Associated with possible climatic changes there is an increasing demand for an easy tool to map the age of vegetation stands, which differ in physiological function and therefore may contribute differently to the global carbon cycle. The analysis of aircraft or satellite Remote Sensing imagery provides a non-intrusive and cost effective method of mapping certain vegetation parameters such as vegetation cover and biomass over large areas. The present study provides the necessary ground truthing for distinguishing stand ages on landscape scale imagery. Using spectral reflectance at leaf and stand scales, we explored the ability of several spectral indices to distinguish physiological differences associated with the stand age of Southern California chaparral vegetation. These indices included NDVI (a "greenness" index), WBI (a measure of water content), and PRI (a measure of the xanthophyll cycle pigment activity and relative photosynthetic rate). Reflectance measurements were made with a field spectrometer at 1-meter intervals along 100-meter transects and leaf-photosynthesis was assessed on two representative species (Adenostoma fasciculatum and Ceanothus sp.) with a portable gas exchange system, which measures photosynthetic uptake of carbon dioxide. Our data suggest that differently aged canopies can be differentiated by spectral indices. However, 20 year old and more than 40 year old sites were hard to distinguish. NDVI, WBI and PRI peak in the wet season and decrease differently during the dry season depending on stand age. NDVI and WBI differentiate best between stand ages during the wet and the early dry season, whereas PRI separates young and mature sites in the wet season. The analysis of NDVI-variability within the stand ages revealed that the heterogeneity in the open canopies of the 3-year-old stands was constantly higher than in the 20 years and older stands. Further work will include a more thorough analysis of scaling effects from leaf to canopy and landscape scales.
 
 

Frequent fires lead to a mosaic of different stand ages in Southern California chaparral vegetation

Frequent fires in Southern California chaparral lead to a mosaic of different stand ages, which vary in species composition and vegetation height (biomass). The present study explored the ability of Remote Sensing to detect the age of Southern California chaparral vegetation at the canopy scale. By comparing the spectral reflectance of differently aged canopies the during the dry and wet seasons, this study provides the needed ground-validation for earlier work at the landscape scale (e.g. Qiu et al., 1998). These studies suggested contrasting optical properties across successional stages of chaparral. Successional change in species composition as well as development and senescence of individual plants may cause physiological and structural differences between stand ages, which are detectable by a unique spectral signature.

Recent fires are easily detectable on an airborne spectral reflectance image. The present study provides the ground validation to distinguish optical properties across stand ages.
 
 

Field Spectrometer (UniSpec, PP Systems, Haverhill, MA.): to measure spectral reflectance at the canopy scale the tip of the fiber optic was mounted on an extendable handle. Holding the handle 2m above the canopy will result in a 1 m diameter field of view.

Portable gas exchange system (LICOR-6400): photosynthetic rates are measured as rates of carbon dioxide- uptake (m mol CO₂ m^-1 s^-1)

Reflectance indices characterize the vegetation spectrum and may be able to distinguish physiological differences:

NDVI = (R (800) – R(680)) / (R(800) + R(680)) : a "greenness" index

WBI = R(970) /R(900) : a measure of water content

PRI = (R(531) – R(570)) / (R(531) + R(570)): a measure of relative photosynthetic rate
 
 

Representative stand ages of Southern California chaparral, ranging from new burns to 3, 20 and more than 40 year old vegetation, were monitored at three replicate study sites in the Santa Monica Mountains near Los Angeles. Using a field spectrometer, canopy spectral reflectance was examined at 1-meter intervals along 100-meter transects during the dry and wet seasons. The study sites were similar in terms of aspect, elevation and distance from the ocean. At the leaf scale, we measured spectral reflectance and photosynthetic gas exchange of top canopy leaves of two representative species (Adenostoma faciculatum and Ceanothus sp.) at three evenly distributed points along the 100-meter transects.
 
 

Stand ages differ in species cover and vegetation height:

Figure 2: Average vegetation height, which serves as an estimate of biomass, was highest at the 20 year old study sites. In >40 year old stands lower perennials reduce the average height.

Depending on the intensity of fire, post- fire succession may result in either a rapid regeneration of the previously existing vegetation or the temporary ascendancy to dominance of species, which are more characteristic of coastal sage scrub (Holland and Keil, 1995).

Commonly herbaceous annuals and perennials, together with shrub-seedlings and sprouting root crowns, form the first successional stage. In mixed chapparral they are successsionally replaced by chamise (Adenostoma faciculatum) shrubs, which have developed a sparse canopy 3 years after fire. During further succession Ceanothus sp. and chamise (Adenostoma faciculatum) form a closed canopy. By the end of the first decade after fire the ground surface covered by shrubs approaches prefire levels (Hanes, 1977). In older stands the shrub canopy becomes sparser again and herbaceous perennials fill the gaps between the shrubs.
 
 

Estimating physiological differences between differently aged canopies by measuring photosynthesis of two representative species on leaf scale
 
 

The successional change in species composition as well as the development and senescence of individual plants most likely cause physiological differences between stand ages. Lacking the opportunity to directly measure canopy photosynthetic rates by flux towers, we measured photosynthetic gas exchange of small branches on Adenostoma fasciculatum and Ceanothus sp. across stand ages and normalized for the measured leaf area (Fig. 3). Overall photosynthetic rates of Adenostoma fasciculatum were lower than those of Ceanothus sp.. Resprouts and seedlings respectively of both species, which occur shortly after fire, showed the highest photosynthetic rates during the wet season, but the rates of seedling photosynthesis sharply decreased after the wet season. Gas exchange rates of branches from 3- and >40-year old plants were lower at the peak of the dry season than those of 20-year old shrubs.

Figure 3: Photosynthetic gas exchanged measured as the CO₂ use per unit leaf area during wet (March-April) and dry season (Jan-Feb, May-July).

Distinguishing stand ages by the seasonal course of spectral reflectance indices

The spectral reflectance indices of the transect canopies followed a clear seasonal trend and demonstrated differences between stand ages (Fig. 4). A repeated measures ANOVA confirmed a significant difference between stand ages. NDVI (a "greenness" index), WBI (a measure of water content) and PRI (indicating photosynthetic activity) peaked in the wet season and continuously decreased during the dry season. New burns and 3-year old stands were clearly distinguishable from 20-year and older stands. Stands older than 40-years were hard to distinguish from the 20-year old sites. The estimated canopy gross photosynthetic flux was consistently higher in 20-year and older sites than that in 3-year old sites and new burns. The new burns showed the lowest photosynthetic fluxes on transect scale. Individual plants in newly burned areas had high photosynthetic rates, which may be due to low competition (Fig. 1), but a sparse cover reduces the average transect photosynthetic activity.

After having completed the data over the entire year, further work will include Principle Component and Discriminant Analysis to quantify the distinction of stand ages.
 

Figure 4: Seasonal course of the spectral reflectance indices across stand ages. Multiplying NDVI by a scaled PRI provides an estimate of the gross photosynthetic flux of the transect-canopy.

Distinguishing stand ages by the seasonal course of the NDVI variability

Young stands of mixed chaparral form a sparse canopy of herbaceous annuals and perennials. With age they contain an increasing amount of shrubs, predominantly Adenostoma fasciculatum and Ceanothus sp., which create a closed canopy by the end of the first decade after fire (Hanes, 1977). In older stands the shrub canopy becomes sparser again and allows herbaceous perennials to invade. Therefore the spectral variability within the stand ages seems to be a good mean to distinguish stand ages. NDVI, which describes differences in vegetation structure, is suitable to express the variability within differently aged stands (Fig. 5).

When comparing samples with very different means, the coefficient of variance is more useful than the variance. Since the standard deviation is an absolute measure of dispersion and thus is a direct function of mean the coefficient of variance provides a relative (rather than an absolute) measure of dispersion and is independent of the mean.

The variability of NDVI increased steadily in the new burns as herbaceous annuals and perennials settled on the sites. Single evergreen shrubs that survived the fire kept heterogeneity high during the dry season. The NDVI variability in the open canopies of the 3 year old stands was consistently higher than in the 20 year and older stands, which may be due to a patchy vegetation cover and eventually blooming annuals and perennials. It decreased after the wet season, when herbaceous annuals and perennials were drying up . The difference in NDVI variability between 3 year and more than 20 year old stands was most clear at the peaks of the wet and the dry season. During the wet season the 20 year and older stands developed a closed green canopy, which drove their heterogeneity to its minimum. In contrast, young stands showed high NDVI variability during the wet season.
 
 

Figure 5: Seasonal course of the NDVI coefficient of variance: CV = (standard deviation * 100) / mean

• Hanes, T.L. (1977). California chaparral. In: M.G. Barbour and J. Major. Terrestrial Vegetation of California. Wiley & Sons. Edition 1988, pp. 417-469.
• Holland, V.L., Keil, D.J. (1995) California Vegetation, Kendall/Hunt, Dubuque, Iowa.
• Qiu H-L, Gamon JA, Luna M, Roberts DA (1998). Monitoring post fire succession in the Santa Monica Mountains using hyperspectral imagery. Proceedings of SPIE Vol. 3502 - Hyperspectral Remote Sensing and Application.