Abstract:
Tropical Mountain Forest (TMF) provides important ecological func-
tions like evapotranspiration (ET) that supplies moisture and energy
to the atmosphere. ET observations are scarce and difficult to
accomplish particularly in areas of high heterogeneity where TMF
are. Remote sensing (RS) allows to quantify and to determine ET
spatial variation at the landscape level. Detail imaginary improves
high spatial variability retrieval. Thought the greater detail intro-
duces cast shadows by trees which hamper image interpretation.
The objective of this study is to characterize ET estimation for the
TMF of the southern Ecuadorian Andes by combining meteorologi-
cal data with high-resolution satellite images. Shadows from high
resolution images were masked out by applying focal statistics. The
analysis included two meteorological periods typical of the area;
a wet period when rain prevails and a dry period when precipitation
is more sporadic. The reference evapotranspiration (ET0) was calcu-
lated using the FAO-Penman Montheid method by applying data
obtained from an automatic weather station. The enhanced vege-
tation index (EVI) was derived from 2 m resolution WorldView2
satellite images. Results showed a lower ET mean value during the
wet period: 1.54 mm day−1
compared to 2.37 mm day−1
. Two forest
types, differentiated from its structural composition and topogra-
phical position (ravine and ridge), marked ET spatial variation.
Ravine forest that has a more dense and closed canopy showed
higher ET values for both meteorological conditions. A comparison
between ET estimations and ET field measurements from
a scintillometer device showed a good agreement (coefficient of
correlation r = 0.89) that proves the validity of the method. This
study demonstrates that the application of high spatial resolution
improves ET estimation in TMF especially when shadows are
removed. Also, emphasizes the importance of analysing spatial
heterogeneity to properly assess ecosystem water flux terms.