An Integrated Study Of Water And Heat Flux Data From Multi-model Simulations In CLDAS In China

The water and heat flux between soil,vegetation and the atmosphere is the basis to research the atmospheric dynamic and thermal process between land and atmosphere.They are important to know the global energy balance and water cycle.However,due to the difficulty of obtaining observational data,research on large-scale area has been restricted.Data assimilation and the development in land surface models provide the possibility for this.The atmospheric driving field products and the fusion analysis products of land surface factors of the CMA Land Data Assimilation System(CLDAS)already had higher temporal and spatial resolution and higher inspection quality in China,but there is no water and heat flux data in its commercial operation.Therefore,this article is based on the water and heat flux data from 2008 to 2017simulated by the three land surface models(CLM,Noah,Noah-MP)in CLDAS.Use site data to evaluate its region in China.And the simulation results were weighted to integrated based on the uncertainty of three kinds of land surface model.At last,the integrated data have been compared with the international similar data set.The main conclusions of the study are as follows:(1)The results of evaluating CLDAS's water and heat flux simulation data using long-term series of observation data from 12 flux towers in China showed that there were certain commonalities in the simulation results of water and heat flux.For example,the simulated value of latent heat flux was overestimated in mixed forests and coniferous forests while the simulated value of sensible heat flux was overestimated at all sites.However,none of the model simulation results performed best under all underlying surfaces.The seasonal and annual average spatial distribution of water and heat flux simulation values had good consistency.Latent heat space showed a decreasing distribution law from southeast to northwest while seasonally showed the highest value in summer and the lowest value in winter.The sensible heat space presented an increasing distribution law from southeast to northwest while seasonally showed the highest value in spring and the lowest value in winter.(2)The study used two methods(arithmetic average and weighted average based on Triple Collocaton(TC))to integrate the simulation results of three land surface models under CLDAS at 12 stations.The results showed that the weighted average method based on TC was slightly superior to the arithmetic average method numerically,and it improved the sensible heat flux greatly when the values are different from each other,but it improved the latent heat flux less when the values are close.Therefore,this paper used the weighted average method based on TC to integrate the water and heat flux data simulated by the three land surface models to obtain the water and heat flux integrated data set(CLDAS-triple collocation,CLDAS_TC).The correlation between the data before and after the latent heat flux integration was higher than 0.9and the numerical difference was less than 8W·m^-2.CLDAS_TC had the highest correlation with CLDAS-Noah-MP(0.97),which was the closest to CLDAS-Noah value.The correlation between data before and after sensible heat flux integration ranged from 0.83 to 0.96 while the numerical difference ranged from 8.06 to 14.93 W·m^-2.CLDASTC had the highest correlation with CLDAS-CLM,and was the closest numerically to CLDAS-Noah-MP.(3)Comparing the integrated data set(CLDASTC)with the latent heat flux data of other international data sets,that is the Noah model simulation data set of the Global Land Surface Data Assimilation System(GLDAS),The fifth-generation global reanalysis data(ERA5)issued by the European Center for Medium-Range Weather Forecast(ECWMF)and the reanalysis data set(MERRA2)of the NASA,etc.,it can be seen that CLDAS_TC performed better on the site(the correlation is 0.80 and the minimum root mean square is 30.41W·m^-2).The correlation between latent heat flux data was spatially higher in humid areas and lower in dry areas.CLDAS_TC was the closest numerically to GLDAS-Noah.The numerical difference between CLDAS_TC and ERA5 was mainly concentrated in the polar regions.The numerical difference between CLDAS_TC and MERRA2 was mainly concentrated in the southern temperate zone.mainly concentrated in the southern region of the temperate zone.From the perspective of time variation,the latent heat flux data all met the increasing interannual variation trend,and the increase range was larger in the tropical and temperate zone while the increase range was smaller in the dry zone and polar zone.From the perspective of time change,the latent heat flux data all met the increasing inter-annual change trend,and the increase in the tropical and temperate zones was larger while the increase in the dry zone and the polar zone was smaller.(4)Comparing the sensible heat flux data in the integrated data set(CLDAS_TC)with other international data sets,we can see that at the site,the correlation between CLDAS_TC and the observation data was the highest(0.63),while the value of ERA5 was the closest to the observation data.The spatial distribution of the correlation between the sensible heat flux data was opposite to that of the latent heat flux.The correlation was higher in the dry areas of northern China and lower in the humid areas of the south.CLDAS_TC was the closest numerically to MERRA2 while the difference between CLDAS_TC and ERA5 in numerical value was the biggest and the difference was mainly concentrated in the dry zone and polar zone.From the perspective of time change,the inter-annual change trend of sensible heat flux was quite different.CLDAS_TC and GLDAS-Noah showed an increasing inter-annual change trend while the two reanalysis data ERA5 and MERRA2 showed a decreasing trend.