Hydrology and Earth System Sciences (Apr 2018)

Statistical forecast of seasonal discharge in Central Asia using observational records: development of a generic linear modelling tool for operational water resource management

  • H. Apel,
  • Z. Abdykerimova,
  • M. Agalhanova,
  • A. Baimaganbetov,
  • N. Gavrilenko,
  • L. Gerlitz,
  • O. Kalashnikova,
  • K. Unger-Shayesteh,
  • S. Vorogushyn,
  • A. Gafurov

DOI
https://doi.org/10.5194/hess-22-2225-2018
Journal volume & issue
Vol. 22
pp. 2225 – 2254

Abstract

Read online

The semi-arid regions of Central Asia crucially depend on the water resources supplied by the mountainous areas of the Tien Shan and Pamir and Altai mountains. During the summer months the snow-melt- and glacier-melt-dominated river discharge originating in the mountains provides the main water resource available for agricultural production, but also for storage in reservoirs for energy generation during the winter months. Thus a reliable seasonal forecast of the water resources is crucial for sustainable management and planning of water resources. In fact, seasonal forecasts are mandatory tasks of all national hydro-meteorological services in the region. In order to support the operational seasonal forecast procedures of hydro-meteorological services, this study aims to develop a generic tool for deriving statistical forecast models of seasonal river discharge based solely on observational records. The generic model structure is kept as simple as possible in order to be driven by meteorological and hydrological data readily available at the hydro-meteorological services, and to be applicable for all catchments in the region. As snow melt dominates summer runoff, the main meteorological predictors for the forecast models are monthly values of winter precipitation and temperature, satellite-based snow cover data, and antecedent discharge. This basic predictor set was further extended by multi-monthly means of the individual predictors, as well as composites of the predictors. Forecast models are derived based on these predictors as linear combinations of up to four predictors. A user-selectable number of the best models is extracted automatically by the developed model fitting algorithm, which includes a test for robustness by a leave-one-out cross-validation. Based on the cross-validation the predictive uncertainty was quantified for every prediction model. Forecasts of the mean seasonal discharge of the period April to September are derived every month from January until June. The application of the model for several catchments in Central Asia – ranging from small to the largest rivers (240 to 290 000 km2 catchment area) – for the period 2000–2015 provided skilful forecasts for most catchments already in January, with adjusted R2 values of the best model in the range of 0.6–0.8 for most of the catchments. The skill of the prediction increased every following month, i.e. with reduced lead time, with adjusted R2 values usually in the range 0.8–0.9 for the best and 0.7–0.8 on average for the set of models in April just before the prediction period. The later forecasts in May and June improve further due to the high predictive power of the discharge in the first 2 months of the snow melt period. The improved skill of the set of forecast models with decreasing lead time resulted in narrow predictive uncertainty bands at the beginning of the snow melt period. In summary, the proposed generic automatic forecast model development tool provides robust predictions for seasonal water availability in Central Asia, which will be tested against the official forecasts in the upcoming years, with the vision of operational implementation.