**4. Conclusions**

Combining TRMM satellite data and MOD16 algorithm enabled mapping the spatial distri‐ bution and evaluating precipitation and evapotranspiration in São Paulo State, as well as analyzing the temporal dynamics of these variables in Atlantic Rainforest remnants for the period between 2000 and 2010. Generally, the precipitation and evapotranspiration trends (considering both São Paulo State and forest remnants) revealed a strong seasonal pattern, with highest values concentrated in the rainy season (October to March) and lowest values in the dry season (April to September).

Regarding to São Paulo State, highest values of precipitation and evapotranspiration were found in southern and eastern regions, while lowest values were located in the northern and western portions of the state. The time series analysis showed that monthly averages for precipitation and evapotranspiration were, respectively, ~200 and ~84% higher during the rainy season when compared to the dry season. Considering annual averages, evapotranspi‐ ration corresponded to ~53% of precipitation in São Paulo State.

In regard to Atlantic Rainforest remnants, time series analysis showed that during the rainy season precipitation and evapotranspiration were, respectively, ~138 and ~65% higher than those observed during the dry season. In terms of annual averages, evapotranspira‐ tion accounted for ~72% of precipitation, indicating a low hydric production (~28%). Con‐ sidering the entire period, monthly averages of precipitation and evapotranspiration were, respectively, ~12 and ~52% higher than the monthly averages for São Paulo State, which demonstrates the contribution of these remnants to the regional hydrologic regime. The higher amounts of precipitation are observed in the coastal region where most of the rem‐ nants are located and maybe there is an artifact, however if we analyze the evapotranspi‐ ration maps it is possible to note that the evapotranspiration is really low in the western part of Sao Paulo state due the presence of large areas of agriculture and pasture and a reduced number of forest remnants. Analysis of wavelet transform for precipitation and evapotranspiration time series in Atlantic Forest remnants showed that the main oscilla‐ tion mode is concentrated between 8 and 16 months, revealing a seasonal or intra‐annual behavior. It is important to note that the wavelets analysis allowed to conduct a more com‐ prehensive evaluation of the behavior of precipitation and evapotranspiration through time. It can be a useful tool to verify trends of temporal shifts in environmental parameters [48–53], which in its turn can affect the ecosystem services delivered by tropical forest rem‐ nants. In this sense, just to point out, the trends of temporal shifts in precipitation and evapotranspiration observed in our study were related, in general, to the SACZ, SALLJ, and El Niño.

Results found in this study demonstrated that the use of remote sensing was an important tool for analyzing hydrological variables in Atlantic Rainforest remnants, which can contribute to better understanding the interaction between tropical forests and the atmosphere, and for generating input data necessary for surface models coupled to atmospheric general circulation models. Accordingly, future studies should be performed to (i) validate MOD16 algorithm for Atlantic Rainforest conditions, (ii) analyze potential artifacts related to the spatial distribution of the land cover and environmental parameters, (iii) identify other phenomena that could be related to intraseasonal and interannual variations in precipitation and evapotranspiration occurred in Atlantic Rainforest remnants, (iv) analyze precipitation and evapotranspiration in specific forest formations of Atlantic Rainforest (e.g., dense ombrophylous forest, mixed ombrophylous forest, and seasonal semideciduous forest), (v) evaluate the differences of precipitation and evapotranspiration between forest remnants and different land use types (e.g., pasture, agriculture, urban areas, etc.), which can play an important role to understand more specifically what is the impact of land use changes in ecosystem services in tropical regions, and (vi) analyze the relationship between precipitation and evapotranspiration in Atlantic Rainforest remnants using other biophysical variables, such as surface albedo and vegetation indices.
