Reconstruction of thermal local signal from statistical downscaling (sd) through artificial neuronal network: detection of local patterns of change in Valencia Region (Spain) (1948-2011)

Abstract

For the area of Valencia (Spain) is available more than 300 daily temperature series of
which only three cover completely the period from 1948 to 2011, and a majority less than
half of that period. This work seeks to reconstruct all missing local thermal signals for the
whole period 1948-2011 from the NCEP/NCAR reanalysis availability in that period. To do
this, a statistical downscaling is performed with the observed data available through an ANN.
This allowed us to detect local patterns of climate change that have been consistent over time
and space. These patterns indicate substantially a greater warming tendency in the upper
parts of the reliefs and mountain areas, and lower in the valley bottoms and lower lands. The
value of the study lies primarily in identifying areas potentially more vulnerable to thermal
change, rather than the accurate determination of a global magnitude of thermal change.