Projects per year
Tidal rivers and lowland wetlands present a transition region where the interests of hydrologists and physical oceanographers overlap. Physical oceanographers tend to simplify river hydrology, by often assuming a constant river discharge when studying estuarine dynamics. Hydrologists, in turn, generally ignore the direct or indirect effects of tides in water level and discharge records. This thesis aims to improve methods to monitor, model and predict discharge dynamics in tidal rivers and lowland wetlands, by focussing on the central and lower reaches of the River Mahakam (East Kalimantan, Indonesia), and the surrounding lakes area. The 980-km long river drains an area of about 77100 km2 between 2°N - 1°S and 113°E - 118°E. Due to its very mild bottom slope, a significant tidal influence occurs in this river. The middle reach of the river is located in a subsiding basin, parts of which are below mean sealevel, featuring peat swamps and about thirty lakes connected to the river via tie channels.
Upstream of the lakes area, at about 300 km from the river mouth, an acoustic Doppler current profiler (H-ADCP) has been horizontally deployed at a station near the city of Melak (Chapter 2). The H-ADCP profiles of velocity are converted to discharge adopting a new calibration methodology. The obtained time-series of discharge show the tidal signal is clearly visible during low flow conditions. Besides tidal signatures, the discharge series show influences by variable backwater effects from the lakes, tributaries and floodplain ponds. The discharge rate at the station exceeds 3250 m3s-1 with a hysteretic behaviour. For a specific river stage, the discharge range can be as high as 2000 m3s-1. Analysis of alternative types of rating curves shows this is far beyond what can be explained from kinematic wave dynamics. Apart from backwater effects, the large variation of discharge for a specified river stage can be explained by river-tide interaction, impacting discharge variation especially in the fortnightly frequency band.
A second H-ADCP station has been setup in the lower reach of the Mahakam, near the city of Samarinda, where the tidal discharge amplitude generally exceeds the discharge related to runoff (Chapter 3). Conventional rating curve techniques are inappropriate to model river discharge at this tidally influenced station. As an alternative, an artificial neural network (ANN) model is developed to investigate the degree to which tidal river discharge at Samarinda station can be predicted from an array of level gauge measurements along the tidal river, and from tidal level predictions at sea. The ANN-based model produces a good discharge estimation, as established from a consistent performance during both the training and the validation periods, showing the discharges can be predicted from water levels only, once that a trained model is available. The ANN models perform well in predicting discharges up to two days in advance.
Chapter 4 addresses the role of backwater effects and tidal influences on discharge time-series used to calibrate a rainfall-runoff model. The HBV rainfall-runoff model is implemented for the Mahakam sub-catchment upstream of Melak (25700 km2). In a first approach, the model is calibrated using a discharge series derived from the H-ADCP measurements from Melak station. In a second approach, discharge estimates derived from a rating curve are used to calibrate the model. Adopting the first approach, a comparatively low model efficiency is obtained, which is attributed to the backwater and tidal effects that are not captured in the model. The second approach produces a relatively higher model efficiency, since the rating curve filters the backwater effects out of the discharge series. Seasonal variation of terms in the water balance is not affected by the choice for one of the two calibration strategies, which shows that backwaters do not have a systematic seasonal effect on the river discharge.
To allow for investigation of the causes of backwater effects, satellite radar remote sensing is employed to monitor water levels in wetlands (Chapter 5). A series of Phased Array L-band Synthetic Aperture Radar (PALSAR) images is used to observe the dynamics of the Mahakam River floodplain. To analyze radar backscatter behavior for different land cover types, several regions of interest are selected, based on land cover classes. Medium shrub, high shrub, fern/grass, and degraded forest are found to be sensitive to flooding, whereas peat forest, riverine forest and tree plantation backscatter signatures only slightly change with flood inundation. An analysis of the relationship between radar backscatter and water levels is carried out. For lakes and shrub covered peatland, for which the range of water level variation is high, a good water level-backscatter correlation is obtained. In peat forest covered peatland, subject to a small range of water level variation, water level-backscatter correlations are poor, limiting the ability to obtain a floodplain-wide water surface topography from the radar images.
Chapter 6 continues to investigate the degree in which satellite radar remote sensing can serve to distinguish between dry areas and wetlands, which is a difficult task in densely vegetated areas such as peat domes. Flood extent and flood occurrence information are successfully extracted from a series of radar images of the middle Mahakam lowland area. A fully inundated region is easily recognized from a dark signature on radar images. Open water flood occurrence is mapped using a threshold value taken from radar backscatter of the permanently inundated areas. Radar backscatter intensity analysis of the vegetated floodplain area reveals consistently higher backscatter values, indicating flood inundation under forest canopy. Those observations are used to establish thresholds for flood occurrence mapping in the vegetated area. An all-encompassing flood occurrence map is obtained by combining the flood occurrence maps for areas with and without vegetation.
Chapter 7 synthesizes the findings from the previous chapters. It is concluded that the backwater effects and subtle tidal influences may prevent the option to predict river discharge using rating curves, which can best be interpreted as a stage-runoff relationship. H-ADCPs offer a promising alternative to monitor river discharge. For a tidal river, an ANN model can be used as a tool for data gap filling in an H-ADCP based discharge series, or even to derive discharge estimates solely from water levels and water level predictions. Discharge can be predicted several time-steps ahead, allowing water managers to take measures based on forecasts. The stage-runoff relationship derived from a continuous series of H-ADCP based discharge estimates may be expected to be much more accurate than a similar rating curve derived from a small number of boat surveys. The flood occurrence map derived from PALSAR images can offer a detailed insight into the hydroperiod, the period in which a soil area is waterlogged, and flood extent of the lowland area, illustrating the added value of radar remote sensing to wetland hydrological studies. In future work, radar-based floodplain observations may serve to calibrate hydrodynamic models simulating the processes of flooding and emptying of the lakes area.
|Qualification||Doctor of Philosophy|
|Award date||23 Oct 2013|
|Place of Publication||S.l.|
|Publication status||Published - 2013|