Analysis of the hydrological response of the small Hitiaa catchment belonging to Tahiti Island, French Polynesia

Abstract:

The volcanic Tahiti Island is prone to heavy rainfalls and flash floods which generate severe damages every year. Due to a complex geological story, the flood processes are still poorly known, so that high uncertainty remains when designing hydraulic structures. This study focuses on the high elevated and trade wind exposed Hitiaa catchment (1.37 km2) as a pilot site to identify the main factors of the hydrological response. A statistical analysis performed on 70 single-peak events showed runoff coefficients (Cr) ranging from 0.32 to 0.99. These events represented observed total cumulated rainfall (Pcum) from 45mm to 170mm and were associated with modal lag-time of 35 min and modal concentration time of 150 min. Cr mainly relied on cumulated rainfall (Pcum, partial correlation pcor = 0.58) and initial baseflow (Q0, pcor = 0.62). We were able to build a multilinear relation for which a R2 of 0.6 was obtained. We also tested different soil humidity proxies such as antecedent precipitation indexes and calculated soil water content from two continuous hydrological models (GR4J and SCS-LR), but their correlation with Cr did not lead to better results. Conceptual distributed event-based SCS-LR model was chosen in an effort to understand the Hitiaa watershed functioning. Its parameter S was found significantly anti-correlated with Q0 (partial correlation coefficient pcor = -0.55) but not influenced by Pcum (pcor = -0.08). In spite of S-optimization, rainfall specialization and flow calibration uncertainties, this result encourages studies towards the inference of S from humidity proxy. A next step will be to undertake a similar study on a downwind, less exposed catchment before investigating model regionalization.