**2.4. Atlantic rainforest remnants**

provide information about the integrated content of the precipitation column, intensity and type of precipitation. The VIRS sensor, derived from the AVHRR/NOAA sensor, has five spectral bands in visible and infrared regions performing observations of clouds, such as cover,

For this study, monthly precipitation data derived from TRMM (3B43 product) version 7 (v7) were used, covering the period from January 2000 to December 2010. The 3B43 prod‐ uct is calculated using data from multiple satellites, in addition to TRMM, as well as me‐ teorological stations data from the Global Precipitation Climatological Center (GPCC) and the Climate Assessment and Monitoring System (CAMS) [19]. 3B43 imagery were ac‐ quired from the electronic address (http://gdata1.sci.gsfc.nasa.gov/daac‐bin/G3/gui.cgi?in‐ stance\_id=TRMM\_Monthly), presenting spatial resolution of ~30 km in mm month−1. Images were processed in ENVI version 4.5, where the steps of reprojecting, resampling of pixels to 1 km (same spatial resolution of MOD16 data) using nearest neighbor meth‐

The MOD16 algorithm [12, 13] was developed in the context of the Earth Observing System/ NASA (EOS/NASA) program, aiming to estimate global evapotranspiration using data from Moderate Resolution Imaging Spectroradiometer (MODIS) sensor (Terra and Aqua) and meteorological data from Global Modeling and Assimilation Office (GMAO). In general terms, MOD16 is a revision of the algorithm proposed by [20], who adapted the Penman‐Monteith

> ( ) ( ) sat / 1 / *p a s a*

+ - <sup>=</sup> + + (1)

of vegetated global areas. Among the products

*sA C e e r <sup>E</sup> s rr* r

where *λE* is the latent heat flux (W m−2) and *λ* represents the latent heat of evaporation (J kg−1), *s* = *d*(*e*sat)/*dT* is the slope of the curve which relates saturated pressure of water (*e*sat) and temperature (Pa K−1), *A* is the energy available at surface (W m−2), *ρ* represents air density (kg m −3), *Cp* is the specific heat of air (J kg−1 K−1), *e* is the real pressure of water vapor (Pa), *rs* is the surface resistance, *ra* is the aerodynamic resistance (s m−1), and *γ* represents the psychrometric

MODIS input data required for MOD16 algorithm have spatial resolution between 500 m and 1 km, and include global products of land use and land cover (MOD12Q1), leaf area index (LAI) and photosynthetically active radiation (PAR‐MOD15A2), and albedo (MCD43B2). Regarding the meteorological parameters required for the algorithm, daily reanalysis data of GMAO referring to incident solar radiation, air temperature, and water vapor pressure, with spatial resolution of 1.00° × 1.25°, are used [13, 21]. In summary, MOD16 data have a spatial

generated, we highlight the potential and actual evapotranspiration and potential and actual latent heat flux products, in intervals of eight (MOD16A2) and 30 days (MOD16A3) [22].

g

od, and clipping to Atlantic Rainforest remnants in SP were performed.

96 Tropical Forests - The Challenges of Maintaining Ecosystem Services while Managing the Landscape

equation (Eq. (1)) to be used with remote sensing data:

resolution of 1 km and covers ~109 million km2

l

type, and top temperature [18].

**2.3. MOD16 data**

constant (66 Pa K−1).

Since the 1980s, the SOS Mata Atlântica Foundation, jointly with the National Institute for Space Research (INPE), is regularly mapping forest cover in the Atlantic Rainforest biome. These institutions use remote sensing imagery to produce the "Atlas of Forest Remnants of Atlantic Rainforest." Resulting maps enable us to determine the spatial distribution of forest remnants and ecosystems associated to Atlantic Rainforest, keep track of changes in vegetation cover, and generate permanently improved and updated information of this biome [23, 24].

The spatial distribution of the forest remnants was obtained from the database provided by [5]. This database was used to update the period corresponding to 2011–2012, that is, to up‐ date changes occurred in the polygons previously classified as forest fragments (forest rem‐ nants, mangrove, or restinga) in previous versions of the "Atlas." To this end, images of the LISS III/RESOURCESAT‐1 orbital sensor corresponding to the second semester of 2012 were used. Vector files of the forest fragments polygons were acquired from the electronic ad‐ dress http://mapas.sosma.org.br/ and edited in ARCGIS version 9.3. The editing process consisted in selecting polygons of forest remnants with area equal or greater than 100 ha. Selecting only ≥100 ha polygons intended to create a spatial homogeneity of the analyzed areas. It should be noted that the study comprehends the period between 2000 and 2010 and it is understood that the forest remnants mapped by [5], referring to the update of 2011– 2012, are representative of the period analyzed.
