Rajarshi Das Bhowmik Assistant Professor, Interdisciplinary Centre for Water Research,Indian Institute of Science, Bangalore, India presents, "Absence of temporal trends in retrospective
forecasts"
Skillful forecasts of hydroclimate variables are essential for
operational water management, agricultural planning, and food
supply. Several studies have attempted to improve the skill of raw forecasts either by post-processing or by incorporating the sea surface state to the raw forecasts. However, to the best of our knowledge, no study has investigated the absence of a temporal trend, that is present in observed records, in retrospective forecasts a.k.a. Hindcasts. The current talk will discuss two potential approaches to introduce a temporal trend in the raw forecasts by i) Updating surface boundary forcing and ii) Applying statistical models for either post-processing or streamflow forecasting. To analytically
derive the relationship between a temporal trend and forecast
performance, three statistical bias-correction approaches for seasonahead hindcasts of the Indian monsoon from a General Circulation Model (GCM) were considered by our study. The study finds that raw hindcasts of the Indian monsoons typically ignore a temporal trend that is present in the observed records. Further, analytical derivations confirm that the absence of a trend in the hindcasts significantly influences bias-correction performance. Finally, the semi-parametric approach could not overcome the limitations of a parametric linear model to yield a temporal trend in the hindcasts. The talk will also discuss one potential reason for the absence of a trend in the hindcasts.