0ð Þ ð1Þwhere B′, G′, and W′ represent the Tasseled-Cap brightness,greenness, and wetness indices normalized by a dense woodyvegetation index result identified for each Landsat scene. Forexample:

B0 ¼ B−μBð Þ.σB ð2Þ

where μB is the mean Tasseled-Cap brightness index ofthe dense woody vegetation class for a particular Landsatscene, and σB is the standard deviation of brightnesswithin the dense woody vegetation class for that samescene. In effect, the DI measures the spectral distance of

a given pixel from the dense woody vegetation “centroid”for that scene, in units of within-class standard deviation.Since LEDAPS DI values are based on the statistics ofwoody vegetation cover reflectance from individual scenes,the DI is relatively insensitive to variability in solar geometrybetween scenes, and lessens the effect of seasonal vegetationvariability among Landsat image dates.

Given the population of mature perennial vegetationpixels identified from recent land cover map products, themean and standard deviation of each Tasseled-Cap compo-nent for the class were calculated. Each Tasseled-Cap imageplane was then normalized as in Eqs. (1) and (2). Thedifference in DI (ΔDI) was next calculated between 1999and 2009. Large positive values ofΔDI have been shown beassociated with a major disturbance event such as wildfire,whereas large negativeΔDI values frequently correspond toregrowth of perennial woody vegetation following distur-bance (Masek et al. 2008).

Thresholds were next applied to the ΔDI values to iden-tify the highest probable areas of ecosystem disturbance orvegetation regrowth/restoration. TheseΔDI thresholds weredetermined using wildfire perimeter boundaries mapped onan annual basis by the California Department of Forestry,Fire and Resource Assessment Program (FRAP), with con-tributions from the USDA Forest Service, the Bureau ofLand Management, and the National Park Service (dataavailable at http://frap.cdf.ca.gov/). Manual adjustments ofthese thresholds were carried out across Central Californiato optimize the spatial correspondence between ΔDI forareas burned by wildfire before 1999 (regrowth pixels) andthose burned between 1999 and 2009 (disturbed pixels).

Since short-term land use transformations may be inadver-tently identified by ΔDI, particularly agricultural rotationpatterns and fallowing of cropland, we routinely excludedthese annual transitions for further evaluations by screeningLEDAPS results with a cropland/non-cropland mask based onUSDA National Agricultural Statistics Service (NASS),California Cropland Data Layer (CDL) from 2010 (availableat http://nassgeodata.gmu.edu/CropScape). The CDL is araster, geo-referenced, crop-specific land cover data layerwith a ground resolution of 30 m. The CDL is producedusing satellite imagery from the Indian Remote SensingRESOURCESAT-1 (IRS-P6) Advanced Wide Field Sensor(AWiFS) collected during the current growing season.Additional land cover maps were used as zonal layersfor classifying LEDAPS results, including: the UnitedStates Geological Survey (USGS) National ElevationDataset (NED), the USGS National Land Cover Database2001 (NLCD 2001), and the National Aeronautics andSpace Administration (NASA) Moderate ResolutionImaging Spectroradiometer (MODIS) 250-m resolution16-day Normalized Difference Vegetation Index (NDVI)composites.

C. Potter

Results

Changes in shrubland, woodland, and forest cover changebetween 1999 and 2009 from LEDAPS analysis inMarin and San Francisco counties were mapped as redshaded pixels for disturbed perennial vegetation cover,

and as blue shaded pixels for regrowing or restoredperennial vegetation cover (Fig. 1). Four principalpoints of interest, each with extensive contiguous areasof similar LEDAPS vegetation change (either disturbedor restored), were identified for further analysis in thesections that follow.

Fig. 1 Results from LEDAPSanalysis of Landsat imagery(30-m pixel resolution) in 1999and 2009 in Marin and SanFrancisco counties. Red shadedpixels were detected asdisturbed vegetation, whereasblue shaded pixels weredetected as regrowing orrestored vegetation. Romannumerals identify selectedpoints of interest examined insubsequent figures

Ten years of land cover change on the California coast

Inverness Ridge, PRNS

Evergreen forests and shrublands dominate the elevatedtopographic area of the eastern PRNS known as InvernessRidge (Evens 2008). Bishop pines (Pinus muricata) pre-dominate the forest cover along the northern portion ofInverness Ridge, with occasional small islands of treesscattered along the ridge southwest toward Drakes Bay.Trees growing in association with Bishop pine or Douglasfir (Pseudotsuga menziesii) on southern portion of InvernessRidge include: madrone (Arbutus menziesii), coast-live oak,(Quercus agrifolia), wax myrtle (Myrica californica), bay

laurel (Umbellularia californica), tanoak (Lithocarpusdensiflorus) and golden chinquapin (Chrysolepischrysophylla) (Forrestel et al. 2011). Shrub species include:California coffee berry (Rhamnus californica), coyote brush(Baccharis pilularis), blue blossom ceanothus (Ceanothusthyrsiflorus), huckleberry (Vaccinium ovatum), salal(Gaultheria shallon), and manzanita (Arctostaphylos spp).

LEDAPS regrowth pixels (in blue shades, Fig. 2) clearlydelineated the northern and eastern sections of burned forestareas from the 1995 Vision Fire in the PRNS (NPS 2005).This wildfire was the largest within the PRNS since it wasestablished in 1962, burning nearly 5,000 ha and more than

Fig. 2 Detail of LEDAPSresults for the Point ReyesNational Seashore (Point I inFig. 1). The perimeter line ofthe 1995 Vision Fire is shownin magenta. Red shaded pixelswere detected as disturbedvegetation, whereas blueshaded pixels were detectedas regrowing or restoredvegetation

C. Potter

15 % of the total park area from October 3-8, 1995. Forrestelet al. (2011) reported that proximity to pre-fire Bishop pinestands was the most important predictor of measured post-fire (Bishop pine) regeneration rates. The LEDAPSregrowth pixels detected in the northeastern-most sectionof the Vision Fire area match closely to the pre-fire Bishoppine and Douglas fir dominated areas (Forrestel et al. 2011).

The line of LEDAPS disturbance pixels (in red shades,Figs. 3 and 4) extending the length of Inverness Ridgeadjacent to and to the southwest of the main pre-fire forestedareas of the Vision Fire were mapped onto the 2001 NLCDmap (from Landsat imagery) and located as predominantlyevergreen forest and shrubland cover. Only 34 % of theLEDAPS disturbance pixels detected in the entire PRNSarea was classified as predominantly grassland/herbaceouscover in the 2001 NLCD map (Table 1), whereas 59 % ofthe LEDAPS disturbance pixels were classified as forest andshrubland cover within the entire park area. Less than 1 % ofthe LEDAPS disturbance pixels detected within the entirePRNS area was classified as predominantly wetland cover.

Based on site visits and first-hand visual observations bythe author along the Bear Valley and Drakes View trails in2012, there appeared to be a high rate of understory treecanopy browning in many of the LEDAPS disturbance pixelareas of northern Inverness Ridge (Photo 1). This LEDAPS-detected disturbance pattern in areas observed with diebackof tanoak and bay laurel trees was consistent with defolia-tion by sudden oak death (Phytophthora ramorum), anintroduced plant pathogen confirmed by Ramage et al.(2011) to have infected trees in survey plots in northernsections of Inverness Ridge. Sporulation of P. ramorum isfavored by high humidity and mild temperatures (Davidsonet al. 2005), suggesting that ideal conditions for this patho-gen spread may exist in the Douglas fir forests of InvernessRidge (Ramage et al. 2011).

For the entire PRNS area, LEDAPS results implied that13 % of all vegetated cover was disturbed in some detect-able manner between 1999 and 2009, where 9 % of allvegetated cover was detected as regrowing over the sameperiod. The majority of regrowing areas were located within

Fig. 3 Overlay of LEDAPSresults for the northern sectionof Inverness Ridge in the PRNSonto NLCD cover types, shownas forest in green shades,shrubland in light brownshades, and grassland/herbaceous in light tan shades.Creeks are shown in dark bluelines, roads as dashed lines. Redshaded pixels were detected asdisturbed vegetation, whereasblue shaded pixels weredetected as regrowing orrestored vegetation

Ten years of land cover change on the California coast

the Vision Fire area, although other isolated areas outsidethe wildfire perimeter in the PRNS that were mapped aspredominantly grassland/herbaceous cover in the 2001NLCD were detected with increases in perennial woodycover. These areas were located along Pierce Point Roadnear the historic L Ranch in the northern section of the park,and also along Home Ranch Creek in the southern section ofthe park. These locations merit further investigation forongoing processes of plant succession.

Olema Valley—Bolinas Ridge, GGNRA

The Olema Valley, extending from CA Highway 1 on itswestern boundary upwards toward Bolinas Ridge on its

Fig. 4 Overlay of LEDAPSresults for the southern sectionof Inverness Ridge in the PRNSonto NLCD cover types, shownas forest in green shades,shrubland in light brownshades, and grassland/herbaceous in light tan shades.Creeks are shown in dark bluelines, roads as dashed lines

Table 1 Percentage of NLCD cover types within LEDAPS classes forthe PRNS area. Rows total to 100 % when minority land cover classes(not shown) are included

LEDAPSClass

Developed Barren Forest Shrub/Scrub

Grassland Wetland

Disturbed 5.4 0.7 28.3 30.3 33.7 0.9

No Change 4.9 0.1 66.0 13.1 14.3 1.3

Regrowing 6.5 0.2 28.1 19.2 43.5 1.4 Photo 1 Tanoak stand near Drakes View Trail of the PRNS in 2012detected as a LEDAPS disturbed vegetation area from 1999 to 2009

C. Potter

eastern boundary, is part of the Northern District of theGGNRA now managed by PRNS in an arrangement betweenthe two park units (Rilla and Bush 2009). This Olema Valleysection of the GGNRAwas established in 1972 and includes anumber of active ranches with livestock grazing.

LEDAPS regrowth pixels (in blue shades, Figs. 5 and 6)were detected over much of the predominantly grassland/herbaceous cover from the 2001 NLCD map of the OlemaValley ranchland. Practically none of the NLCD forestcover areas that border the local creek drainages weredetected with LEDAPS change pixels (either disturbedor regrowing) in the Olema Valley ranchlands (Fig. 6).

Based on first-hand visual observations by the author in2012, LEDAPS regrowth areas were located in produc-tive herbaceous cover being invaded by sparse shrubcover at sites along CA Highway 1 (Photo 2). Theselocations also merit further investigation for processesof plant succession.

Crissy Field and Baker Beach, San Francisco

Crissy Field is located on the northern portion of the SanFrancisco Peninsula in the Presidio of San Francisco.Culminating with preparations for the 1915 Panama-Pacific

Fig. 5 Detail of LEDAPSresults for OlemaValley—Bolinas Ridge withinthe Golden Gate NationalRecreation Area boundaries(Point II in Fig. 1). Red shadedpixels were detected asdisturbed vegetation, whereasblue shaded pixels weredetected as regrowing orrestored vegetation

Ten years of land cover change on the California coast

International Exposition, the historical marsh at Crissy Fieldwas filled and the resulting land was used by the U. S. militaryuntil 1994, when the property was transferred to the NationalPark Service as part of the GGNRA. From 1997 to 2000, 40

acres (16 ha) of Crissy Field were restored, including a tidalmarsh and 22 acres (9 ha) of dune and dune swale habitat(GGNRA 2003). More than 176,000 cubic meters of fill wereremoved and a channel to the bay was opened in November1999. Nearly 100,000 native plants were replanted and mon-itored in the restored sites (Hamingson 2002)

The dune scrub habitat above Baker Beach in SanFrancisco has been identified as a Special Ecological Area(SEA) by resource managers of the GGNRA. Since 1995,the National Park Service and community volunteers havebeen working to restore the native plant cover on the BakerBeach and Lobos Creek dunes.

LEDAPS results showed that there was extensive resto-ration of perennial vegetation cover on Crissy Field, BakerBeach and Lobos Creek dunes between 1999 and 2009(Fig. 7). Over 1 mile (1.6 km) of waterfront in the parkwas detected as regrowth on Crissy Field. The perennialgreen cover on Crissy Field could result in part from irriga-tion of lawns, as well as restoration of new woody cover onthe southern sections and into the Presidio.

Fig. 6 Overlay of LEDAPSresults for OlemaValley—Bolinas Ridge ontoNLCD cover types, shown asforest in green shades, shrublandin light brown shades, andgrassland/herbaceous in light tanshades (Point II in Fig. 1).Creeks are shown in dark bluelines, roads as dashed lines

Photo 2 Grazed rangeland in the Olema Valley in 2012 LEDAPSregrowth vegetation area from 1999 to 2009

C. Potter

Fort Funston, San Francisco

Fort Funston is located west of Skyline Boulevard (Highway35) at the western edge of San Francisco on the Pacific Ocean.The Fort Funston and Nike Battery were transferred to theGGNRA from the Department of the Army in October 1972.The property covers 230 acres (92 ha). During military use inthe 1930s, much of the native plant cover at Fort Funston wasdestroyed. As part of its construction projects, the Armyplanted an exotic ground cover, ice plant (e.g., Carpobrotusedulis), to stabilize sand around the batteries. By the mid1960s, much of the present park was covered with invasiveplants (NPS 2000). Restoration began in 1976 by removing

ice plant and replanting with native propagules collected fromother areas in the park. Dogs have been permitted off‐leashovermuch of Fort Funston and Phillip Burton Beach, but mustbe on-leash in Bank Swallow habitat areas, which has led toconflicts between dog owners, the GGNRA, and conserva-tionists (McCall 2000).

Winter storms of 1997 and 1998 eroded the protectedcliff area where the Bank Swallow colony had traditionallynested. Park biologists tracked their movement to the un-protected cliff areas immediately to the south and, becausethe swallows are a threatened species in the State ofCalifornia, determined the new nesting sites also need pro-tection. The 6-acre (2.4 ha) protected bird nesting area is

Fig. 7 Detail of LEDAPSresults for Crissy Field areawithin the Golden Gate NationalRecreation Area boundaries(Point III in Fig. 1). Red shadedpixels were detected as disturbedvegetation, whereas blue shadedpixelswere detected as regrowingor restored vegetation

Ten years of land cover change on the California coast

located north of the beach trail above Battery Davis andwest of the Coastal Trail.

LEDAPS results showed that there was no detectablerestoration of perennial vegetation cover in the protected birdnesting areas of Fort Funston between 1999 and 2009 (Fig. 8).On the contrary, LEDAPS detected what appears to have beennoticeable erosion and plant cover loss on the coastal dunes inthe northern section of the park that includes much of theprotected bird nesting areas (S. Skartvedt, GGNRA, personalcommunication). Little data exist to determine the causes ofthis apparent dune erosion at Fort Funston, which is naturallysubjected to high winds and shifting sand formations on ayear-round basis.

Discussion

The LEDAPS methodology has proven to be effective formapping disturbance and regrowth patterns from wildfiresacross much of North America (Masek et al. 2008). Thepotential for changing climate along the California coast(i.e., increase in surface temperatures and declines in fogextent; Johnstone and Dawson 2010) makes the monitoringof such vegetation responses to wildfires more urgent thanever for the region. While many more studies are needed ofchanges in post-fire regrowth rates, Wells et al. (2004)reported that coastal scrub and chaparral in southernCalifornia has burned more frequently than other vegetation

Fig. 8 Detail of LEDAPSresults for Fort Funston withinthe Golden Gate NationalRecreation Area boundaries(Point IV in Fig. 1). Red shadedpixels were detected asdisturbed vegetation, whereasblue shaded pixels weredetected as regrowing orrestored vegetation

C. Potter

types during the course of the 20th century. They detected apattern of increased wildfire ignitions along the relatively low-elevation urban-wildland interface. As residential develop-ment continues in coastal communities of western MarinCounty and other counties to the north, how and wherevegetation grows back after wildfires can influence firemanagement plans for many years (Charles 1995).

Over the past decade, sudden oak death (SOD) hasaffected many tree species native to coastal California,but the most susceptible species seems to be tanoak(Rizzo et al. 2005). SOD-induced tanoak mortality isoccurring across a north–south distance of ca. 750 km,generally within 30 km of the Pacific coast (Rizzo et al.2005). There is concern that SOD-induced tanoak declinemay lead to long-term changes in forest stand structure,facilitating the invasion of exotic plants and affecting thefrequency and/or severity of wildfire via increased fuelloads (Ramage et al. 2011). The LEDAPS methodologycan fulfill much of the need for continual, low-cost monitoringof emerging damage to coastal forest stands, especially inlocations where the presence of SOD has been previouslyconfirmed or suspected to invade.

Documentation of vegetation restoration for positiveenvironmental and recreational impacts was demonstratedby the LEDAPS methodology for the GGNRA parks inSan Francisco County. The advantages of routine, large-scale Landsat mapping of changes in perennial vegetationcover (as opposed to occasional ground surveys) includesthe benefit that LEDAPS results are not dependent onrepeat cover type mapping, such as the effort undertakenfor the NLCD. Once the proper thresholds for ΔDI valueshave been determined for a given region, it is possible tocompute the highest probable areas of ecosystem disturbanceor vegetation regrowth/restoration every year for which cloud-free Landsat imagery is available.

Acknowledgements This work was supported by grants from NASAAmes Research. The author thanks Steven Skartvedt and Robert Steersof the National Park Service for assistance with image interpretationsand historical information on the GGNRA.

References

California Beach ErosionAssessment Survey (CBEAS) (2010) CaliforniaCoastal Sediment Management Workgroup, Sacramento, CA

Charles L (1995) Social and environmental issues in developingvegetation, and fire management plans, The Biswell Symposium:fire issues and solutions in urban interface and wildland ecosystems,USDA Forest Service Gen. Tech. Rep. PSW-GTR-158

Cohen WB, Goward SN (2004) Landsat’s role in ecological applicationsof remote sensing. BioScience 54:535–545

Crist EP, Cicone RC (1984) Application of the Tasseled-Cap concept tosimulated Thematic Mapper data. Photogramm Eng Remote Sens50:343–352

Davidson J, Wickland A, Patterson H, Falk K, Rizzo D (2005) Trans-mission of Phytophthora ramorum in mixedevergreen forest inCalifornia. Phytopathology 95:587–596

Evens JG (2008) Natural history of the Point Reyes Peninsula, 2ndedn. California Natural History Guides, University of CaliforniaPress, 978-0-520-25467-1

Forrestel AB, Moritz MA, Stephens SL (2011) Landscape-scale vege-tation change following fire in Point Reyes, California, USA. FireEcol 7(2):114–128

Golden Gate National Recreation Area (GGNRA) (2003) Crissy FieldRestoration Area Monitoring Program, Quality Assurance ProjectPlan. June 27, 2003

Hamingson E (2002) An analysis of Crissy Field year 2000 dunevegetation monitoring. Report completed for SFSU Biology 458(Biometry), December 2002

Hanak E, Moreno G (2012) California coastal management with achanging climate. Clim Chang 111(1):45–73

Healey SP, Cohen WB, Zhiqiang Y, Krankina ON (2005) Comparisonof Tasseled-Cap-based Landsat data structures for use in forestdisturbance detection. Remote Sens Environ 97:301–310

Huang C, Wylie B, Yang L, Homer C, Zylstra G (2002) Derivation of aTasseled-Cap transformation based on Landsat-7 at-satellitereflectance. Int J Remote Sens 23:1741–1748

Johnstone JA, Dawson TE (2010) Climatic context and ecologicalimplications of summer fog decline in the coast redwood region.Proc Natl Acad Sci. doi:10.1073/pnas.0915062107

Kauth RJ, Thomas GS (1976) The Tasseled-Cap—a graphic descrip-tion of the spectral–temporal development of agricultural crops asseen by Landsat. Proceedings, Symposium on Machine Process-ing of Remotely Sensed Data. LARS, West Lafayette, pp 41–51

Kildow J, Colgan CS (2005) California’s ocean economy, report to theresource agency, State of California, National Ocean EconomicsProgram, July 2005

Masek JG, Huang CQ, Wolfe R, Cohen W, Hall F, Kutler J, Nelson P(2008) North American forest disturbance mapped from a decadalLandsat record. Remote Sens Environ. doi:10.1016/j.rse.2008.02.010

McCall R (2000) Dogs vs. birds: negotiated rulemaking at FortFunston. Hastings Coll Law West-Northwest J Environ LawPolicy 13:187–193

National Park Service (NPS) (2005) Vision fire: lessons learned fromthe October 1995 fire. Point Reyes National Seashore, U.S.Department of the Interior, 98 pp

National Park Service (NPS) (2000) Golden Gate National RecreationArea. Notice of proposed year-round closure at Fort Funston andRequest for comments. Brian O’Neill, Superintendent, GGNRA.July 17, 2000

Rahmstorf S (2007) A semi-empirical approach to projecting futuresea-level rise. Science 315:368–370

Ramage BS, Forrestel AB, Moritz MA, O’Hara KL (2011) Sudden oakdeath disease progression across two forest types and spatialscales. J Veg Sci. doi:10.1111/j.1654-1103.2011.01340

Rilla E, Bush L (2009) The changing role of agriculture in Point ReyesNational Seashore. University of California Cooperative Exten-sion, Novato, 25 pp

Rizzo DM, Garbelotto M, Hansen EM (2005) Phytophthora ramorum:integrative research and management of an emerging pathogen inCalifornia and Oregon forests. Annu Rev Phytopathol 43:309–335

Wells ML, O’Leary JF, Franklin J, Michaelsen J, McKinsey DE (2004)Variations in a regional fire regime related to vegetation type inSan Diego County, California (USA). Landsc Ecol 19:139–152

Ten years of land cover change on the California coast