Improvement of Land-surface Hydrological Model TOPX and Its Coupling with Regional Climate Model WRF

Author:Yi Lu

Supervisor:Zhang Mochang


Degree Year:2018





Water is an important material basis for the survival of the earth’s organisms.Regional water cycle is a significant natural cycle related to regional economic development,ecological environment and water resources security.Since the 1970s,global warming has triggered the acceleration of global hydrological cycle,the increase of climatic extreme events and the change of seasonal rainfall patterns,this results in frequent regional droughts and floods and posed a profound impact on many water-related problems such as water resources storage,soil erosion,ecological environment health,and agricultural water resource management.In regional water cycle simulation,how to make full use of numerical simulation,data assimilation,remote sensing observation and other technologies to prompt the advantages of land-atmosphere coupling model with high spatial-temporal resolution,less limitation by observation data,and relative longer forecast period of flood forecasting,so as to improve the precision of rainfall-runoff simulation by land-atmosphere coupling model,has become an urgent need in the relevant studies.Under the background of global change,developing physically based land-atmosphere coupling model with the earth observation information to study the response of water cycle in response of global changes through refining water cycle simulation and clarifying the law of regional water cycle has become a hot topic in hydrometeorology studies.Huaihe River Basin(HRB)located between the Yangtze River and the Yellow River is one of the seven major river basins in China,it is in China’s north-south climate transition zone covering an area about 270,000 square kilometers.The HRB has an important socio-economic status and is one of China’s important agricultural products base since its population density is the first among the other Chinese major river basins(about 660 people/km 2)and it occupies 17%of the country’s cultivated land.Due to the comprehensive influence of geography and climate,the precipitation in the HRB varies greatly and distributes unevenly during the year.The precipitation in flood season(June to September)accounts for 60-80%of the total annual precipitation.The uneven spatial and temporal distribution of rainfall,together with the topography of high hills in its west and most plain in its middle,leads to frequent floods and droughts in the HRB.In recent years,under the background of global change,the extreme precipitation in the HRB during flood season shows a significant increase trend and obvious runoff changes.Therefore,it is necessary to study the simulation of water circulation in the HRB with the basis of land-atmosphere coupling model.Based on the modification of topographic index in the TOPX model and the accuracy improvement for WRF rainfall forecasting,a land-atmosphere coupling model constructed for simulating heavy rainfall-runoff processes in the HRB was proposed.The main research contents and conclusions of the dissertation include the following three aspects:(1)Considering the soil hydraulic conduction and soil erosion resistance,a new topographic index TI’ was constructed and calculated by the modified IMFD algorithm.The rationality of TI’ was demonstrated by the experiments of river network extraction and hydrological simulation.In order to quantify the effect of soil spatial heterogeneity on rainfall-runoff process and improve the simulation performance of topography-based land surface hydrological model,two soil characteristic parameters,the soil saturated hydraulic conductivity(Ks)and the soil erodibility factor(K),were added to the traditional topographic index TI(ln(α/tanβ))to construct a new topographic index TI’(ln(α/tanβ·Ks·K)).In order to calculate the spatial distribution of TI’ in the study areas,the spatial distributions of Ks and K were firstly calculated by the SPAW model and the EPIC model with a I:1 000000 HWSD soil database,thus to avoid time-consuming and poorly point-to-area representative in-situ observations.Secondly,based on the core idea of TI’ that the two underlying surface factors,i.e.soil and topography,jointly determine the flow direction,the existed topographic index calculation method IMFD was modified to obtain the spatial distribution of TI’.In order to prove the rationality of TI’,on the one hand,the flow direction calculation module of the river network extraction tool TauDEM was improved obeying the idea of TI’ that soil and terrain together determine the flow direction,then the different river networks in study regions were extracted based on TI and TI’.The extraction results correctly showed that water is more inclined to flow to the soil with high saturated hydraulic conductivity and strong erodibility,and as the saturated area which is closely related to runoff generation is the extension of river network to a certain extent,this proved that TI could correctly reflect the effect of soil spatial heterogeneity on rainfall-runoff process.On the other hand,TI and TI’ were put into topography-based hydrological models TOPMODEL and TOPX to simulate daily rainfall-runoff processes in the study watersheds located in different climatic regions.Compared with the TI-based hydrological simulation results,the Nash coefficients of the TI’-based hydrological simulation results in Yingluoxia,Wangjiaba and Huangqiao watersheds were increased by 0.063,0.019 and 0.003,respectively.The results showed that TI’ could improve the performance of topography-based hydrological model and the enhancement increased with the spatial heterogeneity of Ks·K increased.(2)Taking TRMM 3B42 and GPM IMERG as observation operators,4D-Var assimilation technique was used to improve the initial and boundary conditions of the WRF model to enhance the rainfall prediction accuracy.Taking the HRB as study basin and two heavy precipitation events in the flood and non-flood season in 2015 as study cases,firstly,the sensitivities of the WRF model to rainfall types,initial boundary conditions and spatial resolution were analyzed.The results showed that with the same parameter configuration in the same research area,the WRF model driven by NCEP ds083.3 had better performance than that driven by NCEP ds083.2 data,and the rainfall simulation results in the high-resolution inner domain were better than those in the low-resolution outer domain.Secondly,NCEP ds083.3 dataset was applied to drive WRF 4D-Var system assimilated with TRMM 3B42 or GPM IMERG.Comparing the simulation results of the assimilation experiments and their corresponding control experiments,the results showed that the WRF 4D-Var assimilation system based on TRMM 3B42 or GPM IMERG could both effectively improve the simulation efficiency of the WRF model,but the 4D-Var assimilation maintained a substantial improvement for about 12 hours because of the error accumulation when WRF running.The accuracy of WRF4D-Var rainfall simulation with direct assimilation of GPM IMERG was higher than that with TRMM 3B42.In the 9-km inner domain,comparing to the daily precipitation observed by the meteorological stations in the HRB,the Pearson correlation coefficients of the WRF 4D-Var system assimilated with GPM IMEG were 0.74 and 0.51 for the study rainfall events in flood season and non-flood season,respectively;comparing to the hourly merged CMORPH data,the HSS values of the simulated results of the two heavy rainfall events both reached 0.31.(3)Realized the coupling of the TOPX and the WRF on 1-km horizontal grid and the simulation of heavy rainfall-runoff process in the Wangjiaba watershed based on this land-atmosphere coupling model.This study took the sub-basin of the HRB,the Wangjiaba(WJB)watershed,as study area and a heavy rainfall-runoff process in the flood season of 2015 as study case.On the basis of the confirmation of the new topographic index and the 4D-Var assimilation with GPM IMERG,in order to ensure the simulation efficiency of the land-atmosphere coupling model based on the TOPX and the WRF,on the one hand,the TI’ was taken as the topographic index input of the TOPX and the TOPX was calibrated and validated with long-term and short-term rainfall-runoff simulations in study area.For the total 7 short-term rainfall-runoff simulations for calibration and verification,the minimum Nash coefficient was 0.747.On the other hand,4D-Var algorithm was used to assimilate the GPM IMERG satellite estimation to improve WRF rainfall prediction.Compared with the daily rainfall data and daily potential evapotranspiration data obtained from the precipitation stations and evaporation stations in the WJB watershed,the daily rainfall and the daily potential evapotranspiration simulated by the WRF 4D-Var system were respectively smaller and larger than the in situ observed values on the whole,and the Pearson correlation coefficients were 0.444 and 0.746,respectively.Finally,the coupling of the TOPX and the WRF was realized on the 1-km grid by using grid nesting technology.The results of the land-atmosphere coupling model showed that the average correlation coefficient between the daily soil moisture simulated by the land-atmosphere coupling model and the daily soil moisture estimated by the SMAP Level 4 in root area was 0.802.The daily outlet discharge simulated by this land-atmosphere coupling model was less than the in situ observations from the hydrological station,and the Nash coefficient was 0.543.The land-atmosphere coupling model based on the TOPX model and the WRF 4D-Var assimilation system could better reproduce the heavy rainfall-runoff process in the WJB watershed.