| China is one of the countries that are most severely and frequently affected by extreme hydrological events in the world.Economic losses caused by floods and droughts are more than tens of billions of RMB per year,with many deaths and injuries.In the last three years,there have been a number of extreme events such as the flooding of the entire Yangtze River basin in 2020,the extraordinarily heavy rainfall and flooding in Henan in 2021,and the extreme drought in the Yangtze River basin in 2022.In the context of climate change,extreme precipitation will increase significantly and tend to intensify,and more extreme heat events are causing higher risks of heat waves and droughts.Reliable seasonal hydrological forecasting is therefore needed to effectively manage water resources and reduce losses due to associated hazards.Seasonal climate predictions based on ocean-atmospheric teleconnections are now widely used for seasonal hydrological ensemble predictions,but the climate model systems developed during the hindcast period are not necessarily applicable to future real-time forecast due to decadal climate change,and their reliability needs to be assessed.This study conducts seasonal hydrological prediction experiments in the Yangtze River basin based on North American Multi-Model Ensemble(NMME)climate prediction data,and evaluates the hindcast and real-time forecast skills for hydrometeorological variables such as precipitation,temperature,and streamflow.The main research content and achievements are as follows:(1)Through comparing and assessing the seasonal climate prediction skills in the Yangtze River basin between the hindcast and real-time forecast periods,it was found that the forecast skills for precipitation and temperature decreased during the real-time forecast period.During the hindcast period,the climate model has a low forecast bias and a high forecast skill.Bias in precipitation during the flood season were less than 0.2 mm/day and bias in temperature were generally less than 1°C.The multi-model ensemble predictions of precipitation and temperature have anomaly correlations larger than 0.3 and 0.5,with unbiased root mean squared errors of 0.9 mm/day and 0.7°C,respectively.Compared to individual models,the multi-model ensemble improved the anomaly correlation by 0.1 and0.14 and reduces the errors by 0.08 mm/day and 0.11°C,respectively.In the real-time forecast period,the climate model has an increased bias and a decreased prediction skill for precipitation and temperature.Compared to the hindcast period,precipitation in the middle and lower reaches of the Yangtze River increased significantly by 1～2 mm/day during the real-time forecast period,but the climate model failed to reflect the decadal variability in precipitation,leading to amplified negative biases in precipitation predictions.Temperatures increase across the basin during the real-time forecast period,and climate models predict the increase to some extent.Compared to the hindcast period,the multi-model ensemble experienced a 14% decrease in skill and a 30% increase in error for real-time precipitation predictions,and a 38% decrease in skill and a 51% increase in error for temperature predictions.(2)The decadal variation of the ocean-atmospheric teleconnection was analyzed,and it was concluded that the failure to reasonably reproduce the variation of the Indian Ocean SST-Yangtze River Basin climate teleconnection led to a decrease in the models’ real-time forecast skill.In El Ni?o and La Ni?a years,the forecast errors of precipitation and temperature for the multi-model ensemble and individual models were reduced by up to 40%compared to normal years,indicating that the climate models could reasonably describe the correlation between SST in the eastern equatorial Pacific and the climate of the Yangtze River basin.However,there was a significant decadal variation in the Indian Ocean SST-Yangtze River Basin climate teleconnection in the last decade,with the Yangtze River temperature-Indian Ocean SST correlation changed from 0.5 to about-0.5.At the same time,the Indian Ocean SST anomaly significantly enhanced the negative anomalies in the 200 h Pa geopotential height and outward longwave radiation in the Yangtze River basin and changed the wind field around.The significant influence of the Indian Ocean SST anomaly on the atmospheric circulation situation in the Yangtze River basin increased its impact on the climate prediction in this area.This decadal variation in the ocean-atmospheric teleconnection is not well represented in climate models,which is an important reason why the forecast skill in real-time period is lower than the hindcast period.(3)Seasonal streamflow predictions in the Yangtze River basin driven by a high-resolution land-surface hydrological model with bias corrected climate model predictions taking into account decadal climate change were found to have improved streamflow forecast skill as compared with traditional hydrological predictions without integration of climate predictions,but hydrological forecast skill in the real-time forecast period was also lower than that in the hindcast period.Firstly,the land surface hydrological model CSSPv2 was evaluated and found to have a good ability to simulate streamflow in the Yangtze River basin.Using an ensemble streamflow prediction method based on random sampling from historical meteorological forcings to drive the CSSPv2 model for streamflow prediction,it was found that streamflow forecast skill was relatively high in the Yangtze River basin in the short lead time.However,the forecast skill decreased significantly at long leads.When NMME climate model predictions were taken into account,the streamflow forecast skill in the Yangtze River basin under long lead times was significantly improved,with an improvement of average KGE about 0.2.However,influenced by the decrease in climate forecast skill during the real-time forecast period,streamflow forecast skill was also relatively low during this period.Further improvements in real-time climate prediction capability are needed to develop more reliable seasonal hydrological predictions. |
PDF Full Text Request