Two decades of inland water monitoring using satellite radar altimetry

Transcription

1 Two decades of inland water monitoring using satellite radar altimetry P.A.M. Berry (1) (1) De Montfort University, Faculty of Computer Sciences and Engineering, Leicester LE1 9BH, United Kingdom ABSTRACT The most well established use of altimetry over land is the measurement of inland water heights. This field has evolved rapidly over the past two decades. Initial work over a handful of large targets has expanded to the current capability to monitor thousands of river and lake heights worldwide. Two factors have been critical to the advances made in inland water monitoring. The first is the inclusion of a designed capability to track rapidly varying land surfaces, deployed on the ERS RA-1 and Envisat RA-2. The second is in the analysis of inland water echoes, with the ability to identify and retrack to that part of a complex return corresponding to the underlying water surface. This paper presents a global overview of the abilities of past and current altimeters to monitor the Earth s changing inland water resources, utilising a unique database of retracked heights from ERS-1, ERS-2, Envisat, TOPEX and Jason-1. The importance of continuing this unique monitoring capability is illustrated, and issues relevant to a future hydrology mission are discussed. 1. INTRODUCTION The analysis and interpretation of satellite radar altimeter data obtained over inland water was pioneered using data from the Seasat mission. Despite its short life, this mission provided a valuable first look at echoes from inland water, with a precision that allowed meaningful measurements to be made over a small number of large lake targets and the Amazon river basin [1]. The advent of Geosat provided a further opportunity, curtailed by the serious off-pointing problem which severely constrained the use of these data for land and inland water applications (ibid). 2. DEVELOPMENT Over the past ten years, a series of altimeter missions have been flown. The primary data-source for most published results obtained over inland water is the longrunning TOPEX mission; data products readily available give a range estimate derived from the onboard tracker, and the user friendly format of the MGDR [2] has meant that these data have been used extensively. The TOPEX mission is succeeded by Jason-1; as will be discussed later, this altimeter has proven to be far less effective at capture of echoes over inland water surfaces than TOPEX. Both these instruments were designed primarily for retrieval of accurate data over the ocean; in consequence they have difficulty in retrieving data over rough terrain, which limits the targets they are able to acquire. However, where echoes are retrieved over an ocean-like target, good vertical precision can be obtained. Of the ESA altimeter missions, the ERS-1 mission gives interesting glimpses into results obtainable from different temporal sampling, and the Geodetic Mission allows land-mapping applications to be addressed; however, the changing orbit configurations make derivation of time series over inland water impossible over all but a small number of very large targets. The dataset from ERS-2 provides a long time-span over inland water, with the ice mode utilised over land greatly facilitating the acquisition of inland water targets at the cost of some vertical precision. Unfortunately, the user hostile format of the WAP product [3] has prevented these data being generally used. This is a great loss to the scientific community, as the RA-1 performed very well over both land and inland water surfaces; the potential of these extremely good data has not been realised to anything like the level which they merit. This data series is now continued by the Envisat RA-2, which is able to acquire data over most of the Earth s land surfaces using three modes of operation which enable it to maintain lock over all but the most mountainous terrain (though the 20Mhz mode is too coarse to obtain meaningful height changes over inland water). 3. CURRENT STATUS Studies have been published by a number of researchers on height retrieval over a limited selection of rivers and lakes, primarily using data from TOPEX and, more recently, Envisat. The most extensive data series available can be found online [4,5,6]. Combining data from different satellites is possible over large lake targets. To illustrate this, Fig. 1 shows an example of combined data-series over Tonlé Sap derived from TOPEX, ERS-2 and Envisat. Intercomparing results derived by different researchers also demonstrates good agreement (Fig. 2). This

2 approach has proven valuable in monitoring of large water targets such as the Aral Sea [7]. There is a fundamental limitation with this approach. In order to obtain good height estimates using the range estimates given in existing altimeter data products, the targets must be restricted to large water bodies which may give consistent results using the range measurements given in the products. Statistical methods must then be employed to assess the probability of the heights being valid. Even with Envisat data, where the results from four retracking algorithms are supplied in the SGDR, biases are, unsurprisingly, observed in the ranges obtained from different algorithms when these are simply intercompared [8]. The potential inland water targets are far more numerous than the handful of large lakes and major river crossings currently published. In order to assess the potential for inland water monitoring from multimission satellite radar altimetry, a global investigation was performed on the echo shapes returned over inland water from TOPEX, ERS-2, Envisat, and Jason-1. A global mask was used to select out inland water echoes from one year of each satellite mission and each waveform was analysed to determine whether basic requirements were met. Criteria included the presence of a leading edge and significant power within the waveform. All locations for which valid time series of echoes are obtained were graphed globally; results are shown in Figs. 3a,3b,3c and 3d. Whilst individual locations blend at this scale, the amount of the continental land mass where products are found can easily be seen. As expected, Envisat obtains the highest number of waveform sequences over inland water; the mode switching capability allows it to maintain lock on the terrain and to some extent compensate for the poor tracker performance over inland water, although many tributaries of major river systems are not successfully acquired. For Jason-1, the tracker performance over inland water targets causes a severe loss of data, in stark contrast to the generally extremely good performance of this instrument over land surfaces. Figure 1. Retracked multi-mission altimeter data over Tonlé Sap Figure 2. Multiple source time series over Tonlé Sap Figure 3a. ERS-2 global echo recovery over inland water

3 Figure 3b. TOPEX global echo recovery over inland water Figure 3c. Envisat global echo recovery over inland water

4 Figure 3d. Jason-1 echo recovery over inland water 4. WAVEFORM SEQUENCES Whilst obtaining decadal time series of echoes over many thousands of inland water targets is clearly possible, many of these echoes are from complex targets, especially in river basins. Each echo may be a composite response from a combination of flowing water, still pools, interruptions to the water surface and surrounding topography. To illustrate this, Fig. 4 shows part of a waveform sequence obtained over the Amazon river basin. Here, the presence of multiple bright targets is apparent in the complex and rapidly changing sequences. Figure 4. Echo sequence over Amazon basin from ERS-2 The key to deriving good height estimates over inland water is to retrack the individual waveforms using an appropriate algorithm for each echo shape. Using an expert system approach [9] allows these echoes to be retracked, isolating and retracking to the inland water element within complex waveform shapes. One consequence of adopting this approach is that a wide target is not required; a small number of echoes gives a good height estimate, as is illustrated in Fig. 5 which shows a comparison over the Amazonas river against gauge data [10] Examples As an illustration of information which can currently be derived, if data are retracked effectively, Fig. 6 shows a series of examples over the Brahmaputra river, using data from TOPEX, ERS-2 and Envisat. For reference, the locations of all good echo sequences are illustrated by circles on the area plot. It is clear that by retracking these echoes effectively, good time series are obtained. As a further illustration Fig. 7 shows time series from the Congo basin. 5. BURST ECHOES An exciting new opportunity has arisen from the Envisat RA-2, with the availability of burst echo sequences over inland water. Fig. 8 shows a sequence obtained over the Amazon basin. The top figure shows the 18Hz dataset from the SGDR, with the location of the burst sequence marked by blue lines. The visualisation of the burst sequence shows detail of the complex response of this surface to Ku band altimetry. Retracking both the burst echoes at full 1800 Hz (shown in red) and the SGDR 18Hz echoes (in blue) demonstrates both good agreement and detail of the response when an off-

5 ranging arc (on the right) begins to dominate the echo power. 6. NEAR-REAL-TIME MEASUREMENTS The RA-2 returns waveforms at 18Hz, which are contained in the level 1B product. Whilst the orbit on this product is greatly in error and cannot currently be used, the IGDR product, available after a short delay of 2-3 days, has a corrected orbit. Fusing these two datasets and retracking the 18Hz waveforms thus allows near-real-time derivation of river and lake heights. A pilot system [5] has been developed to produce near real time heights over rivers and lakes globally. Figure 5. Comparison of ERS-2 data with gauge showing a) heights against gauge b) number of points used in deriving altimeter heights 7. DISCUSSION As all the published results demonstrate, the satellite radar altimeter is a superb instrument for measurement and monitoring of the Earth s inland water resources. In particular, TOPEX and ERS-2 have obtained long time series of data, now being continued by Envisat and Jason-1. The unique multi-mission dataset already gathered, with its varied temporal sampling, frequencies of operation, and differing instrument characteristics has provided a vast database of echoes over more than a decade and sampling many thousands of rivers and lakes worldwide. This resource is only now being exploited, as the advent of increasingly sophisticated retracking techniques allows retrieval of inland water heights from complex river targets as well as lake surfaces. The unique burst mode capability of the RA-2 adds further information. This decadal global dataset holds a vast amount of climate related information; one conclusion from all researchers in this field is that this valuable monitoring capability must be continued by future altimeter missions. Analysis of sequences of echoes from past and current altimeters, together with the burst echoes from the RA-2, all demonstrate the complexity of inland water targets at all spatial scales. This reinforces the requirement for instrument design for a future water mission to model a realistic response of these targets. In conclusion, whilst many applications of radar altimeter data are mature, this comparatively recent area of research is still evolving rapidly, and the potential of this technique is only now being realised. Every year the boundaries of measurement are pushed further back, as the capabilities of the instruments are assessed and techniques for height measurement enhanced. REFERENCES 1. Rapley, C.G., Guzkowska, M.A.J., Cudlip, W., Mason, I.M. Altimeter studies of inland water and land in preparation for ERS-1. (1987) Digest - International Geoscience and Remote Sensing Symposium (IGARSS) p Benada, R. and S. Digby (1997), TOPEX/POSEIDON Altimeter Merged Geophysical Data Record Generation B (NASA/PO.DAAC), JPL PO.DAAC 068.D002, Jet Propulsion Laboratory, Pasadena, CA. 3. Capp, P. (2001), Altimeter Waveform Product ALT.WAP Compact User Guide, Issue 4.0, PF- UG-NRL AL-0001, Infoterra Ltd., UK. 4. Hydrology from space web site [Accessed May 2006] 5. ESA River and Lake project web site [Accessed May 2006] 6. Global Reservoir and Lake Monitor web site ervoir/ [Accessed May 2006] 7. Aladin, N., Crétaux, J.-F., Plotnikov, I.S., Kouraev, A.V., Smurov, A.O., Cazenave, A., Egorov, A.N., Papa, F. Modern hydro-biological state of the Small Aral sea (2005) Environmetrics 16 (4), pp

6 8. Frappart, F., Calmant, S., Cauhopé, M., Seyler, F., Cazenave, A. Preliminary results of ENVISAT RA-2-derived water levels validation over the Amazon basin (2006) Remote Sensing of Environment 100 (2), pp P.A.M. Berry, J.D. Garlick, J.A. Freeman, and E.L. Mathers. Global Inland Water Monitoring from Multi-Mission Altimetry Geophysical Research Letters, 32 (16), L16401, DOI: /2005GL HidroWeb web site [Accessed May 2006] Figure 6. Sample multi-mission time series over the Brahmaputra: red TOPEX, blue ERS-2, pink Envisat. Circles show all possible targets from ERS-2.

7 Figure 7. Sample multi-mission time series over the Congo basin: red TOPEX, blue ERS-2, pink Envisat. Circles show all possible targets from ERS-2. Figure 8. Burst echo sequence over the Amazon basin

8 Figure 9. Retracked burst echo orthometric heights over the Amazon basin