Journal of Coastal Conservation

, Volume 18, Issue 3, pp 213–220

Monitoring the production of Central California coastal rangelands using satellite remote sensing

Authors

Article

DOI: 10.1007/s11852-014-0308-1

Cite this article as:
Potter, C. J Coast Conserv (2014) 18: 213. doi:10.1007/s11852-014-0308-1

Abstract

There is a long history of livestock grazing on the California Central Coast, dating back over 150 years. In this study, methods were reviewed and results presented for analysis of NASA’s Moderate Resolution Imaging Spectroradiometer (MODIS) satellite sensor data to monitor year-to-year variation of forage production on Central Coast rangelands around Big Sur, California. Time series plots from 2000 to 2012 of vegetation greenness for ten rangeland sites showed similar inter-annual patterns in satellite yield index (SYI) values. Most sites reached their maximum greenness levels each year in early May. The year with the highest observed SYI level across most sites was 2005. In the northern portion of the region (north of Pfeiffer Big Sur State Park), 2007 was the year with the lowest observed SYI level, whereas in the southern allotments, 2007 was a year with a relatively high SYI level. These methods have the potential to monitor the differing seasonal growing cycles of rangeland production across the area of individual grazing allotments on the Central Coast. Such a cost-effective and timely approach is required for conservation monitoring in the Big Sur coastal ecosystems where rapid climate change may shift vegetation cover in favor of more extensive rangelands at the expense of forested lands.

Keywords

RangelandsMODISCentral CoastCaliforniaCoastal climate

Introduction

Rangelands on the Central California coast require continual monitoring for human and environmental impacts on native plant communities, soil conservation, riparian zones, and water quality, due to their unique climate and topography. Herbaceous plant growth on the Pacific coast can be highly variable from year-to-year and is generally limited by declines of soil moisture in the summer and by cool temperatures in the winter Li et al. (2012). The annual production pattern for coastal grasses is rapid growth in the late fall (November) after the first rains have returned, slow winter growth (December–February), and rapid growth again in spring (March–May) (George et al. 2001).

The U. S. Forest Service (USFS) on Los Padres National Forest (LPNF) administers a rangeland allotment program through the issuance of term livestock grazing permits (USFS 2004). The permits are generally issued for a period of 10 years (Federal Land Policy and Management Act 1976, as amended). Management of the LPNF rangeland program is authorized by the Multiple Use Sustained Yield Act 1960, the Forest and Rangeland Renewable Resources Planning Act 1974, the Public Rangelands Improvement Act 1978, and the Land Resource and Management Plan (LRMP).

The USFS has identified the need to effectively monitor multiple rangeland properties that are in remote and rugged terrain (USFS 2004). In this study, satellite data analysis from the NASA MODIS sensor was used to develop rapid, low-cost grasslands monitoring methods for rangeland production on the California Central Coast that can be applied anywhere in the world. The objectives of the study were to quantify seasonal growing cycles and assess sources of yearly variation in rangeland forage production for the past 12 years on individual USFS allotments and nearby pasture properties. The long-term purpose is to demonstrate to local conservation managers a cost-effective coastal ecosystem monitoring that can account for climate change and the frequency of extreme weather events, which have the potential to tip the balance in favor or against such vulnerable biological resources and the human activities that sometimes struggle to prosper in a remote and rugged coastal environment.

Site descriptions

The main study area was in the central Pacific coastal region near Big Sur, CA (Fig. 1). The region has a Mediterranean climate of warm, dry summers and cool, wet winters, with localized summer fog near the coast (Hickman 1993). Average annual rainfall varies from 400 to 1,520 mm throughout the region, with highest amounts falling on the higher mountains in the northern extreme of the study area during winter storms. During the summer, fog and low clouds are frequent along the coast. Mean annual temperature ranges between 10.0 and 14.4 °C, from shaded canyon bottoms to exposed ridge tops, respectively. Cattle and sheep ranching began in these coastal grasslands in the 1820s (Burcham 1957). The many unique microclimates of Big Sur result in high natural biodiversity (Henson and Usner 1996), including many endemic rare and endangered species, such as the wild orchid Piperia yadonii, the California condor (Gymnogyps californianus), California southern sea otter (Enhydra lutris nereis), and southernmost habitat of the coast redwood (Sequoia sempervirens).
https://static-content.springer.com/image/art%3A10.1007%2Fs11852-014-0308-1/MediaObjects/11852_2014_308_Fig1_HTML.gif
Fig. 1

Rangeland sites and USFS allotments in the central coastal region of California. The background map shows 2010 USDA land cover types. Inset map of study locations within the shaded area of Monterey County

A total of ten rangeland sites, most of which were listed as priorities by the USFS (2004), were selected for this study, with location coordinates shown in Table 1. Rangeland allotment locations for MODIS analysis were identified for study by the LPNF rangeland program and precisely delineated from topographic maps marked as such by the USFS (2004). Records of daily grazing management from 2000 to the present were not available for most all of the LPNF allotments, but if the Brazil Ranch (Photo 1) can serve as a typical example, then the following stocking and rotation practices would have been commonly followed (J. Kwasny, personal communication). About 30 cow/calf pairs plus 2 bulls would be allowed to graze one pasture section of an allotment from December 1 thru January 15, and then moved to another pasture section along with 30 more cow/calf pairs with bulls until May 15. These 60 total head would be moved back to the first pasture section to graze into until June 15, after which time all grazing on the allotment would be suspended until the following December.
Table 1

Rangeland sites near Big Sur, CA, order by location from north to south

Rangeland site

Latitude (dd N)

Longitude (dd W)

Palo Corona Ranch

36.470253

−121.878208

Brazil Ranch

36.351034

−121.890690

El Sur

36.301831

−121.838530

Torre Allotment

36.215303

−121.734679

Twitchell Allotment

36.027799

−121.516430

Gorda Allotment - Mill Creek Unit

35.980471

−121.469962

Alder Creek Allotment

35.870916

−121.418239

Buckeye Allotment

35.811518

−121.343864

Salmon Creek Allotment

35.811228

−121.346124

San Carpoforo Allotment

35.777669

−121.283760

https://static-content.springer.com/image/art%3A10.1007%2Fs11852-014-0308-1/MediaObjects/11852_2014_308_Fig2_HTML.gif
Photo 1

Rangeland vegetation cover at the Brazil Ranch site

Data sets

Climate data

Annual and seasonal climate summaries from 2000 to 2012 were obtained from the Western Regional Climate Center (WRCC) California Climate Tracker (Abatzoglou et al. 2009) data portal (www.wrcc.dri.edu/monitor/cal-mon). Total yearly precipitation amounts and spring seasonal (MAM) temperature WRCC summaries showed that 2005 was a relatively warm and wet year, 2007 was a relatively dry year, and 2010 was a relatively wet cool year on the Central Coast (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs11852-014-0308-1/MediaObjects/11852_2014_308_Fig3_HTML.gif
Fig. 2

Annual climate summaries for the California central coast a total precipitation, inches × 1,000 b average spring temperature (MAM), degrees F × 100. Data shown as departures from the 1949–2005 base period (Abatzoglou et al. 2009)

Satellite remote sensing data

Data products from the Moderate Resolution Imaging Spectroradiometer (MODIS) were acquired at a 16-day (bi-weekly) interval from 2000 to 2012 for the study sites. The MODIS MOD13Q1 product subset was obtained from the Oak Ridge National Laboratory Distributed Active Archive Center. The 250-m spatial resolution MOD13Q1 collection 5 Enhanced Vegetation Index (EVI) products have been derived from atmospherically corrected bi-directional surface reflectances that have been masked for water, clouds, heavy aerosols, and cloud shadows (ORNL 2011). Bi-weekly composite EVI values were scaled from 0 (barren vegetation cover) to 1,000 (highest potential vegetation cover).

EVI was developed to optimize the greenness signal, or area-averaged canopy photosynthetic capacity, with improved sensitivity in high biomass regions. As a ratio of the red to near infra-red reflectance bands, the EVI incorporates the removal of soil-brightness induced variations, as in the Soil Adjusted Vegetation Index (SAVI) (Huete et al. 2002). EVI has been found useful in estimating absorbed photosyntheically active radiation (PAR) related to chlorophyll contents in vegetated canopies and has been shown to be highly correlated with processes that depend on absorbed light, such as gross primary productivity (GPP) measured from tower flux instruments (Rahman et al. 2005; Xiao et al. 2004).

Validation for California rangelands

The best available validation of MODIS EVI inputs that form the basis of our SYI calculation was first carried out. While a single measured tower flux comparison cannot be a strict representation of how well MODIS represents the entire study region, one can logically select the MODIS pixels closest to the Ameriflux tower for an EVI or carbon model comparison, since those pixels are most likely to best represent the measured tower C flux footprint.

Validation of net primary production (NPP) predicted by from MODIS EVI for California rangeland was carried out by Li et al. (2012) using measured CO2 fluxes from the nearest Ameriflux tower site (Heinsch et al. 2006), the Tonzi Ranch study site (38.42 dd N, −120.95 dd W, at 129 m elevation) located 245 km northeast of Big Sur. This site was an oak-grass savanna consisting of scattered blue oak trees (Quercus douglasii), with occasional gray pine trees (Pinus sabiniana L.), and grazed grassland (Brachypodium distachyon L., Hypochaeris glabra L., Bromus madritensis L. and Cynosurus echinatus L.). Grass grows from November to May, after which it senesces rapidly.

Much like on the Big Sur coast, climate at the Tonzi Ranch site is Mediterranean with a mean annual temperature of 16.3 °C and total annual precipitation of 560 mm. Carbon dioxide fluxes, water vapor, and meteorological variables were measured continuously at Tonzi Ranch for several years using eddy covariance systems. Monthly NPP was calculated based on year-round tower flux measurements (Ma et al. 2007). AmeriFlux eddy-correlation data sets were obtained from the central data repository located at the Carbon Dioxide Information Analysis Center (CDIAC; http://public.ornl.gov/ameriflux/dataproducts.shtml). Level 4 AmeriFlux records contained gap-filled and ustar filtered records, complete with calculated gross productivity on varying time intervals including hourly, daily, weekly, and monthly with flags for the quality of the original and gap-filled data.

The nine closest MODIS EVI 250-m pixel values to the center tower location were extracted for these NPP validation outputs (from the CASA model algorithms reported by Li et al. 2012) and combined as monthly average NPP from 2005 to 2007. Results showed that observed tower fluxes and the predicted monthly NPP fluxes from MODIS EVI were significantly correlated across all seasons of the Tonzi Ranch measurement period by Li et al. (2012), with a coefficient of determination (R2) of 0.89 and root mean square error (RMSE) of 8.8 g C m−2 (Fig. 3). MODIS-predicted peak summer NPP fluxes in 2007 were somewhat lower than the estimated tower flux values. This may have been related to extreme drought tolerance of the mixed savanna ecosystem under the low precipitation amounts recorded in 2007, compared to the previous 2 years (Daly et al. 2008).
https://static-content.springer.com/image/art%3A10.1007%2Fs11852-014-0308-1/MediaObjects/11852_2014_308_Fig4_HTML.gif
Fig. 3

Comparison of NPP predictions (mean value of 250-m MODIS pixel inputs, n = 9) with tower flux measurements for Tonzi Ranch from 2005 to 2007 (Li et al. 2012). Units are g C m−2 month−1

Satellite Yield Index methodology

Bi-weekly MODIS EVI values covering the Central Coast growing season from the September 30 to June 25 were summed for each year (2000 to 2012) to generate a rangeland Satellite Yield Index (SYI) for all locations listed in Table 1. The optimized SYI time period was determined from the observed rangeland production data shown in Fig. 2. Any missing EVI values (due to fog or cloud cover) that were surrounded by valid EVI values were filled-in with the previous bi-weekly EVI value.

Results and discussion

The EVI time series plots for Central Coast rangeland sites showed similar inter-annual patterns in SYI values (Fig. 4). Most sites reached their maximum EVI levels each year in early May. The year with the highest observed SYI level across most sites was 2005. For the four sites in the northern-most portion of the region (north of Pfeiffer Big Sur State Park), 2007 was the year with the lowest observed SYI level, whereas in the southern allotments, 2007 was a year with a relatively high SYI level.
https://static-content.springer.com/image/art%3A10.1007%2Fs11852-014-0308-1/MediaObjects/11852_2014_308_Fig5a_HTML.gifhttps://static-content.springer.com/image/art%3A10.1007%2Fs11852-014-0308-1/MediaObjects/11852_2014_308_Fig5b_HTML.gif
Fig. 4

Inter-annual variations in the MODIS 250 m SYI (orange lines, left axis scale) and monthly EVI (green lines, right axis scale) for ten rangeland locations listed in Table 1

The Brazil Ranch site was estimated to be the most consistently productive rangeland location among those analyzed, with SYI values frequently exceeding 8,000. During the past 100 years, the Brazil Ranch was privately managed for cattle and horses. After the property was proposed for multiple-unit residential development, Brazil Ranch was purchased by the conservation community with public funding in 2002 to protect its scenic and other natural resource values. In September 2002, the Ranch property officially passed to the USFS.

The Alder Creek, Salmon Creek, and San Carpoforo allotments were estimated to be the least consistently productive rangeland sites among those analyzed, with SYI values rarely exceeding 6,000. Herbaceous cover appeared to grow (or recover) more slowly in the spring season (MAM) at these sites, compared to those in the northern portion of the region. These southern allotments were not as variable over the years 2000 to 2011 in the yearly SYI estimates as the northern range sites. Unusually heavy cloud cover in 2012 obscured the EVI observations across several of the southern allotment sites.

As climate changes occurs in the Pacific region (Abatzoglou et al. 2009; Hiatt et al. 2012), the production and health of coastal rangelands and surrounding vegetation may be altered in unexpected ways. The SYI from MODIS satellite imagery provides a rapid, cost effective means to monitor plant growth on rugged, inaccessible hill slopes of the Pacific Coast that even other satellite sensors, such as Landsat cannot match. The continuous image processing and robust atmospheric correction that MODIS EVI data sets offer to scientist and managers is unique (Li et al. 2012).

Conversion of the MODIS SYI to accurately estimate annual forage biomass production is a step many land managers desire and depends on calibration to field measurements of forage at selected rangelands on the Pacific coast. George et al. (2001) reported that the average annual forage biomass production at the University of California Hopland Research Center (39.00 dd N, −123.67 dd W) was 2,300 lb ac−1 (2,576 kg ha−1), with a range from 900 to 3,500 lb ac−1 (1,008 to 3,920 kg ha−1). The average annual precipitation at Hopland is 935 mm, which is well within the mid-range of the 400 to 1,520 mm year−1 typically measured along the Big Sur Coast (Li et al. 2012). Based on the general findings of George et al. (2001) that California forage production on coastal rangelands is most strongly influenced by fall and winter precipitation and winter-early spring temperatures, the range of SYI computed in this study of 5,000 to 9,000 EVI units would correspond to a range of 1,000 to 5,000 kg ha−1 year−1 in forage production for Central Coast grazing allotments. This forage production function is consistent with measurements by the author on the Brazil Ranch rangeland in 2007 of standing herbaceous biomass. The MODIS SYI of 6400 in 2007 corresponded to a measured standing forage biomass mean value of 3,955 kg ha−1 (n = 15 samples) from May–June 2007.

The seasonal distribution of cattle and (moderate) intensity of grazing on the Brazil Ranch grasslands has been closely managed over the past decade. On all LPNF allotments, livestock distribution and grazing intensity presents ongoing challenge because of steep terrain, fresh water shortage, and the (large) size of pastures. It is also worth noting that measurement studies at Brazil Ranch reported measureable summer precipitation in the form of frequent and heavy coastal fog (Hiatt et al. 2012), which may be common in managed grassland locations along the Big Sur coast.

According to the USFS (2004; and J. Kwasny, USFS personal communication), the Torre, Twitchell, and Buckeye allotments were closed to grazing as a result of the 2004 Coastal Rangelands Analysis. The Gorda Allotment - Mill Creek has been available for grazing, but has not been used since 2005. Grazing has been occurring on a regular basis on the Palo Corona, Brazil Ranch, El Sur, Alder Creek, Salmon Creek (west of Highway 1 only), and San Carpoforo rangelands. Examination of the MODIS SYI results (Fig. 4) in this context suggests that the rangeland sites that have been consistently grazed over the past decade also showed SYI values more consistently above 6,000, equivalent to >3,900 kg ha−1 annual forage production.

The remote sensing methods presented in this paper track the differing seasonal growing cycles of rangeland forage production across the individual rangeland properties selected on the Big Sur coast. The MODIS-derived SYI computed in this study can uniquely integrate the effects of precipitation, livestock management, terrain, and soil properties into one consistent historical assessment metric. However, since MODIS satellite data provide a composited estimate of herbaceous green cover for any given month (rather than a cumulative change over the entire grazing period), the SYI may be limited in the capacity to directly estimate the amount of forage biomass consumed by cattle on any of the rangelands sites. That additional forage biomass produced (but rapidly consumed by livestock) would add to any measured production estimate in the field.

It is important nonetheless to place the results from this study within the context of current and future climate change on the Pacific coast. On a statewide basis, surface air temperatures have been warming between 1950 and 2005 (LaDochy et al. 2007). Almost all increases detected were due to changes in daily minimum temperatures (Tmin) in the dry summer season, since daily maximum temperatures (Tmax) showed either no change or cooling. The situation is reversed on the California coast, where a current hypothesis is that observed coastal California cooling of Tmax values derives from greenhouse gas-induced regional warming of the inland Central Valley and Sierra Nevada foothill areas, resulting in increased on-shore sea-breeze flow, which in turn can overwhelm GHG warming of coastal areas (Lebassi et al. 2009). This hypothesis is consistent with increased upwelling of cold ocean water off the California coast that further enhances cool sea-breeze flows and associated low stratus cloud cover.

It is common to find temperature inversions on the Central Coast that frequently result from the persistent marine layer, and which remain very stable between May and October (Johnstone and Dawson 2010). Summer fog is common below about 600 m (2,000 ft) elevation (Hiatt et al. 2012). Nonetheless, diurnal heating and cooling of ecological habitats along the Big Sur coast and higher into the Santa Lucia mountains involves a complex series of elevational gradients and topographic variations. In general, steep slopes of coastal scrub vegetation mixed with open grasslands starting just above sea level give way to mixed hardwood (oak) forests at around 600 m elevation, which in turn transition to pine (Pinus ponderosa and Pinus coulteri) woodlands at around 900 m (3,000 ft) elevation. In the canyon bottoms and on the lower parts of north-facing slopes, coast redwood (Sequoia sempervirens) stands are able to grow. An expansion upward in elevation of these coastal cooling temperature trends could force mixed hardwood and pine forests to contract further into the mountain ridge tops and favor expansion of potential rangeland habitat.

Year-round cooling patterns (as opposed to summer only) may have an effect on maintaining river and steam flows on the Pacific coast, as well as stabilizing the growing season initiation date and duration for native plant species in the era of future global warming. If the cooling patterns observed in on the coast since the mid-1990s (Lebassi et al. 2009) continue as a long-term climate changes along the California coastal region, the productivity and health of coastal rangelands and surrounding vegetation habitats may be altered in unanticipated ways.

In conclusion, freely available satellite image data together with NPP modeling methods (Li et al. 2012) can be used to monitor the differing seasonal growing cycles of rangeland forage production across rangeland properties on the central California coast. With further development, these methods combing MODIS and Landsat images can provide continual, low-cost, and standardized monitoring indices of conservation impacts on residual dry matter dynamics and native plant communities for any rangeland in the Pacific coastal region.

Acknowledgments

The author is grateful to the U. S. Forest Service, Los Padres National Forest, for whom this study was conducted, and particularly to Ecosystem Manager Jeff Kwasny for information on Central Coast rangeland management.

Copyright information

© Springer Science+Business Media Dordrecht 2014