Estimation of soil heat flux in a neotropical Wetland region using remote sensing techniques

• Autor(a) principal: Danelichen,Victor Hugo de Morais
• Data de Publicação: 2014
• Outros Autores: Biudes,Marcelo Sacardi, Souza,Maísa Caldas, Machado,Nadja Gomes, Silva,Bernardo Barbosa da, Nogueira,José de Souza
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Revista Brasileira de Meteorologia (Online)
• Texto Completo: http://old.scielo.br/scielo.php?script=sci_arttext&pid=S0102-77862014000400001
• topic: Mato Grosso; pasture; monodominant forest; orbital sensors

The direct estimation of the soil heat flux (G) by remote sensing data is not possible. For this, several models have been proposed empirically from the relation of G measures and biophysical parameters of various types of coverage or not vegetated in different places on earth. Thus, the objective of this study was to evaluate the relation between G/Rn ratio and biophysical variables obtained by satellite sensors and evaluate the parameterization of different models to estimate G spatially in three sites with different soil cover types. The net radiation (Rn) and G were measured directly in two pastures at Miranda Farm and Experimental Farm and and Monodominant Forest of Cambará. Rn, G, and G/Rn ratio and MODIS products, such as albedo (α), surface temperature (LST), vegetation index (NDVI) and leaf area index (LAI) varied seasonally at all sites and inter-sites. The sites were different from each other by presenting different relation between measures of Rn, G and G/Rn ratio and biophysical parameters. Among the original models, the model proposed by Bastiaanssen (1995) showed the best performance with r = 0.76, d = 0.95, MAE = 5.70 W m-2 and RMSE = 33.68 W m-2. As the reparameterized models, correlation coefficients had no significant change, but the coefficient Willmott (d) increased and the MAE and RMSE had a small decrease.