**1. Introduction**

The Atlantic Rainforest stretches across Brazil, Argentina, and Paraguay, and is considered the second largest tropical forest in the American continent [1]. In Brazil, the Atlantic Rainforest covers 17 states, from Ceará to Rio Grande do Sul, and is located along coastal and inland regions, over mountains and plateaus [2]. The great longitudinal and latitudinal extension and, consequently, the wide variability in rainfall and temperature provide, combined with altitude gradient and ancient vegetation formations, a high degree of biodiversity and endemism [3].

The Atlantic Rainforest was heavily devastated since the beginning of Brazilian coloniza‐ tion, mainly due to wood extraction and rural and urban settlement. Thus, from the original forest cover, only isolated remnants with variable sizes in different successional stages were left [4]. Recent studies show that only 8.5% of the original Atlantic Rainforest, estimated in 1.3 million km2 , still exists [5]. The state of São Paulo (henceforth referred to as SP), despite its high levels of agricultural and urban development, presents the largest remnants of At‐ lantic Rainforest in the country. Estimates show that 13.9% of the original Atlantic Rainfor‐ est still exists in SP [6].

Although Atlantic Rainforest has been reduced and fragmented, its remnants are an impor‐ tant source of heat and water vapor to the atmosphere. This is because latent heat released as evapotranspiration influences the atmospheric circulation in the tropics and the water va‐ por contributes to the regional precipitation regime [7]. In this context, knowing the annual and interannual variability of precipitation and evaporative processes in tropical biomes is necessary for a better understanding of the energy and water partitioning between surface and atmosphere, which allows for a better parameterization of the boundary layer processes used in climate and weather forecasting models [8, 9].

Usually, precipitation and evapotranspiration are measured by instruments equipped in conventional meteorological stations; however, these measurements are expensive and do not represent well the spatial variability of these processes [10]. Hence, the use of remote sensing techniques becomes a methodological alternative since it enables to obtain different biophys‐ ical parameters at the Earth's surface with high temporal and spatial coverage. The Tropical Rainfall Measuring Mission (TRMM) [11] and the MOD16 algorithm [12, 13], developed, respectively, to estimate global surface precipitation and evapotranspiration, have been widely used by the scientific community in large‐scale hydrological studies [14].

The purpose of this study was to characterize and analyze, based on both TRMM and MOD16 imagery, the temporal dynamics of precipitation and evapotranspiration in the Atlantic Rainforest remnants of SP, southeastern Brazil, during a 10‐year period (January 2000 to December 2010). We have implemented a wavelet transform to evaluate the temporal varia‐ bility of these parameters. Wavelet analysis is becoming a common tool for researches involving remote sensing and land‐atmosphere interactions. It provides an efficient method for extracting relevant information from large datasets and has been applied to a wide range of variables and different types of ecosystems [15].
