Brown, Katherine

advertisement
ANALYSING TRENDS IN CLIMATE PROJECTIONS FOR
MAXIMUM SUB-DAILY PRECIPITATION ACROSS TASMANIA
MONTHLY
Katherine Brown,1
1
Pattle Delamore Partners Ltd
Aims
Climate change impacts on the distribution of sub-daily precipitation will affect flooding and the
future performance of stormwater assets and other infrastructure, particularly in ‘flashy’ urban
catchments. The focus of research in Tasmania to date has largely been on daily precipitation
data. Tasmania has many natural and regulated catchments with critical durations ranging from a
few minutes, to at most a few hours (Willems et al., 2012). There is a need for greater
understanding of climate change impacts on precipitation data with greater temporal and spatial
resolution. The design and ongoing operations and maintenance of assets, particularly in an
urban context, places special requirements on climate change impact studies and the data that
informs them (Willems et al., 2012, Arnbjerg-Nielson et al., 2013).
Trend analysis has been undertaken to ascertain climate change effects at a local level on 3hourly precipitation data extracted from six dynamically downscaled global circulation models
(GCMs). Preliminary conclusions have been drawn as to the impact of changes in sub-daily
precipitation resulting from climate change over the period 1961 to 2100 across the entirety of the
state of Tasmania (ACE CRC, 2010).
Method
Only a limited number of GCMs and emissions scenarios could be evaluated due to time
constraints. Six GCMs for the A2 SRES emissions scenario were evaluated as a part of this
study. Climate models have been selected based on their skill at simulating the present climate
as this is a useful indicator as to GCM performance (ACE CRC, 2010). The six GCMs selected
for this study were chosen by the ACE CRC (2010) based on their ability to reproduce presentday precipitation means and variability across Australia. The six GCMs that have informed this
study are CSIRO-Mk3.5, GFDL-CM2.0, GFDL-CM2.1, ECHAM5, MIROC3.2(medres), and
UKMO_HadCM3.
GCMs have a spatial resolution of between 200 and 300 km along their horizontal and vertical
axes. At this resolution, GCMs poorly simulate the highly heterogeneous climate conditions
across Tasmania (Brown, Bennett, Parkyn & Link, 2010). It is imperative that GCMs be
downscaled to improve the spatial resolution of the results and to capture localised weather
patterns and topography.
Data used in this study has been quality checked to ensure that the simulated spatial and
temporal patterns are as expected. Where observed precipitation data is available, the observed
precipitation record has been compared with the simulated record for the period 1961-1990. The
ACE CRC has thoroughly checked the 24-hour precipitation data that is comprised of the threehourly dataset used in this study. Checks undertaken by the ACE CRC (2010) included:
comparing downscaled model outputs with raw outputs from the global climate model to ensure
consistency against large-scale climate variables; comparing downscaled simulated data to the
observed data; checking internal consistency of climate drivers to ensure large-scale pressure
patterns and ocean effects are correct; examining climate change drivers to ensure consistency
with current understanding of climate change; and using climate modelling outputs as inputs to a
range of biophysical and hydrological models.
Removing serial correlation from data is an important step in trend analysis to ensure that results
are reliable and not unduly impacted by serial correlation contained within the data. Pre-whitening
was used to remove serial correlation from the dataset. ARIMA models were fitted to the
ECHAM5 sub-daily monthly maxima precipitation series to remove serial correlation and seasonal
trends. This fitted ARIMA model was then applied to the remaining five GCMs to produce 36 prewhitened datasets. LOWESS and kernel smoothing functions were applied to the pre-whitened
dataset. Kernel smoothing and LOWESS smoothing techniques have been used to test for the
presence of step-trends in the dataset. Agreement between the six GCMs has been analysed to
allow for discussion of uncertainty between climate models.
Results
There is a general agreement at the regional scale between the six GCMs as to where
decreasing and increasing trends are likely to occur in the intensity of three-hourly monthly
maximum precipitation. Figure 1 demonstrates the agreement between GCMs in predicting trends
in sub-daily precipitation data for kernel and LOWESS smoothed sub-daily monthly maximum
precipitation.
Figure 2 below shows percent change in the intensity of sub-daily precipitation across the state.
Increasing trends are observed on the west and east coast of the state, whereas declining trends
are observed in the central plateau region of Tasmania. Changes in the distribution of
precipitation will result in ecological, social and economic challenges for Tasmania. Where
increasing trends in sub-daily monthly maximum precipitation are projected it is important to
evaluate how increased flood risk and associated water quality impacts can be managed into the
future.
Future analysis using other emissions scenarios will likely improve confidence in the results and
allow for a greater understanding of changes to the spatial and temporal distribution of
precipitation across the state. Future work should aim to link these results to average recurrence
intervals. This would improve the accessibility of the results and enable them to be readily used to
optimise the design and management of assets sensitive to changes in the intensity, frequency
and duration of short duration precipitation events.
Figure 1 Agreement between GCMs as to the presence of an increasing or decreasing trend for the
LOWESS (upper row) and kernel smoothed (lower row) mean sub-daily precipitation with durations ranging
from 3 hours to 24 hours between the baseline period and the end of the 21 st century. The colour bar ranges
from zero to six, where six indicates that all six GCMs have predicted a negative step-trend and zero
indicates that no GCM has predicted a negative trend. The red and blue dots are key sites around the state.
(a)
(b)
(c)
(d)
(e)
(f)
Figure 2 Multi-model mean percent change in the intensity of the kernel and LOWESS smoothed 5th, 50th
and 95th percentile 3-hourly monthly maximum precipitation between the baseline period (1961-1990) and
the end of the 21st century (2070-2099). Mapped results are as follows: a) LOWESS Smoothed 5th
Percentile b) LOWESS Smoothed Mean c) LOWESS Smoothed 95% Percentile d) Kernel Smoothed 5 th
Percentile e) Kernel Smoothed Mean f) Kernel Smoothed 95% percentile.
References
ACE CRC. (2010). Climate Futures for Tasmania climate modelling: the summary. Antarctic Climate and
Ecosystems Cooperative Research Centre. Hobart: Antarctic Climate and Ecosystems Cooperative
Research Centre.
Arnbjerg-Nielson, K., Willems, P., Olsson, J., Beecham, S., Pathirana, A., Bulow Gregersen, I., et al. (2013).
Impacts of climate change on rainfall extremes and urban drainage systems: a review. Water Science and
Technology , 68 (1), 16-28.
Bennett, J. C., Ling, F. L., Graham, B., Grose MR, M. R., Corney, S. P., White, C. J., et al. (2010). Climate
Futures for Tasmania: water and catchments technical report. Antarctic Climate & Ecosystems Cooperative
Research Centre, Climate Futures for Tasmania. Hobart: ACE-CRC.
Brown, K., Bennett, J., Parkyn, R., & Ling, F. (2010). Translating climate projections into usable information
for business. A case study from Tasmania. In E. Australia (Ed.), Climate Change 2010: Practical Responses
to Climate Change (pp. 262-271). Melbourne: Engineers Australia.
Willems, P., Olsson, J., Arnbjerg-Nielsen, K., Beecham, S., Pathirana, A., Bulow Gregersen, I., et al. (2012).
Impact of Climate Change on Rainfall Extremes and Urban Drainage Systems. IWA Publishing.
Download