New perspectives in monitoring water resources in large tropical transboundary basins based on remote sensing and radar altimetry FREDERIQUE SEYLER1, STEPHANE CALMANT2, JOECILA DA SILVA3, JUAN-GABRIEL LEON4, FREDERIC FRAPPART5, MARIEPAULE BONNET5, NAZIANO FILIZOLA6, EMMANUEL ROUX1,GERARD COCHONNEAU1, ANA-CAROLINA ZOPPAS COSTI7, EURIDES DE OLIVEIRA8, JEAN-LOUP GUYOT5, PATRICK SEYLER5 1 Institut de Recherches pour le Développement, IRD, US 140 ESPACE, CP 7091, Lago Sul, 71970-320, Brasilia, DF, Brasil, frederique.seyler@ird.fr 2 Université de Toulouse, France, UMR 5566 LEGOS, IRD, CNRS, CNES 3 Universidade Federal do Rio de Janeiro, Brasil 4 Universidade Nacional de Colombia, Colombia 5 Université de Toulouse, France, UMR 5563 LMTG, IRD, CNRS 6 Universidade do Estado de Amazonas, Brasil 7 CPRM Servico Geologico do Brasil, Brasil 8 ANA Agencia Nacional de Aguas, Brasil Abstract: The combined use of remote sensing and radar altimetry is offering entirely new perspectives for the monitoring of water resources in large tropical trans-boundary basins. Based on a study conducted mostly in the region of the Llanos de Mojos, a large complex of wetlands located within the southernmost extension of the Amazon Basin, at the Brazilian border with Bolivia and Peru, but also based on previous studies, we give some results that will illustrate this point of view. First, the results are showing the current limitations of the radar altimetric missions, which were designed primarily for ocean level or ice caps studies: essentially the revisit time, the size of the water bodies that can be monitored, and the lack of reliable data in presence of relief. Nevertheless, they also show that the data processing and the tools we developed to select appropriately the data, allow retrieving quite accurately the seasonal variability of the water elevation within the basin selected. Second, we are stating the absolute necessity of coupling these data with remote sensing images, in order to retrieve useful information on the hydrologic behaviour of this very complex system. The common altitudinal reference of the radar altimetry missions is opening new modelling opportunities, as the river slope is a key parameter for hydrodynamical studies. Last, the spatial distribution reachable nowadays, and the perspectives offered by the future sensors, are going forward a global, detailed capability of monitoring wetlands and floodplains, as well as their relationship with the river flow. The applications resulting from these monitoring tools are of primarily importance for tropical poorly gauged basins, and we will review as a conclusion some of them, as infrastructure monitoring and planning, flood and drought monitoring and forecast, fluvial waterway monitoring and transport planning, fluvial dynamics of the riverbed and discharge modelling. Key words: Water Resources, Remote Sensing, Radar Altimetry, Trans-boundary Basins INTRODUCTION Radar altimetry missions for ocean studies began with GEOS3, launched by NASA in 1975 with both objectives of defining the structure of the earth's gravitational field and mapping the ocean surfaces. It was followed by SEASAT in 1978. SEASAT was managed by JPL and had the first space-borne synthetic aperture radar (SAR) along with a radar altimeter. SEASAT with 60 cm precision on orbit radial component was the first to be scientifically used for ocean mapping, although it flew only three months before switching off. It was followed by GEOSAT, and this successful mission has given way to three lineages of satellite, GFO (GEOSAT Follow-On), ERS/ENVISAT and Topex/Poseidon/Jason. From Topex/Poseidon (T/P) launched in 1992, and its 3 cm precision on orbit radial component (Le Traon et al., 1995), were initiated studies dedicated to inland seas and lakes. In the first PhD dissertation defended in France on continental water altimetry, Mercier (2001) stated "In spite of great potentiality in the domain of continental hydrology, (…) the number of team practising satellite altimetry beyond the border of ocean studies is small". The remark could be made today, eight years later. Basically, three teams are processing and distributing radar altimeter data for continental water studies: (a) The Global Reservoir and Lakes Monitor Project team supported by NASA and USDA (United States Department of Agriculture) and leaded by C. Birkett (1995a, 1995b). The project has used archive data from the T/P mission (1992-2002) and uses now near-real time radar altimeter data from the Poseidon-2 instrument on-board the Jason-1 and Jason-2 satellites which were launched respectively in December, 2001 and in July, 2008. It delivers time series of lakes and reservoir surface height variations for about 100 locations around the world, with an accuracy expected to be better than 10cm rms. (http://www.pecad.fas.usda.gov/cropexplorer/global_reservoir/). (b) The River and Lake product team supported by ESA and De Montfort University (UK) and led by P. Berry (2007).The project is aimed at developing, demonstrating and assessing an information service based on inland water altimetry (tethys.eaprs.cse.dmu.ac.uk/). (c) The Hydroweb team at LEGOS (Laboratoire d'Etudes en Géophysique et Océanographie Spatiale), Toulouse, France and the associated laboratories, in particular the LMTG (Laboratoire d'Etudes et Transfert en Geologie), which has led the CASH project (Contribution of Spatial Altimetry for Hydrology) from 2004 to 2006. The objective of the CASH project was to define the scientific and technological environment necessary to use the altimetric elevation of continental water surface in complement of conventional gauging network data in the scope of hydrological studies. Some of the applications developed during the lifetime of the project will be briefly described hereafter. Today, the CTOH (Centre Topographique des Océans et de l'Hydrosphere), part of LEGOS, has been approved as Observation Service and is supported by CNES (Centre National d'Etudes Spatiales), CNRS (Centre National de Recherche Scientifique) and IRD (Institut de Recherches pour le Développement), from France. It distributes freely altimetric products along track ameliorated specifically for continental water, from the missions: T/P, JASON 1 and JASON 2, GFO, and ENVISAT RA-2. Hydroweb is a database of water levels temporal series over great rivers, lakes and wetlands distributed by the GOHS (Geodésie, Océanographie et Hydrologie Spatiale) team of LEGOS (http://www.legos.obs-mip.fr/fr/soa/hydrologie/hydroweb/). CLS (Collecte Localisation Satellite) also in Toulouse, France, which was partner of the CASH project, is distributing retracked T/P data (retracked with the four algorithms used in the ENVISAT RA-2 altimetry products: OCEAN, ICE1, ICE2 and SEAICE) over the 8 great fluvial basins studied by CASH: Amazon (pilot site), la Plata - Parana, Congo, Doner, Lena, Ganga - Brahmaputra, Mekong, Yellow River. (http://ocean.cls.fr/html/cash/donnees/access_fr.html). CLS is currently pursuing the efforts of adapting the altimetry data retracking algorithms to the specificity of continental waters, through the PISTACH project (Improved Jason-2 Altimetry Products for Coastal Zones and Continental Waters), funded by CNES. Part of the CASH project team has been relocated by IRD in Brazil, in the frame of bilateral cooperation projects, to pursue the applications of radar altimetry for the Amazon basin hydrology, and the transfer of the developed methodologies to Brazilian partners. Three main projects are developed: (i) Amazon basin Hydrological Monitoring from Space (Monitoramento Espacial Hidrológico na Bacia Amazônica), which is a technical cooperation project between IRD and ANA (Agencia Nacional de Aguas) approved by the ABC (Agencia Brasileira de Cooperação). The objective of the project is to build a database of satellite altimetry and imagery products, with the same standards as the hydrological network of gauges maintained by ANA. (ii) Fluvial Dynamics of the system Negro-Solimões-Amazon, developed in the frame of a bilateral understanding agreement between IRD and CPRM (Serviço Geologico do Brasil). The project aims at calibrating/validating altimetric data by the way of field campaigns and installation of permanent equipment under the satellite tracks, with the ultimate objective of using the altimetric data in fluvial dynamics monitoring and fluvial evolution prediction. (iii) Project of Spatial Hydrology for the Amazon Basin (Projeto de Hidrologia Espacial na bacia Amazônica - PHIESAM). The project involves the creation of a team dedicated to spatial hydrology studies within the UEA (Universidade do Estado de Amazonas), with the collaboration of UFAM (Universidade Federal de Amazonas) and INPA (Instituto Nacional de Pesquisas da Amazônia). Basically, these projects are sharing the same vision of radar altimetry for hydrology. Although not optimized for continental water study, water surface elevation measured from space can be used for hydrological monitoring of great trans-boundary basins. After reviewing briefly the main applications of radar altimetry to hydrology, we will draw the perspectives offered for water resources monitoring globally, based on the example of a study developed in the Llanos de Mojos, at the border between Brazil and Bolivia, with ENVISAT RA2 data and satellite imagery. BRIEF HISTORY OF RADAR ALTIMETRY APPLICATIONS TO HYDROLOGY GEOS3 was the first altimeter to be dedicated to ocean mapping, but yet, some authors sought to use these new sensors to land mapping and continental water monitoring. Miller (1979) shown that GEOS3 data could be used to monitor lakes water level. From these pioneer studies, three main applications have been drawn. A review of the application of radar altimetry to hydrology can be found in Calmant et al., in press. We will only give here some salient points: a) The first line of applications concerns water resource monitoring in relation with climate and agriculture. The three web sites cited herein are proposing this application line in their stake statements. The studies in this line are numerous. We can cite Mason et al. (1985), Morris and Gill, (1994), Birkett, (1995), Ponchaut and Cazenave, (1998), Birkett et al., (1999), Birkett (2000), de Oliveira Campos et al., (2001), Mercier et al., (2002). Most of these studies were conducted over great lakes. A review of radar altimetry over lakes can be found in Creteaux and Birkett, (2006). The first study on river was conducted by Koblinsky et al., (1993) from GEOSAT waveforms. They estimated to 70-cm rms the discrepancy between satellite and in-situ measurements at four sites on the Amazon. They attributed partly the uncertainties to the orbit determination, but partly to the in-situ record, and we will comment that in the next paragraph. The altimetric data that have been most used in these studies have been those of T/P (10-day repeat period, since 1992), followed by Jason-1 (10-day repeat period since 2002) and now Jason 2 since July 2008. ENVISAT (following the ERS series started in 1991 with a 35-day repeat period) mission has been used only in Berry et al., (2005), and Frappart et al., (2006) with a purpose of validation on rivers. b) The second great application of radar altimetry for hydrology is coming from the fact that radar altimeters paths are crossing water bodies at all types of hydrological situations. Hydrologic networks in-situ stations are located in narrow, straight sections of the river, as they are to be gauged from time to time, to estimate discharge and be able to relate river stage and discharge, in linear stage-discharge relationship known as rating curve. For this reason the status of wetlands water level is largely unknown globally. The first application of radar altimetry for monitoring wetlands has been by Cudlip et al., (1992). But it is difficult to distinguish river and floodplain with T/P data. Birkett et al. (2002) have succeeded in a number of cases as a phase offset of a few days in stage variations between river and nearby floodplain has occasionally been observed. Frappart et al. (2005) have determined spatiotemporal variations of water volume over the main stream jointly with the floodplain in the Negro River basin, using area variation estimates for a seasonal cycle captured by the Synthetic Aperture Radar (SAR) onboard the Japanese Earth Resources Satellite (JERS-1), and changes in water level from the T/P altimetry at 88 altimetric stations, combined with 8 in-situ hydrographic stations. A volume variation of 331 km3 was estimated for the whole Negro sub-basin, enhancing the complex relationship between the volume potentially stored in the inundated area and the volume flow during the same period. Similarly, Frappart et al. (2006) monitored the flood propagation along the Mekong River by combining satellite altimetry data and SPOT4 vegetation imagery. Lastly Frappart et al., (2008) have used combined satellite imagery to re-estimate the volume of water stored in the inundation plains of the Negro sub-basin with a better temporal resolution. The first study using ENVISAT data to examine the relationship between river and floodplain through the differences in water levels was conducted by Cauhope, (2004). On the same floodplain located in the Amazon basin, Bonnet et al., (2008) have modelled the transfer of water between river and floodplain partly based on altimetric water levels time series. c) The third application is using the unique reference system of the altimetric data for studying the slope of the rivers and therefore be able to model the hydrodynamics. As for the other applications the pioneer works were conducted on the Amazon main stem with SEASAT data (Guskowska et al., 1990; Cudlip et al.,1992; Mertes et al., 1996, and Dunne et al.,1998) and T/P data (Birkett et al., 2002). With mixed T/P and ENVISAT data, Leon et al. (2006a and 2006b) have proposed a methodology to derive stream profiles from the river bed height and slope derived from altimetry, through the estimation of rating curves at the altimetric virtual stations. Some works have estimated discharge at altimetric stations using empirical regressions from in situ gauging stations (Coe and Birkett , 2004; Kouraev et al., 2004; and Zakharova et al., 2006). The present study is mixing both applications: monitoring of wetlands and estimating slope and discharge. DRAWBACKS The major drawbacks in using radar altimetry for hydrology are often reported as being the precision of the measurements and the revisit time of the altimeters (10 days for T/P and Jason and 35 days for ERS/ENVISAT). We can add the size of the water bodies that can be monitored, and the lack of reliable data in case of steep relief close to the margin and before the river in the flight direction (Seyler et al., 2008). Seyler et al., (2008) have shown that seasonal fluctuation of river stage can be captured for rivers width lesser than 100m. As for precision and revisit time, most of the studies attempting to validate altimetric data have used comparison with in situ data. We will discuss this notion in this study. For example, Bercher et al. (2006) evaluated the amount of information lost due to the revisit time of the currently used radar altimeters under-sampling, compared with daily measurements at in situ stations. But for some hydrologic applications, as estimation and monitoring of fast flood events, hourly or even more frequent measurements are needed. Nevertheless, daily in-situ water stage measurements have never been questioned for most hydrological application. In the same way, we will try to show in this study, that for some applications, altimetric measurements are able to monitor seasonal variations of water level in vast inundated transboundary basins, and are very valuable as they are today, without speaking of our need to adapt and prepare for the future altimetric missions, which will be designed for continental waters, as the SWOT mission (Alsdorf et al., 2003). DATA AND METHODS The study is located at the Brazilian border with Bolivia and Peru. In Brazil the region is known as the upper Madeira region. Madeira river has four main tributaries, two flowing through Bolivia: Beni and Mamore rivers, one flowing through Peru and Bolivia: the Madre de Dios River, and one forming the border between Bolivia and Brasil: The Guapore river. Mamore and Beni are meandering rivers, flowing mostly from South to North, West of an extensive inundation plain, called Llanos de Mojos. The Llanos de Mojos region is a large floodplain with variable extension related to the alternating dry and rainy tropical seasons (Ronchail et al., 2005), partly dry during Austral winter and reaching 150 000 km² at the end of the rainy season (Roche and Fernandez, 1988). Altimetric radar data The altimetric data used in this study come from the ENVISAT mission, distributed by the CTOH, specifically the range data retracked by the ICE 1 algorithm. They are along-track recorded. In order to extract the value of water surface elevation, we used a manual selection of the altimetric ranges projected in the plane perpendicular to the flow direction (Roux et al., in revision). One virtual gauge presents generally several points belonging to the same satellite cycle. In order to retrieve the time series for the virtual gauge, we used the median and a median dispersion of the points located across the water body (Santos Da Silva et al., 2008). Finally, the value of the geoid GGM02C has been added to the elevation value, referenced to the ellipsoid WGS84 (Tapley et al., 2004). For purpose of validation, the data at six virtual stations have been compared with the water levels at 6 conventional gauges located on the Madeira and the Guapore rivers gauges (Porto Velho, Principe da Beira, Pimenteiras, Abuna, Pedras Negras, Vila Bela de Santissima Trindade, data distributed by ANA, www.ana.gov.br). Figure 1 is presenting the location of the 31 virtual gauges presented in this study. Satellite image data The satellite images used were: a) JERS-1 (L Band SAR launched by the National Space Development Agency of Japan (NASDA) in February, 1992) image mosaic from the Global Rain Forest Mapping (GRFM) project at 100m resolution (Siqueira et al., 2000). b) C-Band SRTM (Shuttle Radar Topography Mission) data are DEM (Digital Elevation Model) data that have been collected throughout a 10-day acquisition period in February 2000. The DEM has a relative vertical error of 5.5 m over land with 90% significance 4 3 2 Madre de Dios 8 1 20 21 6,7 23,24 22 12 25 26 5 10 18 9 11 27 19 28 to 30 Rio Itonamas Mamore 13 Guapore 31 15 Beni 16 17 14 Fig. 1: Location of the virtual gauges (Rodriguez, 2005). Errors over open water are largely unknown but were estimated to ±5.51 m by Le Favour and Alsdorf, (2005), after removing of the linear trends. Geomorphological parameters and discharge estimation Three main geomorphological parameters have been estimated at the virtual stations, and are presented Table 2. Each virtual station is identified by a sequential number, the satellite track number, the river it crosses, and the latitude-longitude of the virtual station's centre given as the mean of the longitude and latitude for the points constituting the station. The width (L) has been measured from JERS1 images mosaic by the distance measuring tool of ARCGIS, and is therefore given with an uncertainty of 100m, being the resolution of the image. The maximum level difference H (m) has been estimated by the following formula (1): H = MAX(hi) - MIN(hi) (1) hi = median of the water surface altitude at the virtual station for each ENVISAT cycle (m) Slope (m/m) has been estimated by two methods: The first one is given by S1 = MIN(hi) - MIN(hj) / dist(SVi to SVj) (2) SVj being located downstream SVi and dist(SVi to SVj) being the distance between SVi to SVj in m, and measured by the distance measuring tool of ARCGIS. S2 = MEAN(MIN(hi) - MIN(hj) / dist(SVi to SVj); MIN(hh) - MIN(hi) / dist(SVh to SVi) (3) SVh being located upstream and SVj downstream SVi Bankfull discharge Q1* (m3/s) has been estimated for each virtual station by: Q* = 4.00·A*1.21·S0.28 (4) given by Williams (1978) where Q* is the bankfull discharge, A* cross-sectional area at bankfull and S the slope. A* has been approximated by: A* = L * H (5) Another formula is given by Bjerklie et al., (2003): Q2* = 0.1676xL*1.86 (6) The estimation of discharge has been evaluated by comparing with monthly discharges estimated at two conventional gauges (Table 1). Table 1: Minimum (Min Q), Mean (Moy Q), and Maximum (Max Q) of monthly discharges (m3/s) estimated between, respectively, 1970, September - 2007, November at Guaraja-Mirim, on Mamore river upstream the confluence with the Guapore River, 1967 May - 2008 June at Porto Velho, on the Madeira river, 331km downstream the confluence Beni-Mamore. Conventional gauge Guaraja-Mirim Min Q 1039 Moy Q 7688 Max Q 21280 Porto Velho 2220 18394 47400 RESULTS AND DISCUSSION Validation We can analyze the precision of the altimetric data by analysing the discrepancy between water levels at conventional gauges and at virtual gauges. Table 2 gives the rms standing for the discrepancy at 6 couples conventional gauges/altimetric virtual stations The rms is calculated with respect to the best fitting regression line, which coefficient is provided. Table 2: Comparison between in situ gauge and virtual gauge. Conventional gauge Porto Velho Principe da Beira Pimenteiras Abuna Pedras Negras Vila Bela de Santissima Trindade Altimeter track Distance(km) Rms (m) 18 Regression coef. (m/m) 1.06 951 192 5 0.75 1.349 478 278 106 0 30 29 1.02 0.99 0.74 0.170 1.496 0.689 392 0 0.62 0.342 0.395 As we have seen, it is very difficult to assess the precision of the altimetric data by comparison with conventional gauges, as the first one is averaging the surface elevation across the channel of the river, and the second one is measuring at one single point located on one bank of the river. First, the precision of the conventional gauge with respect with the spatial variability of the river surface is never stated. Koblinsky et al., (1993) have attributed part of the error in altimetric measurements to the error at in situ gauges, as they have estimated the discrepancy between two conventional gauges located on the opposite banks of the Tapajós river. But most of the time, the in situ gauge is given as the absolute reference of the river surface plane. Second, it is very uncommon to have an altimeter track right above the conventional gauge taken as reference. When located at some distance from each other, the river section can change, or unknown amount of water can reach the river between the two points from an ungauged tributary, or another unknown amount of water can be temporarily lost in a derivation or an inundation plain. Analysing table 2, we can summarize the validation situation in three categories: a) the regression coefficient is very different from 1, which is the case when the river sections (conventional and altimetric) are not similar: Principe da Beira (regression coefficient of 0.75; the track is partly crossing over a braided part of the river), Pedras Negras (regression coefficient of 0.74; the track is crossing in a 30° angle from the river longitudinal profile, allowing to take in account part of the longitudinal slope of the river in the altimetric measurement). The last case is that of Vila Bela de Santissima Trindade. The track 393 of ENVISAT is crossing the Guapore river right above the conventional gauge, but is averaging the 100m wide channel at this point plus 17km of floodplain, which gives a regression coefficient of 0.62. b) the regression coefficient is near 1, which could indicate a similar river section. It is the case for the comparison made at Porto Velho and Abuna. The Madeira river has an equivalent width at both conventional and virtual station, and the track is crossing in a near perpendicular direction from the river flow. But the distance between the two measurements points is large, 18 km for Porto Velho and 30 km for Abuna. In both cases, there are some derivations by floodplains. For Abuna, the derivations seem very important ones, as seen on JERS images, and a tributary is joining the river between the conventional gauge and the satellite track. c) the regression coefficient is near 1, and the distance between the two measurements points is null. This is the case at Pimenteiras. This is the only case where the precision of the altimetric data can be assessed: 17cm rms. This is within the range given in Calmant et Seyler, (2006) for the ENVISAT data (decimetric precision), taking into account the spatial variability of the river across the channel. General slope across the wetlands of Llanos de Mojos and between the four rivers draining the area Fig 2 presents a comparison between ENVISAT and SRTM data across the Madre de Dios et Beni rivers, and a NNW-SSE cross cut of the Llanos de Mojos. SRTM and ENVISAT data are very close from each other across the floodplain and differs 10 to 50 meters when crossing Beni and Madre de Dios rivers. The difference can be due to a difference of geoid model. We have chosen the GGM02C model to retrieve altitudes from ENVISAT data, and SRTM altitude data are given in reference to WGM96 model. Despite these discrepancies, both dataset are according to place the Llanos de Mojos as a very plane region declining from West to East between 165 to 150m high. The two rivers Madre de Dios and Beni are draining a lowest region separated by various reliefs from the vast inundation plain of the Llanos. Fig 3 presents two cross cuts of ENVISAT tracks. The first one, track 106, at the top of the figure is cutting the llanos de Mojos from NNW to SSW between the Mamore and the Guapore rivers. The second one, track 035 is cutting the llanos from NNE to SSW, West of the Guapore, across the area drained by the Guapore river and its tributaries. Both tracks are showing a general decline of the altitudes from South to North. The profiles are also indicating that most part of the inundation plain is drained by various tributaries of the Guapore river. Llanos de Mojos is then distributed in two vast wetland systems, located East of the Beni and Madre de Dios basins, without evident communication with these two watersheds. The two parts of Llanos de Mojos plateau are: first one located South West of Mamore river, drained by the Mamore and its tributaries mostly oriented SW-NE, direction enhanced by a string of Llanos de Mojos Madre de Dios Beni Envisat Track 121 N Madre de Dios Envisat Track 665 Beni Llanos de Mojos Fig. 2: Cross cut of the rivers Madre de Dios and Beni, and of the Llanos de Mojos from NNW to SSE. Pink line represents the ENVISAT data (track 665 at the bottom and track 121 up). Red line represents a height profile of the SRTM data at the same location than the ENVISAT track. In the middle, the JERS image of the area. Mamore Rio Blanco Guapore ENVISAT track 106 ENVISAT track 035 Rio Machupo Guapore Fig. 3: Cross cut of the rivers Mamore and Guapore, and of the Llanos deMojos. geometrically-shaped lakes. The second one is drained by the Guapore and its tributaries mostly in a direction SSW-NNE. The two parts are gently sloping NNE. There is no evidence of communication between the two parts. Morphological parameters at virtual stations and estimation of discharge Table 3: Morphological parameters at virtual stations: SV: Number of station localized in Fig1. Track: ENVISAT track number; River MD: Madre de Dios, B: Beni; M: Mamore; G: Guapore; B-M confluence Beni Mamore; M-G Confluence Mamore-Guapore L: rivers draining the geometrically-shaped rivers left bank Mamore; I: Itonamas river; aff. G tributaries left bank Guapore; mean lat: mean latitude at the virtual station; mean lon: mean longitude at the virtual station; L: Width (m); H: Maximum level difference (m); S1: Slope (m/m) calculated by (2); S2: Slope (m/m) calculated by (3); Q*1 full bank discharge (m3/s) estimated by formula 4 with S1 (2); Q*2 (m3/s) full bank discharge estimated with formula 4 with S2 (3); Q*3 (m3/s) full bank discharge estimated with formula 6. SV Track River mean lat mean lon L H S1 S2 Q*1 Q*2 Q*3 1 207 MD -11.61 -67.67 600 10.518 0.000119 0.000119 12618 12618 24640 2 665 MD -11.34 -67.05 1400 8.44 0.000111 0.000115 26476 26717 119144 3 121 MD -11.08 -66.39 640 10.752 0.000089 0.000100 12912 13355 27782 4 278 B-M -10.38 -65.38 1420 8.28 0.000260 38426 33351 122329 5 207 B -13.33 -67.32 350 6.805 0.000087 0.000087 3556 3556 9042 6 665 B -12.12 -66.88 360 8.53 0.000054 0.000071 4243 4564 9528 7 665 B -11.88 -66.93 430 8.11 0.000067 0.000061 5253 5106 13259 8 121 B -11.37 -66.33 390 7.02 0.000083 0.000075 4161 4045 11057 9 665 L -13.62 -66.53 <100 1.94 0.000119 0.000119 187 187 880 10 121 L -13.28 -65.90 <100 1.77 0.000162 0.000141 182 175 880 11 579 M -13.21 -65.19 480 12.32 0.000030 0.000023 8183 7395 16270 12 736 M -12.45 -65.13 400 11.5 0.000013 0.000030 4649 5841 11591 13 192 M -14.95 -65.00 370 11.55 0.000046 0.000081 6049 7081 10026 14 650 M -16.91 -64.72 <100 7.15 0.000116 0.000116 900 900 880 15 106 I -15.50 -63.68 <100 1.346 0.000124 0.000127 121 122 880 16 564 I -16.19 -63.12 <100 2.54 0.000131 0.000203 266 301 880 17 20 I -16.65 -62.50 <100 7.192 0.000276 0.000276 1154 1154 880 18 192 aff G -12.87 -64.51 100 11.94 0 0.090018 13083 10776 880 19 650 aff G -12.95 -63.80 100 7.45 0.000037 0.000037 686 686 880 20 736 G-M -11.95 -65.02 1200 13.307 0.000088 0.000059 21812 19490 42075 21 35 G -12.12 -64.72 590 10.845 0 0.000030 7765 8695 23881 22 192 G -12.46 -64.41 430 9.572 0.000039 0.000051 5524 5953 13259 23 493 G -12.50 -63.92 260 6.04 0.000063 0.000058 1968 1918 5201 24 650 G -12.45 -63.69 150 5.18 0.000052 0.000050 796 787 1870 25 951 G -12.63 -63.17 100 5.58 0.000048 0.000054 522 538 880 26 564 G -13.12 -62.41 100 5.39 0.000060 0.000063 531 539 880 27 20 G -13.57 -61.79 100 4.85 0.000067 0.000087 481 518 880 28 865 G -13.54 -61.53 100 3.87 0.000107 0.000088 418 396 880 29 478 Guapore -13.51 -61.06 100 4.6 0.000070 0.000079 457 473 880 30 936 Guapore -13.94 -60.44 100 4.095 0.000088 0.000092 424 430 880 31 392 Guapore -14.88 -59.94 <100 1.54 0.000097 0.000097 133 133 880 Table 4: Full Bank Discharge by rivers MD: Madre de Dios; B: Beni; M: Mamore; G: Guapore; Sum M-G: Sum of value for Mamore and for Guapore; Md: Mamore after the confluence with Guapore; Sum 4tr. Sum of the values for the 4 tributaries; Q*1 full bank discharge (m3/s) estimated by formula 4 with S1 (2); Q*2 full bank discharge (m3/s) estimated with formula 4 with S2 (3); Q*3 full bank discharge (m3/s) estimated with formula 6. MD B D*1 12912 4161 M 8183 G 7765 Sum M-G Md Sum 4tr. 15918 21812 38885 D*2 13355 4045 7395 8695 15090 D*3 27782 11057 16270 23881 40151 19490 36890 42075 80914 Table 5: Contribution (%) of the four tributaries at the full bank discharge of Madeira river MD B M G Md %total D*1 %total D*2 33 11 21 20 36 36 11 20 24 29 %total D*3 34 14 20 30 22 The bank full discharges estimated from the slope downstream the virtual station (S1) and from the mean slope upstream and downstream the station (S2) are very close to each other. Comparing the bank full discharge of the Mamore upstream the confluence with the Guapore (Md), with the maximum discharge at Guajara-Mirim, the closest value is that of D*1, calculated only with the downstream slope. The value of D*1 summed for the four tributaries (sum 4tr., Table 4) is closest too to the values of maximum discharge in Porto Velho. Full bank discharges calculated with formula 6 (only the width of the river) seem highly overestimated. Retaining the values of D*1 as the more realistic, it can be assessed that at full bank, the Madre de Dios contributes for 33% of the discharge of the Madeira river, the Beni for only 11%, the Mamore before the confluence with the Guapore to 21%, the Guapore to 20%, and the sum of the two at the confluence with the Beni river contributes for 56% of the Madeira discharge. There are rough estimations, but it could be useful in an ungauged basin, to have estimates of discharge obtained only from satellite data. CONCLUSION Altimetric data as they are today are not well adapted to continental hydrology. Nevertheless, we can assess that radar altimeter can monitor correctly the seasonal fluctuation of water stage in great trans-boundary basins. From previous studies, we can state that the width of the river is not the only criteria to take into account to predict the reliability of the time series. Rivers less than 100 meters wide can be sampled, provided that they are surrounded by a floodplain, even covered by flooded forest. Steep relief near and before (in the path direction) the river, steep longitudinal slope, islands, flow direction along track are major impediment for obtaining a reliable time series. Except these situations, the time series obtained allow monitoring the seasonal variation of water stage. Although it is very difficult to assess the precision of the altimetric data by comparison with conventional gauges, in cross-track situations at the exact location of a conventional gauge, discrepancy between virtual and conventional gauges can be analysed, and in all cases, they do not exceed 20cm. In this study, full bank discharges have been estimated for the four main tributaries of the Rio Madeira. Estimation of general slope across the Llanos de Mojos have allowed discriminating two main draining area in this vast wetland complex, one by the Mamore river, and the other by the Guapore river, without any evidence of communication, at least by superficial runoff, between the two regions. Beni river, which seems to flow very closely to the Llanos de Mojos complex on remote sensing images is not related to the wetlands, as they are forming a vast overhanging plateau gently sloping North West and separated from the Beni river by high ridges of savannas. Of course, precision, revisit time, size of the water bodies monitored, conditions of shallow relief, impose restrictions to the use of satellite radar altimetry to hydrology. Actually, they limit its use to a range of applications: levelling of conventional unlevelled water gauges; study of the relationship between the river and its floodplain and between the river and the swamps and wetlands within the watershed; study of the elevation profile and river slope. These two last applications have opened entirely new perspectives in the hydrological field: for example: infrastructure monitoring and planning (in particular monitoring of remote dams for producing hydropower), flood and drought monitoring and forecast, fluvial waterway monitoring and transport planning, fluvial dynamics of the riverbed and discharge modelling. All these applications are to be conducted with joined spatial and conventional monitoring, but in remote, tropical forested areas, or in vast wetlands, unreachable most of the time, as the region that is the object of the present study, or in trans-boundary basins where conventional data are unevenly distributed, or lastly in politically troubled region, radar altimetry monitoring is of primarily importance. In this prospect, the SWOT mission, which will be launched around 2015, and will provide a global, quasi-continuous measurement of water surface area and elevation, will provide very valuable new data sets. A lot of improvements are yet to be achieved to process the existing data, and it is necessary to carry on other validation works in distinct environment. REFERENCES Alsdorf D, Lettenmaier D, Vörösmarty C et al (2003). The need for global, satellite-based observations of terrestrial surface waters. EOS Trans., 84 (29), 269–276. Bercher N, Kosuth P, Bruniquel J (2006). Quality of river water level time series issued from satellite radar altimetry: influence of river hydrology and satellite measurement accuracy & frequency. Presented at EGU, General Assembly 2006 Vienna, AUT, 02 07 April 2006 Berry P A M, Garlick J D, Freeman J A, Mathers E L (2005). Global inland water monitoring from multi-mission altimetry. Geophysical Research Letters, vol. 32, L16401, doi:10.1029/2005GL022814 Berry, P.A.M., Freeman, J.A., Smith, R.G., Benveniste, J., (2007) Near Real Time Global Lake and River Monitoring using the Envisat RA-2. "Envisat Symposium 2007", ESA Pub. SP-636 2007. Birkett, C.M. (1995). The contribution of TOPEX/POSEIDON to the global monitoring of climatically sensitive lakes. J. Geophys. Res., 100 (C12), 25, 179–25,204. Birkett, C.M. and Mason, I.M. (1995). A new Global Lakes Database for a remote sensing programme studying climatically sensitive large lakes. J. Great Lakes Research, 21 (3), 307–318. Birkett, C.M., Murtugudde, R., Allan, T. (1999) Indian Ocean climate event brings floods to East Africa’s lakes and the Sudd Marsh, Geophys. Res. Lett. 26 1031–1034. Birkett C M (2000) Synergistic Remote Sensing of Lake Chad: variability of Basin inundation, Remote Sens. of Environ., 72, 218-236. Bjerklie, D. M., Dingman, S. L., Vorosmarty, C. J., Bolster, C. H., Congalton, R.G. (2003) Evaluating the potential for measuring river discharge from space. Journal of Hydrology, 278,1-4, 25 July,17-38. Bonnet, M.P, Barroux, G., Martinez, J.M., Seyler, F., Moreira-Turcq, P., Cochonneau, G., Melack, J.M., Boaventura, G. Maurice-Bourgoin, L., León, J.G., Roux, E., Calmant, S., Kosuth, P., Guyot, J.L., Seyler, P. (2008). Floodplain hydrology in an Amazon floodplain lake (Lago Grande de Curuaí) Journal of Hydrology, Volume 349, Issues 12, 30 January 2008, Pages 18-30. Calmant S., Seyler, F., (2006). Continental surface water from satellite altimetry. C.R. Geosciences 338, 1113-1122. Calmant, S., Seyler, F., Créteaux J.F. Monitoring Continental Surface Waters by Satellite Altimetry. Sous presse à Survey in Geophysics Cauhopé, M. (2004) Hauteurs d’eau d’une plaine d’inondation amazonienne par altimétrie spatiale, rapport de stage de DEA ‘Sciences de la Terre et l’Environnement’, 30 p., in French). Creteaux J.-F. and Birkett, C. (2006) Lake studies from satellite radar altimetry,. Comptes Rendus Geosciences, 338,14-15, 1098-1112. Cudlip, W., Ridley, J.K., Rapley, C.G. (1992) The use of satellite radar altimetry for monitoring wetlands, in: Remote Sensing and Global Change, Proc. 16th Annu. Conf. Remote Sensing Society, London, pp. 207–216. De Oliveira Campos, I., Mercier, F., Maheu, C., Cochonneau, G., Kosuth, P., Blitzkow, D., Cazenave, A. (2001) Temporal variations of river basin waters from Topex/Poseidon satellite altimetry. Application to the Amazon Basin, C. R. Acad. Sci. Paris, Ser.IIa 333 (2001) 633–643. Dunne T, Mertes L A K, Meade R H, Richey J E, Forsberg B R (1998). Exchanges of sediment between the floodplain and channel of the Amazon river in Brazil, GSA Bull., 110(4), 450-467. Frappart F., Seyler, F., Martinez J.-M., León J. G. and Cazenave A., (2005). Floodplain Water Storage in the Negro River Basin Estimated from Microwave Remote Sensing of Inundation Area and Water Levels. Remote Sensing of Environment, 99 (2005) 387 – 399. Frappart, F., Calmant, S., Cauhope, M., Seyler, F. and Cazenave, A., (2006). Preliminary results of envisat ra-2-derived water levels validation over the amazon basin.. Remote Sensing of Environment, 100, 2, 252-264 JAN 2006. Frappart, F., F. Papa, J. S. Famiglietti, C. Prigent, W. B. Rossow, and F. Seyler (2008), Interannual variations of river water storage from a multiple satellite approach: A case study for the Rio Negro River basin, J. Geophys. Res., 113, D21104, doi:10.1029/2007JD009438. Guzkowska M A J, Rapley C G, Riddley J K, Cudlip W, Birkett C M, Scott R F (1990). Developments in inland water and land altimetry. ESA contract report 78391881FIFL. Koblinsky, C.J., Clarke, R.T., Brenner, A.C., Frey, H. (1993) Measurements of river level variations with satellite altimetry, Water Resour. Res. 29 (6) 1839–1848. Leon J.G., Calmant S, Seyler, F., Bonnet M.P., Cauhopé M. (2006a). Rating curves and average water depth estimation at the upper Rio Negro from altimetric spatial data and calculated remote discharges. Journal of Hydrology, 328, 481-496. Leon J.G., Seyler, F., Calmant S., Bonnet M., Cauhope M. (2006b). Hydrological parameter estimation for ungauged basin based on satellite altimeter data and discharge modeling. A simulation for the Caqueta River (Amazonian Basin, Colombia). Hydrology and Earth System Sciences. SRef-ID: 1812-2116/hessd/2006-3-3023 Le Traon, P.-Y., Gaspar, P., Bouyssel, F., Makhmara, H. (1995) Using TOPEX/Poseidon data to enhance ERS1 data, Am. Meteorol. Soc. 161–170. Mason, I.M., Rapley, C.G., Street-Perrott, F.A., Guzkowska, M. (1985) ERS-1 observations of lakes for climate research, in: Proc.EARSeL/ESA Symposium ‘European Remote Sensing Opportunities’, Strasbourg, France, 31 March–3 April 1985. Mercier F. (2001). Altimétrie spatiale sur les eaux continentales: apport des missions TOPEX/POSEIDON et ERS-1&2 à l’étude des lacs, mers intérieures et bassins fluviaux. PhD Thesis, Université Paul Sabatier, Toulouse, France. Mercier, F., Cazenave, A., Maheu, C. (2002). Interannual lake level fluctuations (1993–1999) in Africa from TOPEX/Poseidon: Connections with ocean–atmosphere interactions over the Indian ocean, Global Planet. Changes 32 (2002) 141–163. Mertes L A K, Dunne T, Martinelli L A (1996). Channel-floodplain geomorphology along the Solimões-Amazon river, Brazil, GSA Bull., 108(9), 1089-1107. Miller, L.S. (1979) Topographic and backscatter characteristics of GEOS 3 overland data, J. Geophys. Res. 84 (B8), 4045–4054. Morris, C.S., Gill, S.K. (1994) Variation of Great Lakes waters from Geosat altimetry, Water Resour. Res. 30, 1009–1017. Ponchaut, F., Cazenave, A., (1998) Continental lake level variations from TOPEX/Poseidon (1993–1996), C. R. Acad. Sci. Paris, Ser.IIa 326 13–20. Ronchail, J., Bourrel, L., Cochonneau, G., Vauchel, P., Phillips, L.,Castro A., Guyot, J.-L., de Oliveira, E. « Inundations in the Mamore basin (south-western Amazon—Bolivia) and sea-surface temperature in the Pacific and Atlantic Oceans ». Journal of Hydrology, 302, 223–238, (2005). Roche, M.A., Fernandez, C. « Water resources, salinity and salt yields of the rivers of the Bolivian Amazon ». Journal of Hydrology, 101, 305-331, (1988). Roux, E., Santos da Silva, J., Vieira Getiranaa, A. C., Bonnet, M.-P., Calmant, S., Seyler, F. Producing time-series of river water height by means of satellite radar altimetry – Comparison of methods. En révision à Hydrological Sciences Journal – Journal des Sciences Hydrologiques. Roux E, Cauhopé M, Bonnet M-P, Calmant S, Vauchel P., Seyler, F. (2008) Daily water stage estimated from satellite altimetric data for large river basin monitoring, Hydrological Sciences Journal – Journal des Sciences Hydrologique, 53(1) February 2008 Siqueira, P.; Hensley, S.; Shaffer, S.; Hess, L.; McGarragh, G.; Chapman, B.; Freeman, A. (2000) A continental-scale mosaic of the Amazon basin using JERS-1 SAR. IEEE Transactions on Geoscience and Remote Sensing, 38(6), 2638 – 2644. Santos da Silva, Corrêa Rotunno Filho, O.J., Roux, E., Seyler, F., Calmant, S. (2008), Níveis d’água nas zonas úmidas da bacia Amazônica estimadas por satelites altimétricos. II sympósio de Recursos Hidricos do Sul-Sudeste, 12-18 Outubro de 2008, Rio de Janeiro. Seyler, F., Calmant, S., Silva, J., Filizola, N., Roux, E., Cochonneau, G., Vauchel, P., Bonnet, M.-P. (2008). Monitoring water level in large trans-boundary ungauged basins with altimetry: the example of ENVISAT over the Amazon basin. In: Remote Sensing of Inland, Coastal, and Oceanic Waters, Ed: R.J. Frouin, S.Andrefouet, H.Kawamura, M.J. Lynch, T. Platt and D.Pan. Proceedings of the SPIE Asia Pacific Remote Sensing Symposium, 17-21 Novembre 2008, Noumea, New Caledonia., Vol. 7150, 17p. Tapley, B.D., Bettadpur, S., Watkins, M., Reigber, C. (2004) The gravity recovery and climate experiment: mission overview and early results ». Geophys. Res. Lett. 31. Williams, 1978. Bankfull discharge in rivers, Water Resources Research 14: 1141-1154. Zakharova E. A., Kouraev A. V., Cazenave A., Seyler, F., (2006). Amazon river discharge estimated from TOPEX/Poseidon altimetry. Geosciences Comptes Rendus (French Academy of sciences). 338 (3): 188-196 FEB 2006.