234
5 Discussions and Conclusions
Climate change and consequently warming of high altitude areas has increased the
vulnerability of cryosphere owing its sensitivity towards temperature fluctuations.
However, the observed trends in climate are not always concordant with observed
cryosphere changes owing to point based information on climate and the points in
turn are quite clustered/sparse in high elevated areas. On the other hand, the cryospheric changes are monitored/calculated by use of remote sensing products which
provide a wide spatial coverage but low temporal coverage. In backdrop of above
mentioned facts, gridded datasets are seen as a boon to climatologists who seek
information with good temporal and spatial coverage. This study attempted to evaluate the performance of various gridded datasets in complex areas like NWH where
stark variations in climate can be observed at short distances. Though such attempts
have earlier been made by many researchers (Andermann et al. 2011; Palazzi et al.
2013; Dahri et al. 2016; Hussain et al. 2017 etc.) worldwide but they also reported
that due to lack of observations for areas lying above 5000 m, these areas remain
unexplored. The novelty of present study lies in the validation of datasets from
observations for areas above 5000 m also. Apart from dataset evaluation, we studied
the spatial variability and temporal trends of climatic patterns as captured by
selected datasets.
We evaluated the performance of 08 datasets, i.e. APHRODITE, NCEP-NCAR,
TRMM, ERA-I, CRU-TS, GPCC, GPCP and UDEL. APHRODITE, CRU-TS,
GPCC and UDEL are interpolated datasets with good resolution. APHRODITE has
two limitations, i.e. lack of observations above 5000 m (Kumar et al. 2015) and
ceased temporal coverage after year 2007; due to which its use in hydrological/climatological studies is constrained. The performance of UDEL was not much satisfactory in many ways. GPCC dataset has only precipitation data but no temperature
data is there. On the other hand, satellite based products like TRMM despite having
good spatial resolution lacks in providing long temporal coverage, i.e. datasets are
available from year 1998 only. Merged datasets like GPCP have a limitation of very
coarse spatial resolution (2.5°). Reanalysis dataset like NCEP-NCAR also has a
very coarse spatial resolution which makes them unsuitable for Himalayan terrain.
Thus, two datasets i.e. ERA-I and CRU-TS were selected to study spatial variability
and temporal trends of temperature and precipitation over NWH. Since the fidelity
of datasets counts on their spatial resolution, thus CRU-TS was found to be handicapped compared to ERA-I in many terms.
ERA-I and CRU-TS could effectively capture the amount and trends of wintertime temperature and precipitation but with inherent biases which suggest the need
for bias correction of datasets. Spatial distribution of temperature and precipitation
was also well captured by these trends. Long term (1985–2015) trends by both datasets reveal significant warming by varying rates which is in phase with observations.
However, ERA-I showed maximum rate of warming for LH followed by GH and
KH. LH has undergone maximum urbanization during recent decades because of its
accessibility as compared to GH and KH. Increased urbanization/population/tourH. S. Negi and N. Kanda
5 Discussions and Conclusions
Climate change and consequently warming of high altitude areas has increased the
vulnerability of cryosphere owing its sensitivity towards temperature fluctuations.
However, the observed trends in climate are not always concordant with observed
cryosphere changes owing to point based information on climate and the points in
turn are quite clustered/sparse in high elevated areas. On the other hand, the cryospheric changes are monitored/calculated by use of remote sensing products which
provide a wide spatial coverage but low temporal coverage. In backdrop of above
mentioned facts, gridded datasets are seen as a boon to climatologists who seek
information with good temporal and spatial coverage. This study attempted to evaluate the performance of various gridded datasets in complex areas like NWH where
stark variations in climate can be observed at short distances. Though such attempts
have earlier been made by many researchers (Andermann et al. 2011; Palazzi et al.
2013; Dahri et al. 2016; Hussain et al. 2017 etc.) worldwide but they also reported
that due to lack of observations for areas lying above 5000 m, these areas remain
unexplored. The novelty of present study lies in the validation of datasets from
observations for areas above 5000 m also. Apart from dataset evaluation, we studied
the spatial variability and temporal trends of climatic patterns as captured by
selected datasets.
We evaluated the performance of 08 datasets, i.e. APHRODITE, NCEP-NCAR,
TRMM, ERA-I, CRU-TS, GPCC, GPCP and UDEL. APHRODITE, CRU-TS,
GPCC and UDEL are interpolated datasets with good resolution. APHRODITE has
two limitations, i.e. lack of observations above 5000 m (Kumar et al. 2015) and
ceased temporal coverage after year 2007; due to which its use in hydrological/climatological studies is constrained. The performance of UDEL was not much satisfactory in many ways. GPCC dataset has only precipitation data but no temperature
data is there. On the other hand, satellite based products like TRMM despite having
good spatial resolution lacks in providing long temporal coverage, i.e. datasets are
available from year 1998 only. Merged datasets like GPCP have a limitation of very
coarse spatial resolution (2.5°). Reanalysis dataset like NCEP-NCAR also has a
very coarse spatial resolution which makes them unsuitable for Himalayan terrain.
Thus, two datasets i.e. ERA-I and CRU-TS were selected to study spatial variability
and temporal trends of temperature and precipitation over NWH. Since the fidelity
of datasets counts on their spatial resolution, thus CRU-TS was found to be handicapped compared to ERA-I in many terms.
ERA-I and CRU-TS could effectively capture the amount and trends of wintertime temperature and precipitation but with inherent biases which suggest the need
for bias correction of datasets. Spatial distribution of temperature and precipitation
was also well captured by these trends. Long term (1985–2015) trends by both datasets reveal significant warming by varying rates which is in phase with observations.
However, ERA-I showed maximum rate of warming for LH followed by GH and
KH. LH has undergone maximum urbanization during recent decades because of its
accessibility as compared to GH and KH. Increased urbanization/population/tourH. S. Negi and N. Kanda