Decomposition of Sources of Errors in Monthly to Seasonal Streamflow Forecasts in a Rainfall–Runoff Regime

Abstract

Despite considerable progress in developing real-time climate forecasts, most studies have evaluated the potential in seasonal streamflow forecasting based on ensemble streamflow prediction (ESP) methods, utilizing only climatological forcings while ignoring general circulation model (GCM)-based climate forecasts. The primary limitation in using GCM forecasts is their coarse resolution, which requires spatiotemporal downscaling to implement land surface models. Consequently, multiple sources of errors are introduced in developing real-time streamflow forecasts utilizing GCM forecasts. A set of error decomposition metrics is provided to address the following questions: 1) How are errors in monthly streamflow forecasts attributed to various sources such as temporal disaggregation, spatial downscaling, imprecise initial hydrologic conditions (IHCs), climatological forcings, and imprecise forecasts? and 2) How do these errors propagate with lead time over different seasons? A calibrated Variable Infiltration Capacity model is used over the Apalachicola River at Chattahoochee in the southeastern United States. The model is forced with a combination of daily precipitation forcings (temporally disaggregated observed precipitation, spatially downscaled and temporally disaggregated observed precipitation, ESP, ECHAM4.5 forecasts, and observed) and IHCs [simulated and climatological ensemble reverse ESP (RESP)] but with observed air temperature and wind speed at 1 /88 resolution. Then, errors in forecasting monthly streamflow at up to a 3-month lead time are decomposed by comparing the forecasted streamflow to simulated streamflow under observed forcings. Results indicate that the errors due to temporal disaggregation are much higher than the spatial downscaling errors. During winter and early spring, the increasing order of errors at a 1-month lead time is spatial downscaling, model, temporal disaggregation, RESP, large-scale precipitation forecasts, and ESP.

Identification of dominant source of errors in developing streamflow and groundwater projections under near‐term climate change

“…In this analysis, we considered the error due to GCM and downscaling into one single source, but one could estimate the downscaling error alone, following the approach of Sinha et al . [] and Mazrooei et al . [], by upscaling the observed precipitation and temperature to the spatial resolution of BCSD (i.e., 1° × 1°) and then applying the downscaling procedure, quantile mapping, adapted by BOR.…”

“…We did not estimate the downscaling error separately here, since we expect it to be very small as found by Sinha et al . [] and Mazrooei et al . [].…”

“…This provides us the basic framework and a general set of equations for understanding the magnitude and direction of errors arising from the model chain. In this analysis, we considered the error due to GCM and downscaling into one single source, but one could estimate the downscaling error alone, following the approach of Sinha et al [2014] and Mazrooei et al [2015], by upscaling the observed precipitation and temperature to the spatial resolution of BCSD (i.e., 1°× 1°) and then applying the downscaling procedure, quantile mapping, adapted by BOR. We did not estimate the downscaling error separately here, since we expect it to be very small as found by Sinha et al [2014] and Mazrooei et al [2015].…”

“…Following the decomposition procedures suggested by Sinha et al [2014] and Mazrooei et al [2015] for understanding the dominant sources of error in seasonal streamflow forecasts, the difference between observed and projected hydrologic variables could be linearly decomposed into three different sources of errors based on equation (1). Readers refer to Text S1 and Figure S1 in the supporting information for the validation of the linear decomposition.…”

Hybrid Models Combining EMD/EEMD and ARIMA for Long-Term Streamflow Forecasting

“…Physical models are based on the physical hydrological dynamic process, combing spatio-temporal distribution of precipitation [4,5], meteorological conditions [6,7], and the underlying surface condition [8,9]. The complexity of the runoff generation and flow concentration processes with various influence factors such as climate, geographical environment and human activities leads to the difficulty to build a hydrological models with high accuracy.…”