River Floodplain Vegetation Scenario Development Using Imaging ...

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1179

River Floodplain Vegetation Scenario Development Using Imaging Spectroscopy Derived Products as Input Variables in a Dynamic Vegetation Model M.E. Schaepman, G.W.W. Wamelink, H.F. van Dobben, M. Gloor, G. Schaepman-Strub, L. Kooistra, J.G.P.W. Clevers, A. Schmidt, and F. Berendse

Abstract

Introduction

River floodplains are becoming increasingly subject to multifunctional land-use. In this contribution, we are linking imaging spectrometer derived products with a dynamic vegetation model to improve the simulation and evaluation of scenarios for a river floodplain in the Netherlands. In particular, we are using airborne HyMap imaging spectrometer data to derive Leaf Area Index (LAI), spatial distribution of Plant Functional Types (PFT), and model dominant species abundances as input for the ecological model. We use the dynamic vegetation model (DVM) SMART2-SUMO to simulate vegetation succession under scenarios of changing abiotic conditions and management regimes. SMART2 is a soil chemical model whereas SUMO describes plant competition and resulting vegetation succession. We validate all remote sensing derived products and the DVM calibration independently using extensive field sampling. We conclude that the dynamic vegetation models can be successfully initialized using imaging spectrometer data at currently unprecedented accuracy. However, all efforts undertaken for validation in this contribution may significantly exceed capacities for national or continental scale application of the proposed method.

River floodplains are becoming increasingly subject to multifunctional land-use. In hydraulics, floodplains are assessed for the suitability of flood control strategies in relation to their roughness for reducing and delaying flood peaks (Ghavasieh et al., 2006). In climate change, floodplains are under investigation due to changing discharge regimes, potentially resulting in an increase in peak and low flows (Middelkoop et al., 2001), as well as increased pressure on hard infrastructures protecting large fractions of a growing population living in the affected areas (Kabat et al., 2005). However, a particular challenge remains that river floodplain systems are among the most complex ecosystems on Earth. The lack of detailed information about functional relationships and processes at the landscape and catchments scale currently hamper assessment of their ecological status (Jungwirth et al., 2002). Regional-based investigations point to complex interactions, affecting soils (Kooistra et al., 2004), vegetation (Clevers et al., 2004), as well as animal life (Kooistra et al., 2005) in floodplains, but may not be limited to these. An increasing number of studies point to the fact, that medium range projections of land-use in general (Feddema et al., 2005), and river floodplain multifunctional use in particular will be primarily driven by anthropogenic impact, affecting insect (Nickel and Hildebrandt, 2003) and fish population (Peterson and Kwak, 1999). This human influence may also occasionally override the expected medium range impact of climate induced changes (Schaepman et al., 2005) on floodplains (e.g., extreme events and rising temperatures). The goal of this contribution is to combine remote sensing-derived products with dynamic vegetation modeling (DVM) to improve the simulation and evaluation of future scenarios for a river floodplain. In order to do so, we link imaging spectrometer-derived products as variables in a DVM to assess the development of a river floodplain that has been taken out of agricultural production and is allowed to undergo a natural succession.

M.E. Schaepman, L. Kooistra, and J.G.P.W. Clevers are with the Wageningen University, Centre for Geo-Information, Droevendaalsesteeg 3, NL-6708 PB Wageningen, The Netherlands, ([email protected]). G.W.W. Wamelink and H.F. van Dobben are with Alterra, Landscape Centre, Wageningen, The Netherlands. M. Gloor is with the Earth and Biosphere Institute, School of Geography, Leeds Univ., UK. G. Schaepman-Strub is with the Wageningen University, Nature Conservation and Plant Ecology, Wageningen, The Netherlands, and KNMI, Atmospheric Research Division, DeBilt, The Netherlands. F. Berendse is with the Wageningen University, Nature Conservation and Plant Ecology, Wageningen, The Netherlands. A. Schmidt is with Alterra, Centre for Geo-Information, Wageningen, The Netherlands. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

Photogrammetric Engineering & Remote Sensing Vol. 73, No. 10, October 2007, pp. 1179–1188. 0099-1112/07/7310–1179/$3.00/0 © 2007 American Society for Photogrammetry and Remote Sensing O c t o b e r 2 0 0 7 1179

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1180

We compare therefore the impact of two different management scenarios on biomass production in the floodplain for the year 2050.

Materials and Methods Test Site We focus on a large-scale nature development area between the cities of Arnhem (NL), Nijmegen (NL) and Emmerich (GER), named the “Gelderse Poort” (Figure 1). This border-crossing natural reserve is located along the river Rhine where it splits into three branches, namely the Waal, the Nederrijn and the IJssel. The floodplain Millingerwaard (51.84° N; 05.99° E; WGS84; 700 ha) is part of the Gelderse Poort. The site rises 12 m ASL with a minimum of 8.8 m ASL and a maximum of 15.6 m ASL. Before the 1990s, the main function of the floodplain was agriculture, namely cultivated grassland and arable land (e.g., maize). In the period 1990 to 1993, the agricultural function was gradually changed into a combined nature conservation and flood protection function. Since then, the floodplain is allowed to undergo natural vegetation succession. Nature management measures are limited to removing the fences between former agricultural parcels and grazing by cattle and horses in low densities. This converted the Millingerwaard into a heterogeneous landscape with river dunes stretching along the river, a large softwood forest in the eastern part along the winter dike, and in the intermediate area a pattern of different vegetation succession stages (pioneer, grassland, and shrubs). In addition, several clay pits are present. Nature management (grazing) within the floodplain is aiming at increasing the biodiversity (namely the diversity of species and ecosystems), given the condition that the discharge capacity of the river should be above the critical safety levels during flooding events. To stimulate the development of a heterogeneous landscape (see Figure 2), a low stocking density of one animal (e.g., Galloway, Koniks) per 2 to 4 ha has been chosen. This stocking density allows grazing all year round and allows the development of forested areas. Data Acquisition on the Ground Simultaneous to the acquisition of airborne imaging spectrometer data, extensive ground measurements were carried out. These included various sampling schemes and measurement

Figure 1. Location and current land use for the floodplain Millingerwaard along the river Rhine in the Netherlands.

1180 O c t o b e r 2 0 0 7

Figure 2. Oblique photograph of the Millingerwaard natural reserve towards the East. Vegetation succession stages are ranging from grasslands (foreground) to shrubs and forests (background).

techniques, to comply with the requirements to satisfy relevant input variables for data preprocessing, radiative transfer modeling (RTM), as well as DVM. Table 1 summarizes the measurements performed for this study. We performed sun-photometer measurements using a Solar Microtops II instrument to characterize the atmospheric conditions during the HyMap airborne data acquisition. The setup of the instrument included the use of five collimators, each with a field of view (FOV) of 2.5 degrees (sun looking), covering the wavelengths 440 nm, 675 nm, 870 nm, 936 nm, and 1,020 nm. Aerosol optical thickness (AOT) at 440, 675, 870, and 1,020 nm, respectively, is retrieved using the Bouguer-Lambert-Beer law: V(l) Vo(l)D2eM(l)

(1)

where center wavelength, V() measured detector voltage at , Vo() extraterrestrial voltage at , D Earthsun distance, () total optical thickness, and M air mass. AOT is obtained after subtraction of the optical depth due to Rayleigh scattering from the total optical depth. AOT is subsequently used in the atmospheric correction of the HyMap data. Figure 3 depicts the locations of ground measurements performed. The sun-photometer was located in the center of the floodplain. Nineteen reference sites consisting mainly of natural and man-made materials (e.g., roads, artificial clay pit, sandy beach) have been selected and measured using an Analytical Spectral Devices FieldSpec® Pro FR spectroradiometer (ASD) for vicarious calibration purposes as well as validation of the atmospheric correction. ASD data were spectrally convolved to HyMap spectral response functions for direct comparison. Leaf and canopy reflectance spectra were measured using a second ASD incorporating a high intensity contact probe with a leaf clip for non-destructive reflectance and transmittance measurements of in vivo leaves. Leaf spectra ad-axial and ab-axial of dominating species (e.g., Calamagrostis epigejos, Rubus caesius, and Urtica dioica) were measured using the contact probe. We have further chosen to measure canopy reflectance at 21 vegetation relevés (2 m 2 m each). The location of the 21 sites were derived based on an existing vegetation map created in 2002 for the PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1181

TABLE 1.

OVERVIEW

OF

GROUND

AND

AIRBORNE DATA ACQUISITIONS No. Sample Locations

Instrument Atmospheric condition Vicarious calibration

Solar Microtops sunphotometer ASD Fieldspec Pro FR

Leaf and canopy reflectance spectra Canopy structure

ASD Fieldspec Pro FR Hemispherical camera Braun-Blanquet method Laboratory analysis

Vegetation relevés Destructive biomass sampling + chemical analysis Soil characteristics Imaging spectrometer data

Theta probe, temperature gun HyMap

1

MILLINGERWAARD

Date (2004) 02 August

19 (5 5 m)

28 July and 02 August

21 (5 5 m)

28 July

13 (20 20 m), VALERI scheme 21 (2 2 m)

28 July to 06 August 13–16 August

21 (0.5 0.5 m)

13–16 August

86 2 flight-lines

Figure 3. Geocoded HyMap image indicating sample locations of ground based measurements for radiometric correction and calibration, as well as determining leaf and canopy spectra, canopy structure, soil, and water within the Millingerwaard floodplain.

Millingerwaard (Van Geloof and de Ronde, 2002), complemented by a field survey in May 2004 to adjust for potential changes. In addition, the vegetation relevés served for different purposes, such as performing a detailed vegetation description, visual estimates of the fractional coverage, as well as destructive biomass sampling following all optical measurements. The vegetation description (Schmidt et al., 2005) was performed following the method of Braun-Blanquet (1951). Abundance per species was estimated optically as percentage soil covered by living biomass in vertical projection, and scored in a nine-point scale. The vegetation relevés covered PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

IN THE

28 July 28 July and 02 August

Measurements (Aerosol optical thickness) Reflectance spectra (sand, clay, asphalt, water) TOC (Top-of-canopy) and leaf spectra LAI, gap fraction, leaf angle distribution Vegetation structure, species composition Wet/dry biomass, N and P concentration AOT

Soil moisture and temperature 126 spectral bands, 5 m spatial resolution, 512 pixels across track, acquisition time 11 38 UTC (13 38 MEST)

the most important plant communities as described by Schaminée et al. (1998) present in the area. All bryophytes and lichens, and vascular species that were not readily recognizable in the field, were collected for later identification. Taraxacum species were taken together as T. vulgare, and Rubus species were taken together as R. fruticosus, except R. caesius. No subspecific taxa were used. Nomenclature follows van der Meijden et al. (1990), Touw and Rubers (1989), and Brand et al. (1988) for vascular species, mosses and lichens, respectively. Syntaxonomic nomenclature follows Schaminée et al. (1998). Following optical measurements at the vegetation relevés, destructive aboveground biomass sampling in three 50 cm 50 cm subplots of each of the 21 sample sites was performed. Vegetation biomass was sampled in a relatively homogeneous (vegetation) cover, located at three of the corners of each main plot. Biomass was clipped at 0.5 cm above the ground level and stored in plastic bags. The collected material was air-dried, first for five days at room temperature in open bags, and subsequently dried for 24 hours at 70°C, and weighed. Sampled vegetation material for the 21 vegetation plots was also chemically analyzed for N, P, K, Ca, and Mg content (mmol/kg). The forested part of the Millingwaard is dominated by willow trees, having dominant species of Salix fragilis L. (crack willow), Salix alba L. (white willow), and Populus nigra L. (Lombardy poplar). A dense undergrowth is present with Urtica dioica L. (common nettle), Calamagrostis epigejos (L.) Roth (wood small-reed), and Rubus caesius L. (European dewberry) being the dominant plant species. Since canopy reflectance measurements in the forest were not possible with the ASD due to presence of dense undergrowth and water bodies, a hemispherical camera was used to estimate the gap fraction, leaf inclination angle, and leaf area index (LAI). In the forest, thirteen sample plots were selected following the VALERI sampling scheme (Baret et al., in review). Stratified random sampling was used to position the sample plots in various softwood canopy densities. The coordinates of the sample plots were registered using GPS at each center point. In total, 156 points (i.e., 13 elementary sampling units (ESU’s) according the VALERI sampling scheme with each ESU having 12 subsampling points) were measured (Mengesha et al., 2005b). O c t o b e r 2 0 0 7 1181

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1182

Airborne Data Acquisition Airborne imaging spectrometer data (HyMap (Cocks et al., 1998)) were acquired on 28 July and 02 August 2004 in 126 spectral bands ranging from 400 to 2,500 nm (spectral bandwidth of 15 to 20 nm) over the Millingerwaard. The data is processed to surface reflectance or Hemispherical-Directional Reflectance Factor (HDRF) following the terminology of Schaepman-Strub et al. (2006). Preprocessing is partially compensating for adjacency effects as well as directional effects induced by the atmosphere using the model combination PARGE/ATCOR-4 (Richter and Schlapfer, 2002; Schlapfer and Richter, 2002). However, there is no particular treatment of the surface induced anisotropy in this approach, resulting in the surface reflectance data to approximate HDRF. Using the same correction scheme, HyMap data was also geocoded and remapped to UTM (Zone 31 N, geodetic datum WGS84) at an equally spaced ground sampling distance of 5 m in both axes. The flight was performed close to the local solar noon (11 38 hrs coordinated universal time (UTC), 13 38 hrs Middle European Summer Time (MEST)) at a solar zenith angle of 33° and solar azimuth angle of 178°. SMART2-SUMO Model The model SMART2-SUMO (Kros 2002, Wamelink et al., 2005; Wamelink et al., 2003) is used to simulate vegetation succession under scenarios of changing abiotic conditions and management regimes. SMART2 is a soil chemical model that describes chemical processes like weathering, adsorption, desorption, mineralization, and immobilization. Fertilization and atmospheric deposition of nitrogen compounds are also accounted for. SUMO describes plant competition and resulting vegetation succession. In SUMO for a vegetation structure type (e.g. grassland or forest), the biomass for five functional types is simulated: grassland and herbs, dwarf shrubs, shrubs, pioneer tree, and climax tree. All five functional types are always present, though the amount of biomass simulated for each functional type may vary enormously. SMART2-SUMO contains a full description of the nutrient cycle through root uptake, investment in biomass (divided over root, shoot and leaf), litter fall and nitrogen mineralization. Vegetation management (e.g., mowing and grazing) is described as the removal of part of the biomass at the end of the growing season. SMART2-SUMO is a point model, which does not describe spatial (horizontal) interaction. Therefore, the model can be applied on various spatial scales, provided the necessary input data are available. The SUMO-SMART2 model is calibrated using field data (on biomass, soil pH, etc.). In order to run the SUMO-SMART2 model at a spatially explicit scale, remote sensing-derived products serving as input variables for the DVM are needed. We use imaging spectroscopy derived biomass for the initialization of the model. We first use this data to validate and then to initialize the SMART2-SUMO model for two different model runs. We predict the vegetation succession of the river floodplain by assuming different management scenarios, resulting in estimating biomass production divergence in the year 2050. The two management scenarios that were used are: (a) agricultural management, where the grassland is mown once a year and 100 kg/ha nitrogen is supplied every year, and (b) extensive nature management, where the grassland is grazed by cows and horses at a density of approximately one grazing unit per hectare. Imaging Spectroscopy Derived Products as Input Variables for Dynamic Vegetation Models The spatially explicit biomass input variable for the SMART2-SUMO model is based on imaging spectrometer data. Biomass derived from HyMap data is based first on a sensitivity analysis comparing several LAI estimation methods 1182 O c t o b e r 2 0 0 7

using the HyMap surface reflectance data in combination with four retrieval methods (Mengesha et al., 2005a) (Plate 2). These methods have been refined to account for the significant shadow fraction present in high-resolution airborne (imaging spectrometer) data as well as for spatial heterogeneity of the species composition. Estimates of LAI were retrieved most successful in the forest using the method proposed by Chen et al. (2002), including a comparison to the methods proposed by Roujean and Lacaze (2002) and Weiss et al. (2002). Validation was performed by using a simple kriging approach to interpolate within the ESUs (Baret and Rossello, 2006) in the softwood forest stand and comparing the aggregated HyMap pixels (5 m to 20 m) with this interpolation. For the estimates of biomass of non-forested areas (grasses and herbs, dwarf shrubs, and shrubs), a forward spectral linear mixture modeling of modeled and measured leaf spectra has been applied to approximate dominant plant species composition in the Millingerwaard. Leaf spectra were either measured in the field using the ASD contact probe, or modeled using the leaf model PROSPECT (Jacquemoud and Baret, 1990). Missing input parameters for PROSPECT of dominant plant species not measured in vivo were extracted from literature (Jacquemoud, 2006). The combined radiative transfer model PROSPECT/SAIL (Verhoef and Bach, 2003) was used to create homogenous, single plant species TOC reflectance. Forward linear mixture modeling based on dominant plant species per plant community (Schaminée et al., 1998) was used to create a spectral library at plant community level. Finally, spectral unmixing (Hu et al., 1999; Keshava and Mustard, 2002) was used on the HyMap data to unmix for abundances of shadow, plant functional types, and other relevant endmembers (e.g., water, gravel, clay, sand, artificial/man-made materials). The LAI product generated using the above approach was directly compared to the destructive biomass sampled in the 21 plots. The initialization of SMART2-SUMO assumed standard biomass values for countrywide applications for both agricultural and natural grasslands in 1975. We implemented two management scenarios to evaluate the effect of management measures on vegetation succession (biomass production): an agricultural management scenario, including yearly mowing, and a natural succession scenario under extensive grazing. We also assumed the hydrology to be constant over time. The soil type (fraction of sand or clay) was derived from spectral unmixing of ASD measurements and using field observations per plot. The stocking density of cattle and horses in the natural scenario was also estimated from field observations in the plots. Model validation was performed by comparing the simulated biomass in 2004 in the nature management scenario with the actual measured biomass in the field. Next, the simulated biomass was improved by replacing the simulated biomass of the grasses and herbs functional type in 2004 by the biomass estimated from the HyMap-derived LAI, producing a forecast until 2050.

Results HyMap Processing HyMap data were evaluated for geometric and radiometric quality. Since most of the ground data acquired in the field was registered in the Dutch reference system (Rijksdriehoekstelsel (RD)), a co-location procedure was established to re-project the HyMap data (WGS84, UTM zone 31N coordinate system) into RD, and vice versa. The initial co-registration uncertainty was found to be 1 to 2 pixels (5 to 10 m). Since PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1183

Figure 4. HyMap minimum, maximum, mean and standard deviation signal for 28 July 2004.

the vegetation maps have been established, using DGPS-based methods, HyMap data was re-projected to the ground data using a polynomial transformation, eventually reducing the uncertainty to 1 m. HyMap data did not suffer from saturation or excess noise (Figure 4), and a principal component based approach was performed including the use of Eigenvalues distribution to determine the dimensionality of the data. The dimensionality was considered to be very high (10 at half the sum of all Eigenvalues, 60 at a quarter of the maximum) for HyMap, with a few (4 to 6) noisy bands to be expected. Surface reflectance data were retrieved by using the sunphotometer derived values for aerosol transmittance and

TABLE 2.

CLASSIFIED VEGETATION SYNTAX

horizontal visibility (v 15.5 km) as well as geometrical illumination and observation conditions. The ATCOR inherent iterative validation scheme (IFCALI) was used to determine the quality of the atmospheric correction on the vicarious calibration measurements (see Schaepman et al. (2004)). Vegetation Classification In total, 79 plant species were registered in the 21 vegetation relevés made in the Millingerwaard. The relevés were syntaxonomically identified into plant communities (Schaminée et al., 1998) by using the program ASSOCIA (Table 2; Figure 5: see Wamelink et al., 2003). The chemical analysis of biomass samples showed that the N-content was significantly

FOR THE INDIVIDUAL

PLOTS

IN THE

MILLINGERWAARD

RD-coordinates

Plot Number

X

Y

Syntax Code

02 03 04 05 06 07 08 09 10

196826,397 196814,358 196795,228 196817,681 196863,64 196905,185 196900,856 196852,201 196768,893

431032,034 431034,655 431076,810 431082,001 431170,789 431185,534 431209,847 431313,554 431466,838

16BC01 31CA01B 31CA01B 31A 33RG03 29AA03C 31CA01B 31CA01B 38AA01B

11 12 13 14 15 16 17 18 19 20 21

196791,704 196764,162 196712,193 196640,748 196643,353 196504,145 196512,661 196485,196 196559,239 196587,126 196528,714

431480,965 431620,923 431648,921 431448,257 431367,244 431230,224 431189,218 431115,084 431122,375 431140,347 430948,963

37AB01A 37AC02A 31CA 31CA02 31CA02 31AA01 31CA02 14CA 31CA02 31CA02 31RG08

22

196360,958

430753,776

31A

PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

Syntax Name Lolio-Cynosuretum Echio-Melilotetum typicum Echio-Melilotetum typicum Chenopodio-Urticetalia RG Petasites hybridus-[Galio-Urticetea] Chenopodietum rubri rorippetosum Echio-Melilotetum typicum Echio-Melilotetum typicum Artemisio-Salicetum agrostietosum stoloniferae Pruno-Crataegetum typicum Hippophao-Ligustretum typicum Dauco-Melilotion Bromo inermis-Eryngietum campestris Bromo inermis-Eryngietum campestris Bromo-Corispermetum Bromo inermis-Eryngietum campestris Tortulo-Koelerion Bromo inermis-Eryngietum campestris Bromo inermis-Eryngietum campestris RG Cichorium intybus-[Agropyretalia repentis/Arrhenatheretalia] Chenopodio-Urticetalia

O c t o b e r 2 0 0 7 1183

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1184

Figure 5. Detrended Correspondance Analysis (DCA) of the 21 vegetation samples in the Millingerwaard using the CANOCO software (Ter Braak and Smilauer, 2002). Number of species: 79, Eigenvalues: 1 0.64, 2 0.58, 4.78; the plot therefore represents 26 percent of the variance in the species data. Detrending performed by using second order polynomials. The species plot (a) shows the position of the 58 species whose weight is 5 percent of the maximum weight; after overlaying the species plot with the sample plot (b), the Euclidian distance of a species to a sample is an inverse measure for the probability to find a species in a plot. An explanation of the abbreviated species names is in Appendix A.

Plate 1. Canopy reflectance spectra measured for the 21 vegetation plots in the Millingerwaard.

(P 0.001) higher in the grazed plots (2.0 percent) than in the un-grazed plots (1.3 percent). For other nutrients (P, K, Mg), these differences were not significant. Corresponding TOC reflectance measurements of the plots are listed in Plate 1. 1184 O c t o b e r 2 0 0 7

Ground and Airborne-derived LAI Ground based hemispherical photographs of forest communities were processed using the CAN_EYE® software (Baret and Weiss, 2006), and resulted in an estimation of both, effective and true LAI. LAI values were ranging from 4.7 to PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1185

Plate 2. Spatially distributed LAI of the Millingerwaard based on four different retrieval schemes: (a) Weighted Difference Vegetation Index (WDVI), (b) Reduce Simple Regression (RSR), (c) Vegetation Continuous Fields (FVC), and (d) Normalized Difference Vegetation Index (NDVI). LAI units are in m2/m2.

6.5 m2/m2 and 2.9 to 4.0 m2/m2 for true and effective LAI respectively (see Plate 2). The simple kriging approach comparing interpolated in situ LAI with HyMap-derived RSR LAI proved to be of very good quality (r2 0.88). Linking Remote Sensing and the Dynamic Vegetation Model There appears to be a satisfying agreement between the measured by clipping and weighing and modeled aboveground biomass, although at very low biomass, the simulated values are sometimes significantly lower than the actually measured ones (Figure 6). This may be due to an over-estimation of the stocking intensity (note that the plots with low biomass values are the grazed ones). The relationship is improved, when relating the HyMapderived LAI and the biomass (r2 0.61) while excluding the grazed plots. Next, we attempted to improve the biomass simulated by SUMO for 2050 by re-initializing SUMO in 2004 using the PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

HyMap derived biomass. Plate 3 shows the result for two scenarios: continuation of the agricultural management and natural succession; the latter with and without re-initialization in 2004. Figure 7 shows that grazing also influences the relation between reflectance and biomass; the grazed plots appear to have a much higher LAI at a given biomass compared to the un-grazed ones. Subsequently, these plots were excluded from our analysis because their low number prevented a separate calibration. An explanation for the high LAI of the grazed plots might be their N-content; a chemical analysis of plant material showed that its N-content was significantly higher (P 0.001) in the grazed plots (2.0 percent) than in the un-grazed plots (1.3 percent). For other nutrients (P, K, Mg) theses differences were not significant. Apparently management (e.g. grazing, mowing) will also have to be taken into account if biomass is to be estimated from reflectance. O c t o b e r 2 0 0 7 1185

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1186

Figure 6. Measured versus modeled aboveground biomass. Modeled biomass values taken for 2004 from the SMART2-SUMO run for natural succession; measured values were determined by clipping and weighing.

Figure 7. Measured biomass for the plots versus HyMapderived LAI. The regression line pertains to the un-grazed plots only.

Plate 3. Total Biomass simulated by SUMO for two plant functional types, under a scenario of fertilization and mowing (agricultural management) or grazing (nature management), the latter both with (dotted) and without (drawn) re-initialization in 2004 (“re-init”). The left y axis denotes the biomass of the herbs and grasses in tons per hectare, the right vertical axis plots that of all other plant functional types (woody species, also in tons/ha). The starting point of the dotted line represents the re-initialized value. The oscillations in the first years after (re-) initialization are due to model instability.

Conclusions Our results demonstrate that imaging spectrometer-derived products can be used for the initialization of the dynamic vegetation model SMART2-SUMO at a regional scale. In particular the preprocessing using all relevant information available increased the reliability of the input data, and therefore the final result. The separation of general PFT into the functional types herbs and grasses, dwarf shrubs, shrubs, and forest (trees) proved to have a major effect on the 1186 O c t o b e r 2 0 0 7

simulations for the development of the vegetation. However, since imaging spectrometer data are not yet available on an extended, multi-temporal basis, the performance of the model in predicting the temporal development can only be judged by expert knowledge. After re-initialization, the increase in grass and herb biomass over time decreased, while the woody biomass increases more rapidly. This seems to agree better with the actual vegetation succession in the area, where scrub cover rapidly increases at the expense of grass cover. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1187

An extensive regional-based study was carried out to link imaging spectrometer products to a dynamic vegetation model. Due to the regional flavor of the work performed, its scalability to national or even continental scales needs to be investigated. A refinement of the regional-based approach will include in the future more accurate vegetation mapping using advanced spectral libraries and directional leaf optical properties measurements. A PFT-based unmixing approach is proposed, weighting the abundances per pixel, allowing for multiple PFTs present in one pixel (in particular the mixture of shrub and grassland was identified to be a challenge). The data driven estimate of the LAI-biomass relation should be replaced by a quantitative, physical based approach (e.g., inversion of a radiative transfer based approach), as well as estimating NPP-based on a model including the use of local climatology. The detailed assessment of PFTs may also serve to estimate their spatial frequency distribution helping to estimate the ecosystem’s roughness for reducing and delaying flood peaks.

Acknowledgments The airborne data acquisition was performed in the frame of the STEREO-I HyEco project. The support of the Belgian Space Office (Belspo) and VITO (B) is acknowledged. The contribution of G. Schaepman-Strub is supported by an European Space Agency (ESA) external fellowship. We thank the reviewers for their substantial comments.

References Baret, F., and P. Rossello, 2006. VALERI – VAlidation of Land European Remote sensing Instruments, INRA, URL: http://www.avignon.inra.fr/valeri/ (last date accessed: 27 June 2007). Baret, F., and M. Weiss, 2006. CAN_EYE – Processing digital photographs for canopy structure characterization, INRA, URL: http://www.avignon.inra.fr/can_eye/ (last date accessed: 27 June 2007). Baret, F., M. Weiss, S. Garrigue, D. Allard, J.P. Guinot, M. Leroy, H. Jeanjean, H. Bohbot, R. Bosseno, G. Dedieu, C. Di Bella, M. Espana, V. Gond, X.F. Gu, D. Guyon, C. Lelong, E. Mougin, T. Nilson, F. Veroustraete, and R. Vintilla, In review, VALERI: A network of sites and a methodology for the validation of medium spatial resolution land satellite products, Remote Sensing of Environment. Brand, A.M., A. Aptroot, A.J. De Bakker, and H.F. Van Dobben, 1988. Standaardlijst van de Nederlandse korstmossen, Wet Med KNNV, Utrecht. Braun-Blanquet, J., 1951. Pflanzensoziologie, Grundzüge der Vegetationskunde, Springer, Wien. Chen, J.M., G. Pavlic, L. Brown, J. Cihlar, S.G. Leblanc, H.P. White, R.J. Hall, D.R. Peddle, D.J. King, J.A. Trofymow, E. Swift, J. Van der Sanden, and P.K.E. Pellikka, 2002. Derivation and validation of Canada-wide coarse-resolution leaf area index maps using high-resolution satellite imagery and ground measurements, Remote Sensing of Environment, 80:165–184. Clevers, J.G.P.W., L. Kooistra, and E.A.L. Salas, 2004. Study of heavy metal contamination in river floodplains using the rededge position in spectroscopic data, International Journal of Remote Sensing, 25:3883–3895. Cocks, T., R. Jenssen, A. Steward, I. Wilson, and T. Shields, 1998. The HyMap airborne hyperspectral sensor: The system, calibration and performance, Proceedings of the 1st EARSeL Workshop on Imaging Spectroscopy (M.E. Schaepman, D. Schlapfer and K.I. Itten, editors), EARSeL, Zurich (CH), pp. 37–42. Feddema, J.J., K.W. Oleson, G.B. Bonan, L.O. Mearns, L.E. Buja, G.A. Meehl, and W.M. Washington, 2005. Atmospheric science: The importance of land-cover change in simulating future climates, Science, 310:1674–1678.


Ghavasieh, A.R., C. Poulard, and A. Paquier, 2006. Effect of roughened strips on flood propagation: Assessment on representative virtual cases and validation, Journal of Hydrology, 318:121–137. Hu, Y.H., H.B. Lee, and F.L. Scarpace, 1999. Optimal linear spectral unmixing, IEEE Transactions on Geoscience and Remote Sensing, 37:639–645. Jacquemoud, S., 2006. OPTICLEAF – The Database on Leaf Optical Properties, University of Paris 7, URL: http://teledetection.ipgp. jussieu.fr/opticleaf/ (last date accessed: 27 June 2007). Jacquemoud, S., and F. Baret, 1990. Prospect – A model of leaf opticalproperties spectra, Remote Sensing of Environment, 34:75–91. Jungwirth, M., S. Muhar, and S. Schmutz, 2002. Re-establishing and assessing ecological integrity in riverine landscapes, Freshwater Biology, 47:867–887. Kabat, P., W. Van Vierssen, J. Veraart, P. Vellinga, and J. Aerts, 2005. Climate proofing the Netherlands, Nature, 438:283–284. Keshava, N., and J.F. Mustard, 2002. Spectral unmixing, IEEE Signal Processing Magazine, 19:44–57. Kooistra, L., M.A.J. Huijbregts, A.M.J. Ragas, R.S.E.W. Leuven, and R. Wehrens, 2005. Spatial variability and uncertainty ecological risk assessment: A case study on the potential risk of cadmium for the little owl in a Dutch river flood plain, Environmental Science and Technology, 39:2177–2187. Kooistra, L., R.S.E.W. Leuven, P.H. Nienhuis, R. Wehrens, L.M.C. Buydens, E.A.L. Salas, and J.G.P.W. Clevers, 2004. Exploring field vegetation reflectance as an indicator of soil contamination in river floodplains, Environmental Pollution, 127:281–290. Kros, J., 2002. Evaluation of Biogeochemical Models at Local and Regional Scale, Ph.D. Thesis, Alterra Scientific Contributions 7, Alterra, Wageningen. Mengesha, T., L. Kooistra, R. Zurita-Milla, S. De Bruin, and M. Schaepman, 2005a. Methodology comparison of quantitative LAI retrieval using imaging spectroscopy and geo-spatial interpolation in a softwood forest, Proceedings of the 4th Workshop on Imaging Spectroscopy (B. Zagajewski, M. Sobczak, and W. Prochnicki, editors), EARSeL, Warsaw, Vol. 1, pp. 141. Mengesha, T., M.E. Schaepman, S. De Bruin, R. Zurita-Milla, and L. Kooistra, 2005b. Ground validation of biophysical products using imaging spectroscopy in softwood forests, Proceedings of the VALERI Workshop (F. Baret and M. Weiss, editors), INRA, Avignon, France, unpaginated CD-ROM. Middelkoop, H., K. Daamen, D. Gellens, W. Grabs, J.C.J. Kwadijk, H. Lang, B.W.A.H. Parmet, B. Schädler, J. Schulla, and K. Wilke, 2001. Impact of climate change on hydrological regimes and water resources management in the Rhine basin, Climatic Change, 49:105–128. Nickel, H., and J. Hildebrandt, 2003. Auchenorrhyncha communities as indicators of disturbance in grasslands (Insecta Hemiptera) – A case study from the Elbe flood plains (northern Germany), Agriculture, Ecosystems and Environment, 98:183–199. Peterson, J.T., and T.J. Kwak, 1999. Modeling the effects of land use and climate change on riverine smallmouth bass, Ecological Applications, 9:1391–1404. Richter, R., and D. Schlapfer, 2002. Geo-atmospheric processing of airborne imaging spectrometry data, Part 2: Atmospheric/topographic correction, International Journal of Remote Sensing, 23:2631–2649. Roujean, J.L., and R. Lacaze, 2002. Global mapping of vegetation parameters from POLDER multiangular measurements for studies of surface-atmosphere interactions: A pragmatic method and its validation, Journal of Geophysical Research D: Atmospheres, 107(D12):1261–1282. Schaepman-Strub, G., M. Schaepman, T. Painter, S. Dangel, and J. Martonchik, 2006. Reflectance quantities in optical remote sensing – Definitions and case studies, Remote Sensing of Environment, 103:27–42. Schaepman, M., R. Zurita-Milla, M. Kneubühler, J.G.P.W. Clevers, and S. Delwart, 2004. Assessment of long-term vicarious calibration efforts of MERIS on land product quality, Sensors, Systems, and Next-Generation Satellites VIII (R. Meynart, S.P. Neeck, and H. Shimoda, editors), SPIE, Maspalomas, Vol. 5570, pp. 363–371.

O c t o b e r 2 0 0 7 1187

PMSRS-07.qxd

9/14/07

11:22 PM

Page 1188

Schaepman, M.E., G.W.W. Wamelink, H. van Dobben, M. Gloor, G. Schaepman-Strub, L. Kooistra, A. Schmidt, and F. Berendse, 2005. Regional scale ecosystem modeling for vegetation scenario development – Demonstrated in a river floodplain (The Netherlands) using imaging spectroscopy, Proceedings of the 9th International Symposium on Physical Measurements and Signatures in Remote Sensing (ISPMSRS) (S. Liang, J. Liu, X. Li, R. Liu, and M.E. Schaepman, editors), ISPRS, Beijing, Vol. XXXVI, pp. 667–670. Schaminée, J.H.J., E.J. Weeda, and V. Westhoff, 1998. Plantengemeenschappen van de kust en van binnenlandse pioniermilieus, Opulus Press, Uppsala – Leiden, 346 p. Schlapfer, D., and R. Richter, 2002. Geo-atmospheric processing of airborne imaging spectrometry data, Part 1: Parametric orthorectification, International Journal of Remote Sensing, 23:2609–2630. Schmidt, A., H. van Dobben, G.W.W. Wamelink, L. Kooistra, and M.E. Schaepman, 2005. HYECO’04: Using hyperspectral reflectance data to initialize ecological models, Imaging Spectroscopy – New Quality in Environmental Studies (B. Zagajewski, and M. Sobczak, editors), EARSeL, Warsaw, Vol. 1, pp. 247–254. Ter Braak, C.J.F., and P. Smilauer, 2002. CANOCO Reference Manual and Canodraw for Windows User’s Guide: Software for Canonical Community Ordination, (Version 4.5), Microcomputer Power, Ithaca, New York. Touw, A., and W.V. Rubers, 1989. De Nederlandse bladmossen – Flora en verspreidingsatlas van de Nederlandse Musci (Sphagnum uitgezonderd), KNNV Uitgeverij, Utrecht. van der Meijden, R., E.J. Weeda, W.J. Holverda, and P.H. Hovenkamp, 1990. Heukel’s flora van Nederland, Wolters-Noordhoff, Groningen. Van Geloof, I., and I. de Ronde, 2002. De vegetatie van de Millingerwaard na 10 jaar “natuurontwikkeling,” Wageningen University, Wageningen. Verhoef, W., and H. Bach, 2003. Simulation of hyperspectral and directional radiance images using coupled biophysical and atmospheric radiative transfer models, Remote Sensing of Environment, 87:23–41. Wamelink, G.W.W., J.J. de Jong, H.F. Van Dobben, and M.N. Van Wijk, 2005. Additional costs of nature management caused by deposition, Ecological Economics, 52:437–451. Wamelink, G.W.W., C.J.F. Ter Braak, and H.W. Van Dobben, 2003. Changes in large-scale patterns of plant biodiversity predicted from environmental economic scenarios, Landscape Ecology, 18(5):513–527.

1188 O c t o b e r 2 0 0 7

Weiss, M., F. Baret, M. Leroy, O. Hauteceur, C. Bacour, L. Prevot, and N. Bruguier, 2002. Validation of neural net techniques to estimate canopy biophysical variables from remote sensing data, Agronomie, 22:547–554.

Appendix A: Explanation of the abbreviated species names in Figure 5.

Achilmil Achillea millefolium; Agrossto Agrostis stolonifera; Arctilap Arctium lappa; Brassnig Brassica nigra; Bromuine Bromus inermis; Calamepi Calamagrostis epigejos; Calyssep Calystegia sepium; Cardunut Carduus nutans; Carexhir Carex hirta; Cerasfon Cerastium fontanum; Cirsiarv Cirsium arvense; Cirsivul Cirsium vulgare; Cynoddac Cynodon dactylon; Dactyglo Dactylis glomerata; Elymurep Elymus repens; Epilohir Epilobium hirsutum; Epilotet Epilobium tetragonum; Erigecan Erigeron canadensis; Eryngcam Eryngium campestre; Euphoesu Euphorbia esula; Festurub Festuca rubra; Galeotet Galeopsis tetrahit; Galiuapa Galium aparine; Galiumol Galium mollugo; Glechhed Glechoma hederacea; Hernigla Herniaria glabra; Loliuper Lolium perenne; Lycopeur Lycopus europaeus; Lythrsal Lythrum salicaria; Matrimar Matricaria maritima; Mediclup Medicago lupulina; Melilalt Melilotus altissima; Menthaqu Mentha aquatica; Odontver Odontites vernus; Oenotbie Oenothera biennis; Phleupra Phleum pratense; Plantlan Plantago lanceolata; Poa ann Poa annua; Poa pra Poa pratensis; Poa tri Poa trivialis; Polynper Polygonum persicaria; Potenans Potentilla anserina; Potenrep Potentilla reptans; Ranunrep Ranunculus repens; Rubuscae Rubus caesius; Rumexace Rumex acetosa; Rumexcri Rumex crispus; Rumexobt Rumex obtusifolius; Saponoff Saponaria officinalis; Senecina Senecio inaequidens; Senecjac Senecio jacobaea; Solidcan Solidago canadensis; Stellaqu Stellaria aquatica; Tanacvul Tanacetum vulgare; Taraxoff Taraxacum officinale s.s.; Triforep Trifolium repens; Urticdio Urtica dioica; Verbanig Verbascum nigrum.)
