Possibilities of long range drought forecasting using Standardized
Possibilities of long range drought forecasting using Standardized Precipitation Index Lovro Kalin, Domagoj Mihajlović, Ksenija Cindrić Kalin, Blazenka Matjačić
outline • • • Drought - Introduction and motivation Seasonal forecasts – ECMWF SPI - Data and methods Verification Future plans
drought • highest economic losses (39%) among all hydro-meteorological events in Croatia • mainly initiated by a precipitation deficit at short time scales – 1 or 2 months - meteorological drought – 3 to 6 months – agricultural – longer than 6 months - hydrological
drought - trends • trend analysis of dry spells during the 19612010 period revealed an increase in their duration (Cindrić et al. 2013) • in line with the trend of precipitation extremes (Gajić-Čapka et al. 2014) • Fifth Assessment Report (AR 5) of the IPCC increased number of extreme events (IPCC, 2013)
drought monitoring
drought monitoring - SPI
SPI • SPI - Standardized Precipitation Index (Mc. Kee et al. 1993) • measure of meteorological drought, based on precipitation amount only • gamma dist. to normal -2, -1, 0, 1, 2 • a forecasting component? • ECMWF seasonal forecast (monthly anom. ) • probability = fraction of ensemble members that predict a desired event (SPI<-1)
ECMWF seasonal forecast system • run every 2 weeks • 7 months forecast range • consists of an ocean analysis and a global coupled ocean-atmosphere general circulation model • 51 members of the ensemble
ECMWF seasonal forecast system
ECMWF seasonal forecast system
SPI forecast • different combinations of observed and forecasted data • SPI 1 • SPI 3 comb=2 months obs. +1 month forec. • SPI 3+3. . .
final product
final product
final product
Drought prediction SPI 1 < -1 BS=. 151 BSS=. 265 - good reliability - some overcofidence - decent sharpness
Drought prediction SPI 3 comb < -1 BS=. 11 BSS=. 52 - almost perfect reliability - slight overcofident - poor sharpness
Drought prediction SPI 3 < -1 BS=. 22 BSS=-. 006 - low reliability - strongly overconfident, poor resolution
to conclude • ECMWF long-range ensemble forecasts employed to predict drought (SPI) • different combinations of observed and forecasted period are proposed • good skill, relatively strong signal – worse skill for mountain areas – worse skill for more extreme SPIs (-2, +2) • introduce to operational forecast in DHMZ in 2016
final product
Thank you! Refferences • • • Cindrić, K. , Pasarić, Z. , Gajić-Čapka, M. 2013. Time trends in dry and wet spells in Croatia (1961 -2010). In Milan, Dimkić (ed. ), Climate Change Impacts on Water Resources; Proc. intern. symp. , Belgrade, 17 -18 October 2013. Belgrade: Jaroslav Černi Instite for the Development of Water Resources. Balkema. Gajić-Čapka, M. , Cindrić, K. , Pasarić, Z. 2014. Trends in precipitation indices in Croatia, 1961– 2010. Theoretical and Applied Climatology. DOI: 10. 1007/s 00704 -014 -1217 -9. IPCC 2013. Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Mc. Kee T. B. , Doeksen N. J. , Kleist J. 1993. The relationship of drought frequency and duration on time scales. Proceedings of the 8 th Conference of Applied Climatology. American Meteorology Society: Anaheim CA, Boston MA, 179– 184. WMO (2012) Standardized Precipitation Index User Guide (Svoboda M, Hayes M , Wood D). WMO-No. 1090, Geneva.
- Slides: 20