13.3 Data and Methods
Different types of datasets have been used to accomplish the objectives of this work:
• For studying climate trends and variability, we used the Spain02 database (Herrera
et al. 2012), which is a regular 0.2º latitude/longitude (approximately 20 km) daily
gridded dataset containing precipitation, maximum and minimum temperatures
covering continental Spain and the Balearic Islands during the period 1950–2007.
From this, we extracted only pixels located in the Pyrenees range, with an elevation
(>1200 m a.s.l.) that ensured the specific climate characteristic of mountains to be
captured. A validation of Spain02 against observations shows that temperatures are
positively biased (because grid points are generally at lower elevations than the
AEMET (Agencia Estatal de Meteorología—the Spanish Meteorological Agency)
stations selected for validation). However, this does not affect the computation of
trends since the temporal variability of temperature and precipitation is well represented by the gridded dataset (R
2 > 0.6).
• Snow thickness measures were provided by the ERHIN program (Evaluación de
los Recursos Hídricos procedentes de la Innivación—assessment of water
resources from snow) of the Spanish Ministry of Environment. The ERHIN
program has been taking snow measurements in the Spanish mountains since
mid-1980 with fixed snow poles; three measurements were carried out for every
year: the first in late January or early February, the second in March and the
third in mid-April or early May. Quality criteria, including a maximum number
of three data gaps were set up to obtain reliable snow data series. The data
period was set from 1986 (first year of snow sampling) to 2007 (to make it
coincide with the last year of the climatic series). Some of the selected series still
had few data gaps, and only those with less than 15% of missing data were filled
using the results of linear regressions with the best-correlated series (i.e. those
where R
2 > 0.7). A total of 84 series of snow depth in April–May, covering
most of the Pyrenees area, were finally selected.
• Daily streamflow data was collected from the national water agency of Spain,
Centro de Estudios Hidrográficos (CEDEX, http://hercules.cedex.es/anuarioaforos/
default.asp). To make sure that snowmelt pulses were present in all river regimes,
we selected only rivers located in the foothills of mountain systems whose drainage
watersheds had a mean elevation exceeding 800 m.a.s.l., and had no presence of the
reservoirs or impoundment systems upstream of the gauge station. Data from seven
gauge stations corresponding to six rivers were finally selected.
Some climatic and hydrological indices were computed in order to capture the
conditions that can favour snow accumulation and melt and their signal in the
hydrographs.
• The daily resolution of Spain02 database enabled the behaviour of far-from-normal
events to be investigated. We computed a series of indices that combine moisture
and heat conditions based on the tails of the frequency distributions of daily
precipitation and mean temperature series. Based on the frequency distributions of
13 Changes in Climate, Snow and Water Resources in the Spanish …
309
Different types of datasets have been used to accomplish the objectives of this work:
• For studying climate trends and variability, we used the Spain02 database (Herrera
et al. 2012), which is a regular 0.2º latitude/longitude (approximately 20 km) daily
gridded dataset containing precipitation, maximum and minimum temperatures
covering continental Spain and the Balearic Islands during the period 1950–2007.
From this, we extracted only pixels located in the Pyrenees range, with an elevation
(>1200 m a.s.l.) that ensured the specific climate characteristic of mountains to be
captured. A validation of Spain02 against observations shows that temperatures are
positively biased (because grid points are generally at lower elevations than the
AEMET (Agencia Estatal de Meteorología—the Spanish Meteorological Agency)
stations selected for validation). However, this does not affect the computation of
trends since the temporal variability of temperature and precipitation is well represented by the gridded dataset (R
2 > 0.6).
• Snow thickness measures were provided by the ERHIN program (Evaluación de
los Recursos Hídricos procedentes de la Innivación—assessment of water
resources from snow) of the Spanish Ministry of Environment. The ERHIN
program has been taking snow measurements in the Spanish mountains since
mid-1980 with fixed snow poles; three measurements were carried out for every
year: the first in late January or early February, the second in March and the
third in mid-April or early May. Quality criteria, including a maximum number
of three data gaps were set up to obtain reliable snow data series. The data
period was set from 1986 (first year of snow sampling) to 2007 (to make it
coincide with the last year of the climatic series). Some of the selected series still
had few data gaps, and only those with less than 15% of missing data were filled
using the results of linear regressions with the best-correlated series (i.e. those
where R
2 > 0.7). A total of 84 series of snow depth in April–May, covering
most of the Pyrenees area, were finally selected.
• Daily streamflow data was collected from the national water agency of Spain,
Centro de Estudios Hidrográficos (CEDEX, http://hercules.cedex.es/anuarioaforos/
default.asp). To make sure that snowmelt pulses were present in all river regimes,
we selected only rivers located in the foothills of mountain systems whose drainage
watersheds had a mean elevation exceeding 800 m.a.s.l., and had no presence of the
reservoirs or impoundment systems upstream of the gauge station. Data from seven
gauge stations corresponding to six rivers were finally selected.
Some climatic and hydrological indices were computed in order to capture the
conditions that can favour snow accumulation and melt and their signal in the
hydrographs.
• The daily resolution of Spain02 database enabled the behaviour of far-from-normal
events to be investigated. We computed a series of indices that combine moisture
and heat conditions based on the tails of the frequency distributions of daily
precipitation and mean temperature series. Based on the frequency distributions of
13 Changes in Climate, Snow and Water Resources in the Spanish …
309