Low flow prediction Flashcards
1
Q
Predictability
A
Droughts are less dynamic and develop slowly. So, long term forecasts are possible and of practical relevance. (Lead time: weeks to months)
Even alone the monitoring of groundwater levels, soil moisture, status of reservoirs, discharge etc. can help to assess the future drought potential.
Simple forecasts for low flows are often scenario based concerning the input (e.g. scenario „no precipitation“ provides lower limit of discharge)
Better forecasts consider meteorological predictors like circulation pattern, teleconnection, anomalies of circulations (El- Niño)
2
Q
Methods
A
- Conceptual hydrological models:
Forecast running the model without precipitation input (“Scenarios”)
Application of recession curves (see section 4.1.4) - Stochastic time series and regression models:
Application of ARMA – models for flow forecasts
Application of regression models with various predictors (soil
moisture, precipitation, snow, clime variables, etc.) - Soft-Computing methods:
Artificial neural networks (ANN) and fuzzy rule based methods (FRBM); (see also “Geostatistics & Softcomputing”)
Consideration of fuzzy knowledge and prediction of categorical variables (e.g. wet, dry, mean) is possible