Patrice Bicheron

GlobCover: ESA service for global land cover from MERIS

The Globcover initiative comprises the development and demonstration of a service that in first instance produces a global land cover map for year 2005/2006. Globcover uses MERIS fine resolution (300 m) mode data acquired between mid 2005 and mid 2006 and, for maximum user benefit, the thematic legend is compatible with the UN land cover classification system (LCCS). This new product updates and complements the other existing comparable global products, such as the global land cover map at 1 km resolution for the year 2000 (GLC2000) produced by JRC. It is expected to improve such previous global product, in particular because of the finer spatial resolution. The Globcover project is an initiative of ESA in cooperation with an international network of partner including EEA, FAO, GOFC-GOLD, IGBP, JRC and UNEP.

A method of biophysical parameter retrieval at global scale by inversion of a vegetation reflectance model

Abstract The objective of the paper is to study a physically based method of retrieval of leaf area index (LAI) and fraction of absorbed photosynthetically active radiation (fAPAR) parameters from global data sets of new generation wide field of view optical satellite sensors, such as POLDER/ADEOS, VGT/SPOT4, MODIS/EOS, MISR/EOS, MERIS/ENVISAT, and so forth. The method uses the bidirectional reflectance distribution function (BRDF) model of Kuusk (1995) , which simultaneously predicts the spectral and directional behavior of reflectances, as a function of LAI, chlorophyll concentration, ratio of leaf size to canopy height, and other optical or structural parameters of the soil and vegetation. The same model is used irrespective of surface type, and no ancillary data is needed. This approach is evaluated with field and airborne data acquired over three different types of surfaces: Sahelian vegetation in the HAPEX-Sahel (1992) experiment, boreal forest in the BOREAS (1994) experiment, and cultivated areas in the Alpilles (1996) experiment. The results show that the LAI is restituted with a fair accuracy [root-mean-square (rms) difference between model results and observations of 0.70], better than that obtained with a semiempirical relation LAI-vegetation index. The daily fAPAR is restituted accurately, with a rms difference between measured and modeled fAPAR of 0.097. In the example of HAPEX, the model reproduces to some extent the temporal evolution of measured LAI and fAPAR. Reflectances reconstructed with the model are in acceptable agreement with observed reflectances, with a rms difference between observed and measured values of 0.017 on average. It is concluded that the retrieval of biophysical parameters from inversion of a BRDF model is promising from the perspective of a quantitative characterization of the terrestrial biosphere from space.

Investigation of directional reflectance in boreal forests with an improved four-scale model and airborne POLDER data

Airborne Polarization and Directional Earth Radiation (POLDER) data acquired during the boreal ecosystem-atmosphere study (BOREAS) and the four-scale model of Chen and Leblanc (1997) are used to investigate radiative transfer in boreal forest. The four-scale model is based on forest canopy architecture at different scales. New aspects are incorporated into the model to improve the physical representation of each canopy, as follows: 1) Elaborate branch architecture is added. 2) Different crown shapes are used for conifer and deciduous forests. 3) Bilayer version of the model is introduced for forest canopies with an important understory. 4) Natural repulsion effect is considered in the tree distribution statistics. Ground measurements from BOREAS sites are used as input parameters by the model to simulate measurements of bidirectional reflectance distribution function (BRDF) from four forest canopies (old black spruce, old aspen, and old and young jack pine) acquired by the POLDER instrument from May-July 1994. The model is able to reproduce with great accuracy the BRDF of the four forests. The importance of the branch architecture and the self-shadowing of the foliage is emphasized.

Bidirectional reflectance distribution function signatures of major biomes observed from space

Land surface bidirectional reflectance distribution function (BRDF) measurements were acquired from November 1996 to June 1997 at global scale and 6km spatial resolution with the POLDER instrument onboard the ADEOS-I satellite. We selected 395 BRDF data sets on areas distributed on the 17 biomes of the IGBP 1-km land cover classification (DISCover data set) at 443, 670, and 865nm, at several periods (November and December 1996, May and June 1997). The selected BRDF data are characterized by a low noise level, a sufficient number of clear days during the month, and a roughly even sampling of directional space. The data show large differences of the directional and spectral signatures of the various land cover classes, both in shape and in magnitude. Except for the desert and ice classes, all signatures present a peak in the backscattering direction, with sometimes an additional strong peak in the specular direction for wetlands. The data permit an assessment of the BRDF temporal evolution due to changes of surface state or Sun elevation, as well as a quantification of the BRDF variability within a land cover class. The maximum error level of the BRDF database is estimated to be of the order of 0.01 and 0.03 (in units of reflectance) in the red and near infrared, respectively. The BRDF database is available to the science community through the Internet. It should be helpful for the prototyping of various science applications, including the test of radiative transfer models and algorithmic schemes of corrections of angular effects on remote sensing data. As an example of application of the database, various semiempirical BRDF models published in the literature are tested and intercompared. Whereas all tested models catch reasonably well the overall shape of the BRDF, some differences appear between the red and the near infrared, between classes, and between models, which the use of the database permits to quantify.

Enhanced discrimination of boreal forest covers with directional reflectances from the airborne polarization and directionality of Earth reflectances (POLDER) instrument

During the Boreal Ecosystem-Atmosphere Study (BOREAS), directional and spectral reflectance measurements were acquired from May to July 1994 with the polarization and directionality of Earth reflectances (POLDER) instrument on board a NASA C-130 aircraft. The instrument has a wide field-of-view optics, a two-dimensional CCD array, and a rotating wheel carrying filters in the visible and near infrared. Measurements were obtained (1) over coniferous forests at the young and old jack pine and old black spruce sites, (2) over a deciduous forest at the old aspen site, and (3) over a fen at the fen site. A prominent hot spot feature was apparent at each site, with an additional strong peak in the specular direction for the fen site. Strong variations of the bidirectional reflectance distribution function (BRDF) with sun zenith angle were observed. For a constant sun zenith angle, the variation of the BRDF of conifer stands between May and July was relatively weak. A key objective of this paper is to quantify the improvement of discrimination of various forest covers when remotely sensed directional signatures are added to the more conventional spectral signatures. The experimental protocol consisted of the following steps. First, 150 pixels pertaining to five different classes of forest covers were selected on land cover maps available in the BOREAS Information System (BORIS) data base. Second, the BRDF measurements acquired by POLDER at each pixel were adjusted against a three-parameter semiempirical BRDF model and processed to retrieve the reflectance seen in three different viewing directions. Third, the results of supervised classifications were compared on all selected pixels, using as input, either the reflectances in only one direction (this simulates the case of conventional spectral signatures), or reflectances acquired in three directions (this simulates the case of spectral + directional signatures). The results showed that when only one spectral band was used, the proportion of correctly classified pixels increased from 36–59% with one viewing direction to 64–84% with three viewing directions. When three spectral bands were considered, this proportion improved from 72–87% to 83–97%. These results demonstrate that the account of directional information enhances the ability to discriminate forest covers by remotesensing.

Evidence of hot spot directional signature from airborne POLDER measurements

The POLDER instrument was flown during the BOREAS experiment over various sites and at various altitudes in the Canadian boreal forest and other nearby targets. The instrument design permits the acquisition of the directional signature of any surface cover. In particular, the high directional resolution of POLDER allows it to measure, with an unprecedented accuracy, the hot spot signature of natural targets. The authors present some typical examples of such highly anisotropic reflectance directional signatures. The ratio of the maximum reflectance (hot spot direction) to the minimum reflectance (broad area in the forward scattering hemisphere) varies with wavelength and canopy. It can be as large as six in the visible and three in the near IR.

Geolocation Assessment of MERIS GlobCover Orthorectified Products

The GlobCover project has developed a service dedicated to the generation of multiyear global land cover maps at 300-m spatial resolution using as its main source of data the full-resolution full-swath (300 m) data (FRS) acquired by the MERIS sensor on-board the ENVISAT satellite. As multiple single daily orbits have to be combined in one single data set, an accurate relative and absolute geolocation of GlobCover orthorectified products is required and needs to be assessed. We describe in this paper the main steps of the orthorectification pre-processing chain as well as the validation methodology and geometric performance assessments. Final results are very satisfactory with an absolute geolocation error of 77-m rms and a relative geolocation error of 51-m rms.

An assessment of three candidate compositing methods for global MERIS time series

Compositing has long been used to produce complete, cloud-free images over large areas. Several compositing methods have been developed and implemented for global time series, and each method corrects for angular effects and atmospheric variations differently. This study assesses the performance of three compositing methods on real and simulated medium-resolution imaging spectrometer (MERIS) images. The three methods considered are best index slope extraction (BISE), mean compositing (MC), and the method of Hagolle et al. The optimal method was selected using two criteria, namely a qualitative examination of the temporal MERIS reduced resolution (RR) profiles, and a quantitative analysis of the noise introduced into composite images of the reflectance data time series. The latter calculation relies on the standard deviation of the normalized error based on simultaneous SPOT VEGETATION images. All three methods succeed in reproducing the long-term temporal evolution of the normalized difference vegetation index (NDVI). The BISE method, however, produces a temporal profile with more short-term variations. The other two methods produce very similar noise levels, with MC holding a small but consistent advantage. Owing to its performance and simplicity, the MC method has been selected to process global MERIS time series.

LAI and fAPAR mapping at global scale by model inversion against spaceborne POLDER data

The extraction of the leaf area index (LAI), or the fraction of absorbed photosynthetically active radiation (fAPAR), is one the most challenging issues related to the use of existing (AVHRR/NOAA, POLDER/ADEOS) and forthcoming (VGT/SPOT4, MODIS-MISR/EOS) satellite sensors operating in the reflected solar radiation system. Efforts in this area are at an early stage. An algorithm for the retrieval of LAI and fAPAR based on the inversion of a BRDF model against airborne POLDER data acquired over various biomes has been successfully tested. In this paper, the authors present the first performance of the algorithm against available spaceborne level 2 data (cloud screened, atmospherically corrected) over Austral Africa.

Multi-sensor data fusion for deriving bio-physical variables in the Cyclopes project

The objective of the Cyclopes project is to provide coherent estimates of bio-physical variables such as land surface albedo and leaf area index at regional and global scales and to demonstrate their application in the fields of land cover change detection and carbon cycle modeling. An essential element of the project strategy is to provide sensor-independent estimates by exploiting the synergy of an ensemble of medium resolution imaging instruments. The differences in the orbital configurations (angular sampling, frequency of observations) and sensor characteristics (spectral sensitivity, pixel size) potentially offer complementary information allowing us to increase the precision of the variable estimates. However, from a technical point of view these differences pose a number of problems and for the implementation in operational processing chains a compromise between the optimal use of the available information and the practical feasibility needs to be found. In this contribution we outline possible scenarios for achieving the multi-sensor data fusion.