An Approach To Simulate Root Growth And Its Effect On The Outputs Of Land Surface Models

Root system is an important part of the soil-plant-atmosphere continuum. Rootparameters are important parameters of the land surface model (LSM). Hence, thestudy of root model and its improvement, as well as its effect on the outputs ofLSM, sucn as sensible heat flux, lantent heat flux and so on, is remarkable work.Therefore, the work of the paper is as follows:Firstly, a scheme for the improvement of a root model was proposed, based onprevious researchers s work and field experiments in Jinzhou.(1) The first step is to validate and analyse the root growth module ofCERES-Maize crop growth model. We used the root growth module of CERES-Maize model,to simulate root length density in two different kinds of soil water conditions,respectively, and then compared the simulated value with observed value in Jinzhouexperimental field. The simulation results showed that: a. In well-wateredcondition, linear fitting results showed that correlation coefficients (R) androot mean square error (RMSE) between simulated and measured root length densitywere0.816and1.417, respectively, indicating that the module worked well. b.In the condition of water control in trefoil-jointing stage, R was0.879, and RMSEwas1.296, which also showed that the module worked well. However, in the conditionof water control in spin silk-milk ripe stage, R was0.392and RMSE was2.087,which showed a poorer simulation.(2) For the poorer simulation in the condition of water control in spinsilk-milk ripe stage, according to researchers s work and our further simulationverification, we found that the root distribution weighting factor (wr) had somedefects. Combined typical root distribution theory with actual observation data,we tried to present a method to improve the parameters in the root module, thatwas: when the soil water condition was unfavourable, the parameterδ (reflected declining rate of root length density with soil depth) should be decreased or theparameter Zmax(reflected maximum rooting depth) should be increased, in order tobetter depict the effect of unfavourable soil water conditions on root growth a nddistribution. In fact, this method was verified by the following simulationresults: in the condition of water control in spin silk-milk ripe stage, with Zmaxincreasing from200cm to210cm and δ decreasing from4to1, R was increased by0.105and RMSE was decreased by0.119, which proved a better simulation than before.Secondly, the application of the root module in the LSM SSiB2and its effecton the outputs of SSiB2, such as sensible heat flux, lantent heat flux and so on,were studied.In order to adapt to the local maize field in Jinzhou, the SSiB2model shouldfirstly be parameterized by taking CERES-Maize model s outputs as the dynamicparametes of it. After that, we used the SSiB2model that had been parameterizedto simulate sensible heat flux, lantent heat flux, soil moisture, soil temperature,canopy air temperature and so on, in the whole growing season of spring maize.The results showed that: for the variables above, the simulated values and observedvalues coincided well, which proved that the parameterized SSiB2model was suitableto the study on land surface process in the maize field in Jinzhou. Based on allabove, we nested the root module to SSiB2model, to discuss the effect of twodifferent root treatment schemes as follows, on the simulation results of SSiB2,such as sensible heat flux, lantent heat flux and so on:(1) When the root module provided SSiB2model with root depth only, simulationaccuracy of some variables was enhanced for the whole growing season: R betweenthe simulated upward longwave radition flux, net radition flux, sensible flux,latent flux and the corresponding observed ones were increased by0.002,0.002,0.003and0.007, and meanwhile RMSE decreased by0.099,0.425,0.38and1.84,respectively. Simulation accuracy of CO2flux was not enchanced for the wholegrowing season, but was enhanced for late growth period (from121days after sowingto mature period), with R increased by0.013and RMSE decreased by0.26.(2) When the root module provided SSiB2model with both root depth and rootlength density simultaneously, simulation accuracy of some variables was alsoenhanced for the whole growing season, compared with the original parameterizedSSiB2model: R between the simulated net radition flux, latent flux, sensible flux,and the corresponding observed ones were increased by0.0005,0.006, and0.008,and meanwhile RMSE decreased by0.097,0.36, and0.62, respectively. Simulationaccuracy of CO2flux was also enchanced for the late growth period (from121days after sowing to mature period), with R increased by0.006and RMSE decreased by0.103. Although this scheme was more comprehensive, compared with the one in (1),simulation accuracy of the variables was reduced.In all, the study showed that: putting root module into LSM, could enhancesimulation accuracy of sensible heat flux, latent heat flux and so on. But differentroot parameterization methods performed differently. Compared with only conceringroot depth, more comprehensive root scheme (concerning both root depth and rootlength density) performed worse in enhancing simulation accuracy of sensible flux,latent flux, and other flux. The reason of this should be further studied on.