Monitoring the photosynthetic activity of vegetation from remote sensing data

Monitoring the photosynthetic activity of vegetation from remote sensing data

Advances in Space Research 38 (2006) 2196–2202 www.elsevier.com/locate/asr Monitoring the photosynthetic activity of vegetation from remote sensing d...

970KB Sizes 1 Downloads 53 Views

Advances in Space Research 38 (2006) 2196–2202 www.elsevier.com/locate/asr

Monitoring the photosynthetic activity of vegetation from remote sensing data N. Gobron *, B. Pinty, M. Taberner, F. Me´lin, M.M. Verstraete, J.-L. Widlowski Global Vegetation Monitoring Unit, Institute for Environment and Sustainability, Joint Research Center, TP 440, via E. Fermi, 21020 Ispra (VA), Italy Received 1 October 2002; received in revised form 30 June 2003; accepted 10 July 2003

Abstract The state of terrestrial vegetation has been monitored using remote sensing data for decades. Information was often derived from empirical tools, like vegetation indices, which are very sensitive to perturbations and often depend on the spectral properties of the sensor. Advances in the understanding of radiation transfer and the availability of higher performance instruments have stimulated the development of a new generation of geophysical products poised to provide reliable, accurate information on the state and evolution of terrestrial environments. A series of optimized algorithms have been developed for documenting biophysical activities, using a physically based approach (specifically, to estimate Fraction of Absorbed Photosynthetically Active Radiation (FAPAR)) for various instruments. The outline of the methodology will be summarized and the results from an application conducted with SeaWiFS data will be presented.  2005 COSPAR. Published by Elsevier Ltd. All rights reserved. Keywords: FAPAR; Remote Sensing; SeaWiFS; Vegetation

1. Introduction Remote sensing space can help monitor the characteristics of the landscape and their temporal evolution at a variety of spatial scales. The algorithms used to retrieve surface properties from satellite data, in the optical domain, focus on the interpretation of the spectral signature of the various objects that interact with the solar radiation ultimately collected by a sensor. A generic approach to rationalize the design of spectral indices has been described in Verstraete and Pinty (1996). Exploiting the physics of the signal and radiative transfer models, new optimized algorithms for monitoring land surfaces have been developed for various sensors (Gobron et al., 2000, 2001, 2002). The design of these algorithms is based on the following requirements: *

Corresponding author. Tel.: +39 332 78 6338; fax: +39 332 78 9073. E-mail address: [email protected] (N. Gobron).

(1) they should exhibit a maximum sensitivity to the presence and changes in the properties of healthy live green vegetation and (2) they should not be sensitive to atmospheric scattering and absorption effects, to soil color and brightness changes, and to temporal and spatial variations in the geometry of illumination and observation. The algorithms must also be equivalent to each other in the sense of yielding the same bio-geophysical information when they are applied to their respective data. These constraints imply that the land products should summarize the status of a given terrestrial system with a single number whose value must, by necessity, reflect only broad characteristics of the terrestrial ecosystems and their main changes. The state and evolution of terrestrial ecosystems are characterized by a large number of physical, biochemical and physiological variables. The Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) acts as an integrated indicator of the status of the plant canopy that can be retrieved by remote sensing techniques with acceptable

0273-1177/$30  2005 COSPAR. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.asr.2003.07.079

N. Gobron et al. / Advances in Space Research 38 (2006) 2196–2202

accuracy. This has motivated the hypothesis that the FAPAR variable can be used to quantify the presence of vegetation with good reliability on a global scale, a variable of major importance for researchers. After presenting the mathematical optimization based on the physical modeling to retrieve the FAPAR value, the processing system developed to generate an ensemble of relevant information on the basis of the SeaWiFS data, to support various application projects dealing with the monitoring of land surfaces, is summarized. The last section shows some profiles corresponding to the time series of the FAPAR products over three years and indicates the practical potential of remote sensing to address environmental issues.

2. Physical modeling and mathematical optimization The estimation of a state variable from the signals measured by a given sensor is constructed from sensor-specific simulated data sets, representative of various land surfaces, using radiative transfer models of the coupled surface atmosphere system. This approach (see Fig. 1) defines a large number of simulated radiance fields, which can be sampled by a virtual instrument similar to the actual one in terms of the spectral and angular

2197

observing schemes. Similarly, the corresponding FAPAR values for the various terrestrial systems under investigation can be simultaneously estimated. In this case, simulations are made with a homogeneous canopy model (Gobron et al., 1997) representing land surfaces, coupled with the atmospheric model 6S (Vermote et al., 1998). Since green vegetation strongly absorbs solar radiation in the red spectral region, and strongly scatters it in the near infrared, these two bands are the main ones used to characterize land surfaces from remote sensing data. The reflectance in the blue band is sensitive to the aerosolsÕ optical thickness and, therefore, is used to decontaminate the red and the near-infrared bands from atmospheric effects. The design of the FAPAR algorithm is based on a two steps procedure where the spectral radiances measured in the red and near-infrared bands are, first, rectified in order to ensure their decontamination from atmospheric and angular effects and, second, combined together in a mathematical formulae to generate the FAPAR value. The top of atmosphere (TOA) channel values are first normalized by the anisotropy function, Eq. (1), to take into account the angular effects: ~ðki Þ ¼ q

qtoa ðX0 ; Xv ; ki Þ ; F ðX0 ; Xv ; k ki ; XHG ki ; qkic Þ

ð1Þ

where ki stands for the wavelength (blue, red or nearinfrared) of spectral band i, and qtoa(X0, Xv, ki) denotes the BRF at values measured by the sensor in the spectral band ki, as a function of the actual geometry of illumination (X0) and observation (Xv). These angular coordinates are fully defined by the zenith (h) and relative azimuth (/) angles for the incoming and outgoing radiations, respectively, for a plane-parallel system. The spectral anisotropy reflectance function, F ðX0 ; Xv ; k ki ; XHG ki ; qkic Þ represents the shape of the radiance field. The triplet ðk ki ; XHG ki ; qkic Þ are the RPVs parameters (Rahman et al., 1993) either optimized a priori for each spectral band ki, for mono angular instruments, or retrieved in the case of the multi-angular data, like with the MISR data (see Gobron et al., 2002). The rectification process of the red and near-infrared bands is performed as follows: ~ðkred Þ; qRred ¼ g1 ½~ qðkblu Þ; q

ð2Þ

~ðknir Þ; qRnir ¼ g2 ½~ qðkblu Þ; q

ð3Þ

where ~ðkj Þ ¼ P ðki ; kj Þ=Qðki ; kj Þ; gn ½~ qðki Þ; q

Fig. 1. Design of the optimization algorithm using a coupled atmosphere-vegetation radiation transfer model to produce sensorspecific simulated measurements and the associated surface FAPAR values.

qðki Þ þ ln2 Þ2 þ ln3 ð~ qðkj Þ þ ln4 Þ2 P ðki ; kj Þ ¼ ln1 ð~ ~ðki Þ~ þ ln5 q qðkj Þ; 2

2

qðki Þ þ ln7 Þ þ ln8 ð~ qðkj Þ þ ln9 Þ Qðki ; kj Þ ¼ ln6 ð~ ~ þ ln10 qðki Þ~ qðkj Þ þ ln11 .

ð4Þ

2198

N. Gobron et al. / Advances in Space Research 38 (2006) 2196–2202

The polynomial coefficients lnm have been optimized in such a way that the values generated by each spectral ~ðkj Þ correspond to the bi-direcpolynomial gn ½~ qðkblu Þ; q tional reflectance factors measured at the top of the canopy and normalized by the spectrally appropriate anisotropic reflectance function. In other words, the rectification process yields estimated values of spectral reflectances emerging at the top of the canopy, optimally decontaminated from atmospheric and angular radiative effects. The FAPAR itself is then computed on the basis of these rectified channel values, and its formula is: g0 ðqRred ; qRnir Þ ¼

l01 qRnir  l02 qRred  l03 2

2

ðl04  qRred Þ þ ðl05  qRnir Þ þ l06

; ð5Þ

where the coefficients l0m of polynomial g0 have been optimized a priori to force g0(qRred, qRnir) to take on values as close as possible to the FAPAR associated with the plant canopy scenarios used in the training data set. The numerical values of the various coefficients resulting from these successive optimizations are given in Gobron et al. (2000) for MERIS (Envisat), GLI (Adeos-II) and VEGETATION (SPOT 4); Gobron et al. (2001) for SeaWiFS (SeaStar), and Gobron et al. (2002) for MISR (Terra). These studies indicate that the estimation of FAPAR can be obtained with an average accuracy of ±0.05.

3. Global and regional land products from SeaWiFS data SeaWiFS was launched on the SeaStar spacecraft on August 1, 1997. Since mid-September, 1997, it has delivered multi-spectral BRF values collected over all regions of the globe. The continuous coverage, in both time and space, enable errors to be reduced to a minimum, giving increased confidence in interpretation and identification of ecological patterns that can be easily recognized and delineated. Following the development of the algorithm itself, a processing system using SeaWiFS data has been designed and used to produce two-kilometer resolution maps of FAPAR at the global scale. These maps take advantage of new methodology developed by Me´lin et al. (2002) for merging efficiently the Local Area Coverage and Global Area Coverage SeaWiFS data sources. The system includes a set of algorithms to: (1) classify each SeaWiFS pixel on the basis of multi-spectral BRF measurements into broad categories of geophysical targets such as clouds and bright objects, vegetated surfaces and water bodies and (2) compute the rectified red and near-infrared bands as well as the FAPAR value for those pixels corresponding to vegetated surfaces. The pixel classification is performed on the basis of an ensemble of thresholds using only the values in the

bands centered at 443, 670, and 865 nm. These tests were established on the basis of a priori knowledge of the multi-spectral signatures of each geophysical system. The approach efficiently assigns the vast majority of pixels to these classes without requiring any other ancillary information. A more sophisticated scheme was not deemed necessary or justified given the scientific objectives and computer processing constraints. Ensuring that the values of rectified bands must be within predefined intervals allows a further screening of undesirable geophysical conditions. In practice, for every available individual terrestrial SeaWiFS observation (pixel, date), the algorithm yields either a simple label or, in the case of vegetated surfaces, a string of values including all TOA Bidirectional Reflectance Factors (BRFs), the geometry of illumination and observation, the two rectified bands and the FAPAR value. For a number of surface applications, it is desirable to ensure a good geographical coverage, which implies the temporal compositing of time series over periods long enough to fill the gaps created by clouds. Such a procedure is justified as long as surface changes occur on a time scale much longer than the period adopted for the compositing. The procedure selects the most representative conditions during the compositing period on the basis of a simple statistical analysis (Pinty et al., 2002). This analysis, based on the inspection of the daily FAPAR values retrieved during each period of 10 consecutive days (or monthly period), is implemented as follows: The temporal average and corresponding deviation of the FAPAR values over the 10 day (monthly) periods are first estimated: T 1 X Fapar ¼ FaparðtÞ; T t¼1 ð6Þ T 1 X T DFapar ¼ jFaparðtÞ  Faparj; T t¼1 where T is the number of available clear sky values during the compositing period (10-day or monthly). Fapar is the temporal average index value and DTFapar is the average deviation of the distribution. The value selected as the most representative for the given ten day (monthly) period is the actual value which minimizes the quantity jFaparðtÞ  Faparj. This procedure thus generates maps of geophysical products for every 10day period (and monthly period) where each individual value represents the actual measurement or product for the day considered the most representative of that period. The geometry of illumination and observation for the particular day selected is saved as part of the final product, which is thus fully documented and traceable. The global products are available at either 2 km (nominal) resolution grids for three years but can also be remapped into a 10-km (nominal) grid map for modeling applications.

N. Gobron et al. / Advances in Space Research 38 (2006) 2196–2202

2199

Fig. 2. FAPAR maps in plate carre´e projection over Europe for May (top left), July (top right), September (bottom left) and November (bottom right) during year 1998.

As an example, Fig. 2 illustrates seasonal FAPAR maps in a plate carre´e projection over Europe during the year 1998. The colour scale indicates the value of FAPAR, such that white areas correspond to soils with little or no active vegetation cover, and red signals regions where the plants are particularly active. Between these two extremes, the various shades of green provide an indication of the relative photosynthetic activity of the biosphere. The first month considered here is May for which the FAPAR composite uses the days from 121 to 151. The deep red patches in Central France, much of England, Northern Germany and Slovenia/Croatia indicate very strong chlorophyll activity, therefore active growing. In the Alps, only the snow-covered mountaintops appear black (i.e., no valid estimate). During July (days 182 to 212), it can be seen that the harvest season has started in Central France; maximum growth is noticeable in much of the former Yugoslavia as well as on the northern slopes of the Pyre´ne´es. In contrast, the start of the dry season in Central Spain and the decrease in activity in the Baltic region can also be noted. On the other hand, rice growing has now peaked in the P^ o valley in Italy Vegetation activity decreases in Eastern Europe in September (days 244–273), and during November (days 305–334).

4. Results for various Carbo-Euro sites The annual time series of FAPAR can be used to assess photosynthetic activities over specific land surfaces. Figs. 3 and 4 illustrate the annual photosynthetic activity during three years for two Carbo-Euro sites which are both covered by forest vegetation. The profiles on the top (bottom) panel of each figure are drawn from the 10 day (monthly) products for three years (1998, 1999, and 2000) and represent the mean value of the 3 · 3 pixels around the central pixel defined by the siteÕs latitude and longitude. The superimposed error bars correspond to the average standard deviation, DTFapar within the composite period. The 3 · 3 pixels spatial resolution degradation is done to reduce possible problems of co-registration of the space remote sensing data. The dashed lines give the value of the mean spatial absolute deviation computed over these 9 pixels, i.e., DTFapar . These examples illustrate the consistency of the algorithm itself since the seasonal patterns of the FAPAR values are well reproduced during the three years both for the 10-day and monthly composite. In the case of the site at Castelporziano (Italy), the plots on Fig. 3 show a level around 0.4 with a sinusoidal shape with a maximum value in April and November for the three years The monthly composite time series is, as expected,

2200

N. Gobron et al. / Advances in Space Research 38 (2006) 2196–2202

Fig. 3. The 10-day (monthly) composite time series for Castelporziano are plotted on the top (bottom) panel. The red, green and blue line denotes the FAPAR annual series during 1998, 1999, and 2000. The superimposed error bars correspond to the values of the temporal mean absolute deviation, DTFapar over the time compositing period. The dashed lines illustrate the spatial mean absolute deviation, DSFapar around the central pixel.

ÔsmootherÕ than the 10-day composites. The spatial and temporal averaged mean absolute deviations are less than 0.1, which means that when comparing the annual time series, no dramatic change occurs over the three years. The site has Quercus Ilex as the dominant species, with evergreen shrubs understory (see http:// www.bgc-jena.mpg.de for more information about this site) The profiles in Fig. 4 correspond to a dominant vegetation species of Fagus sylvatica at Hainich (Germany). The seasonality profiles follow similar patterns and levels during the three years. The values increase

from April to May–June from a level of 0.2 to 0.9. The decreases of photosynthetic activity are similar for the three years until the minimal value reaches 0.2 in December. The differences in vegetation properties between these two sites are the stem density and canopy height, which are, respectively, 1500 (334) trees/ha and 10–15 (33) m for Castelporziano (Hainich) site. This diversity could explain the differences of the FAPAR profiles and further analysis should be done to better assess the relationships between FAPAR and ecological phenomenon.

N. Gobron et al. / Advances in Space Research 38 (2006) 2196–2202

2201

Fig. 4. Same as Fig. 1 over Hainich Site.

5. Conclusion This paper outlines the methodology for and shows applications of biophysical monitoring over land surfaces. The presence of live green vegetation over large diversity of terrestrial surfaces can be identified and monitored using this approach which is based on a two steps procedure: The first one aims at rectifying the red and near-infrared bands from the perturbing effects caused by the atmosphere and the changes in the relative geometry of illumination and observation. The second step consists of optimizing the formula to approximate a one-to-one relationship between the index value and the Fraction of Absorbed Photosynthetically Active Radiation, used as a proxy for detecting the

presence of healthy vegetation. The procedure capitalizes on the availability of advanced, coupled, surfaceatmosphere radiation transfer models and uses them to construct the training data set against which the optimization is achieved for various instruments. A methodology has been developed to merge efficiently LAC and GAC SeaWiFS data sources and to generate an ensemble of land products at a spatial resolution close to that of the sensor for global scale. These terrestrial products constitute an improvement in spatial resolution compared to existing products, and can be remapped in a 10-km grid for global change modeling purposes. The two-kilometer resolution maps over Europe and the global products are accessible from the first author.

2202

N. Gobron et al. / Advances in Space Research 38 (2006) 2196–2202

A quantitative evaluation of the seasonal information of FAPAR has been established through an analysis of actual profiles using SeaWiFS data. This application has shown the capability of the biophysical variable to link the information with two types of forest. It also illustrates the potential ability of the FAPAR to distinguish between their temporal dynamics. These geophysical products are well suited to address a number of issues related to the identification and monitoring of land surfaces. Acknowledgements The authors are grateful to the SeaWiFS Project (Code 970.2) and the Distributed Active Archive Center (Code 902) at the Goddard Space Flight Center, Greenbelt, MD 20771, for the production and distribution of the SeaWiFS data, respectively. The first author thanks the University of Strasbourg for providing the financial support to attend the 34th COSPAR symposium. References Gobron, N., Pinty, B., Verstraete, M.M., Widlowski, J.L., Diner, D. Uniqueness of multi-angular measurements. Part 2: joint retrieval of vegetation and photosynthetic activity from MISR. IEEE Transactions on Geoscience and Remote Sensing MISR Special Issue 40, 1574–1592, 2002.

Gobron, N., Me´lin, F., Pinty, B., Verstraete, M.M., Widlowski, J.L., Bucini, G. A global vegetation index for SeaWiFS: design and applications. in: Beniston, M., Verstraete, M.M. (Eds.), Satellite Remote Sensing Data and Climate Model Simulations: Synergies and Limitations. Kluwer Academic Publishers, Dordrecht, pp. 5– 21, 2001. Gobron, N., Pinty, B., Verstraete, M.M., Widlowski, J.L. Advanced spectral algorithm and new vegetation indices optimized for up coming sensors: development, accuracy and applications. IEEE Transactions on Geoscience and Remote Sensing , 38, , pp. 2489– 2505, 2000. Gobron, N., Pinty, B., Verstraete, M.M., Govaerts, Y. A semidiscrete model for the scattering of light by vegetation. Journal of Geophysical Research 102, 9431–9446, 1997. Me´lin, F., Steinich, C., Gobron, N., Pinty, B., Verstraete, M.M. Optimal merging of LAC and GAC data from SeaWiFS. International Journal of Remote Sensing 23, 801– 807, 2002. Pinty, B., Gobron, N., Me´lin, F., Verstraete, M.M. A time composite algorithm theoretical basis document. Institute for Environment and Sustainability, EUR Report No. 20150 EN, European Commission, Luxembourg, 8 p., 2002. Rahman, H., Pinty, B., Verstraete, M.M. Coupled surfaceatmosphere reflectance CSAR model. Part 2: semiempirical surface model usable with NOAA advanced very high resolution radiometer data. Journal of Geophysical Research 98, 20791–20801, 1993. Vermote, E., Tanre´, D., Deuze, J.L., Herman, M., Morcrette, J.J. Second simulation of the satellite signal in the solar spectrum: an overview. IEEE Transactions on Geoscience Remote Sensing 35 (3), 675–686, 1998. Verstraete, M.M., Pinty, B. Designing optimal spectral indices for remote sensing applications. IEEE Transactions on Geoscience and Remote Sensing 34, 1254–1265, 1996.