PhD Defense

Precipitation results from atmospheric phenomena, which are characterized by a large space and time variability. The distribution of rainfall, in particular of strong rainy events, has various impacts in surface hydrology over the different regions in the world (e.g. floods). Any change in the Tropical climate is associated with a modification of the water and energy cycle over those regions. Therefore, in a context of climate change, it is important to develop new tools able to provide quantitative precipitation measurements, both over land and over the open oceans. The work presented hereafter deals with precipitation estimation from space. Indeed, measuring rainfall requires a high density of observations, which, over the whole tropical belt, can only be provided from space. For several decades, the availability of satellite observations has greatly increased and offers an increasing number of measurements. Thanks to newly implemented missions like the Megha-Tropiques mission and the forthcoming GPM constellation (Global Precipitation Measurement mission), measurements from space become available from a set of observing systems. Quantitative precipitation estimation were only available at the monthly scale, it is now possible to estimate rainfall from space at increasingly fine scale. In this work, we focus on the 1°/1-day scale, key scale of meteorological and hydrological studies.

Various methods exist to estimate rainfall from space but they provide estimates of unequal quality. First, a meteorological benchmark is set up with ground-based observations from the African Monsoon Multidisciplinary Analysis (AMMA) program. The analysis shows that the last generation of combined infrared-microwave products is describing the variability of rainfall similarly to ground measurements at meteorologically relevant scales. It also appeared that at these scales, rain accumulation estimations should be used taking into account their uncertainties. A novel methodology for quantitative precipitation estimation is introduced; its name is TAPEER (Tropical Amount of Precipitation with an Estimate of ERrors) and it aims to provide 1°/1-day rain accumulations and associated errors over the whole Tropical belt. This approach is based on a combination of infrared imagery from a fleet of geostationary satellite and passive microwave derived rain rates from a constellation of low earth orbiting satellites. Modelling techniques are developed in order to associate an error with the individual rain accumulations. An investigation of the error budget of the TAPEER method shows that the two main contributions to the total error are related to sampling and systematic errors on rain rates of medium intensity.

A study on the summer 2009 period reveals the importance of using error bars when analyzing the distribution of rainfall, especially for the most important rain accumulations of the tropics.