Hydrol. Earth Syst. Sci., 22, 2225–2254, 2018 https://doi.org/10.5194/hess-22-2225-2018 © Author(s) 2018. This work is distributed under the Creative Commons Attribution 4.0 License. Statistical forecast of seasonal discharge in Central Asia using observational records: development of a generic linear modelling tool for operational water resource management Heiko Apel1, Zharkinay Abdykerimova2, Marina Agalhanova3, Azamat Baimaganbetov4, Nadejda Gavrilenko5, Lars Gerlitz1, Olga Kalashnikova6, Katy Unger-Shayesteh1, Sergiy Vorogushyn1, and Abror Gafurov1 1GFZ German Research Centre for Geoscience, Section 5.4 Hydrology, Potsdam, Germany 2Hydro-Meteorological Service of Kyrgyzstan, Bishkek, Kyrgyzstan 3Hydro-Meteorological Service of Turkmenistan, Ashgabat, Turkmenistan 4Hydro-Meteorological Service of Kazakhstan, Almaty, Kazakhstan 5Hydro-Meteorological Service of Uzbekistan, Tashkent, Uzbekistan 6CAIAG Central Asian Institute for Applied Geoscience, Bishkek, Kyrgyzstan Correspondence: Heiko Apel (
[email protected]) Received: 15 June 2017 – Discussion started: 21 June 2017 Revised: 13 February 2018 – Accepted: 27 February 2018 – Published: 11 April 2018 Abstract. The semi-arid regions of Central Asia crucially els are derived based on these predictors as linear combi- depend on the water resources supplied by the mountain- nations of up to four predictors. A user-selectable number ous areas of the Tien Shan and Pamir and Altai moun- of the best models is extracted automatically by the devel- tains. During the summer months the snow-melt- and glacier- oped model fitting algorithm, which includes a test for ro- melt-dominated river discharge originating in the moun- bustness by a leave-one-out cross-validation. Based on the tains provides the main water resource available for agricul- cross-validation the predictive uncertainty was quantified for tural production, but also for storage in reservoirs for en- every prediction model.