Watershed Short-term Hydrological Forecast Coupling With Weather Forecasting

Watershed hydrological forecast has been a key research issue in hydrology field,as well as an important support for the planning and construction of water conservancy project,high-efficiency use of water resource and safe operation of hydropower system.With the combined impact of global climate change and large-scale development of hydropower,the water regime of the watersheds and spatial and temporal distribution characteristic of water resources have changed deeply in recent years,which brings new challenge for watershed hydrology forecast research.The short-term hydrological forecast of watershed coupling weather forecast is a major way to increase the precision and forecast period of runoff forecast,so it belongs to multi-disciplinary domain.Considering the key problem occurred in the high-efficiency use of water resource in changing climate and taking Jinsha River Basin as the main research object,this study explores advanced theories and methods in the fields of watershed hydrologic analysis,one-way coupled hydro-meteorological forecast and runoff interval forecast.The achieved conclusions and innovative research results are as follows:(1)In this research,we first analyzed the meteorological and hydrologic temporal and spatial changing situations of a basin under various conditions.For the fact that conventional univariate trend analysis can't test the significant changing trends in the whole hydrologic event,multivariable Mann-Kendal trend analysis method is introduced in this paper to test the trend of rainfall and runoff joint variants in the Jinsha river basin main control sites.Results show that annual precipitation and runoff of Jinsha river was on the rise generally,and the annual precipitation in the control areas of Longjie,Wudongde and Baihetan passed the univariate MK significance test and showed significant increasing trend in the average annual rainfall in these areas.Further researches using multivariate MK trend analysis showed that there's a significant increasing trend in annual precipitation and runoff of Longjie station,which indicated a gradually strengthened meteorologic and hydrologic circulation strength in this area.However,we failed to find an increasing trend in the annual runoff of Longjie station in the univariate MK significance test.From the results listed above,we can infer that multivariate MK trend test is efficient in compelling the joint trend of precipitation and runoff,which is very helpful in fully describing the characteristics of hydrologic events.(2)Combining the evolving characteristics of meteorologic and hydrologic factors in river basin,we optimized the parameterization scheme and nesting model of physical process of precipitation in Jinsha river basin,and established the WRF numerical weather forecasting model suitable for Jinsha riverbasin.We also explored the precipitation ensemble forecast considering different initial fields and parameterization schemes,thus providing high temporal and spatial resolution short-term precipitation forecast with a certain precision.Study showed that the optimal parameterization scheme combination of WRF model in the Jinsha river basin is as follows:the Lin et al.scheme,Mellor-Yamada-Janjic(Eta)TKE scheme,Kain-Fritsch(new Eta)scheme was adopted as cloud microphysics scheme,boundary layer scheme and cumulus parameterization scheme respectively.For the Jinsha river basin,there wasn't an obvious advantage in the double nested simulation comparing to single nested simulation,and the three downscaling methods achieved comparatively unanimous results when mesh accuracy was divided properly.The results also show that mesoscale precipitation ensemble forecast system with multiple initial values and physical processes in Jinsha river basin can get the ensemble forecast results that reflect the precipitation forecast uncertainty.Precipitation interval prediction and mid-value forecast resulting from ensemble forecast has an important guiding significance to prediction dispatch.(3)On the basis of adopting the numerical weather model WRF in the basin's short-term weather forecasting,we established the short-term weather-hydrological process coupling forecast system in Jinsha River basin.In order to overcome the shortcoming of conventional hydrological model single-target parameter calibration method,we designed triple-target MOSCDE algorithm taking relative error of total runoff,deterministic target and flood peak relative error as the targets.We conducted multiple target parameter optimization calibration of Xinanjiang model,and introduced the fuzzy theory and comentropy theory to the multiple target noninferior solution fuzzy optimization principle.Results showed that the runoff forecast system coupling with numerical weather prediction in the Jinsha river basin guaranteed the accuracy of hydrologic forecasting,and it also increased the runoff forecast period efficiently.Multiple target parameter optimization calibration can reach the overall optimization among different targets,and realize accurate simulations of multiple factors including peak discharge,flood peak appearance time,flood volume,rising and declining of flood.Furthermore,in order to take the influence of precipitation forecast results in to full consideration,we conducted runoff ensemble forecast study in Jinsha river basin.The results show that ensemble forecast is able to synthesize the forecasting advantages of different members,and the runoff forecast interval based on the ensemble forecast can fairly reflect the varying pattern of prediction error,which provides abundant forecasting information for watershed management.(4)Short-term hydrologic forecasting coupling weather forecast can provide runoff prediction with certainty,while the evolution of weather-hydrologic process has the characteristic of uncertainty.Prediction interval(PI)can give the upper and lower boundaries of the predictive variables at each moment,which can quantify and represent uncertainty of the forecast results and is more advanced than the deterministic forecast.In this paper,we not only introduced the symmetry index to improve the traditional single-objective Lower Upper Bound Estimation(LUBE)method,but also extended the LUBE interval forecasting method into the multi-objective framework.The results show that the proposed LUBE method achieves good performance in PICP,PIARW and PIS simultaneously,and shows great performance in the midpoint forecasting by introducing symmetry.At the same time,after extending the LUBE method to the multi-objective framework,the workload of choosing parameters in the CWSC objective function was greatly reduced.By introducing the multiple target noninferior solution fuzzy optimization principle,it would be easier for researchers to make decisions among PICP,PIARW and PIS,therefore getting the runoff prediction interval that meets actual needs.(5)Based on the theoretical researches listed above,and considering the practical engineering requirements of watershed runoff forecast management,this paper investigated the cross-platform data transmission and database efficient reading technology,and realized the background automatic forecasting function.Furthermore,on the basis of distributed parallel computing,loosely-coupled service processing and high-performance human-computer interaction technology,the short-term weather-hydrological process coupling forecast system in Jinshajiang River Basin was developed on the C/S structure model and it has been intergrated into the water managing and control platform of Jinsha River basin cascade dispatching center for deployment and application.