Determination of instantaneous and daily net radiation from TM – Landsat 5 data in a subtropical watershed

• The short-wave radiation model used presented good accuracy.
• It was possible to determine net radiation in heterogeneous area with satellite data.
• Rn in sugarcane plantations impact local climate, since they present higher Ts.
• The Rn provides valuable information to explain land use changes.
• The results obtained will support the determination of evapotranspiration in the area.

Remote sensing makes it possible to identify the changes that occur on the surface of the Earth as a result of natural and/or man-made phenomena. Such changes impact on the net radiation at surface which in turn controls the Earth's climate. The present study aims to determine the impact of land use changes on net radiation at surface in a tropical watershed in Brazil, based on satellite images. The instantaneous net radiation (Rn,ins) (at the time of the satellite overpass) and the daily net radiation (Rn,24 h) were both estimated by TM – Landsat 5 images and complementary weather data. The net radiation (Rn) estimated from remote sensing data was compared to the measurements taken from two micrometeorological towers located in the study area. Most Rn,ins values were found to be between 457.4 W m−2 and 760.0 W m−2 during the months with more intense solar radiation (February, March, and November), especially in the areas with more vegetation cover (sugarcane and eucalyptus plantations and areas with woody savanna vegetation, locally called Cerradão). The months with the highest thermal and radiative contrast (June and November) were selected to show the spatial distribution of the daily (Instantaneous) Rn, which ranged from 28.0 (420) W m−2 to 98.0 (520) W m−2 in June and from 83.0 (450) W m−2 to 264.0 (800) W m−2 in November 9. The model used to calculate Rn,24 h provided values close to those taken at surface, even on days with higher cloud cover after the satellite overpass. The Mean Absolute Error (MAE), Mean Relative Error (MRE), and Root Mean Square Error (RMSE) associated with the Rn,24 h computations in the sugar cane plantation were 8.3 W m−2, 8.4%, and 10.4 W m−2, respectively, confirming the applicability and accuracy of the results. The Rn patterns registered on the woody savanna throughout the year differ very much from those found in cropped areas, particularly in sugar cane plots. This may cause an impact on the watershed climate.