Seasonal Variations And Simulations Of Water-Heat-Carbon Flux In The Seasonal Arid Region Kiwifruit Orchard Of Southwest China

The variations and simulations of farmland water-heat-carbon flux,involving hydrology,ecology,geography and other disciplines,are the hot spot and frontier of the international academic community.Understanding the process and transmission mechanism of farmland water-heat-carbon cycle is of great significance to perfect agricultural ecological hydrology theory,optimize agricultural water management and improve agricultural water use efficiency.In recent years,kiwifruit planting has become a pillar industry for agricultural economic development and industrial poverty alleviation in Southwest China.In this study,Pujiang County,the main kiwifruit-producing area locating in Chengdu Plain,was selected as the research area.The comprehensive observation of water-heat-carbon flux was conducted using eddy covariance system（EC）,sap flow systems,micro-lysimeter and portable photosynthesis measurement system during the three consecutive kiwifruit growing seasons of 2018-2020.The process and coupling mechanism of water-heat-carbon transport were explored,and the transport or conversion processes of water-heat,light-carbon and water-carbon were simulated by mathematical models.The main achievements are as follows:（1）The variations of water,heat and carbon flux at various time scales of kiwifruit orchard were analyzed based on the measured data of EC and sap flow.It was found that the water-heat flux of different components changed with the single peak curve during the growing seasons.Latent heat flux（LE）was the main energy consumer of net radiation（R_n）,the distribution and conversion of energy were regulated by the underlying surface conductance（G_c）.The average values of evapotranspiration（ET）were 2.34,1.95 and 1.99 mm d^-1,the average values of canopy transpiration（T）were 1.19,1.00 and 0.91 mm d^-1,and the average values of surface evaporation（E’=ET-T）were 1.15,0.9 and 1.08 mm d^-1 during the growing seasons of 2018-2020,respectively.ET was most sensitive to the variation of water vapor pressure difference（VPD）based on the Penman-Monteith model,followed by the underlying surface conductance（r_c）and radiation term（R_n-G）.T was the most sensitive to canopy conductance(r_cp),followed by VPD and radiation term(R_nc).Kiwifruit crop coefficient（K_c）varied from 0.75 to 0.92,the basic crop coefficient(K_cb)varied from 0.35 to 0.48 and the surface evaporation coefficient（K_e）varied from 0.34to 0.45 during the growing seasons of 2018-2020.The correlation between crop coefficient and leaf area index（LAI）was significant（p<0.05 or 0.01）.The diurnal variations of gross primary productivity（GPP）,soil respiration（R_e）and net ecosystem productivity（NEE）of the kiwi orchard showed a bell-shaped change.The average value of GPP in the whole growing season varied from 5.88 to 6.77 g C m^-2 d^-1,NEE varied from-0.85 to-1.79 g C m^-2 d^-1 and R_e varied from 4.78 to 5.15 g C m^-2 d^-1during the 2018-2020 growing seasons.Net radiation,air temperature and VPD were the key meteorological factors affecting GPP,NEE and R_e,LAI also played an important role in regulating carbon flux during the three growing seasons.（2）Transpiration rate（T_r）,photosynthetic rate（P_n）and water use efficiency（WUE）between exposed and sunlit leaves were different due to the influence of light.Photosynthetically active radiation（PAR）and stomatal conductance（g_s）had significant effects on T_r and P_n（p<0.01）.The indirect effects of T_a,PAR and leaf VPD（VPDl）through T_r on WUE_i is greater than that of P_n;adding VPD_l(especially VPD_l^0.5)into P_n increases its coupling with T_rat leaf scale.As the ratio of P_n*VPD_l^0.5 to Tr,WUE_u can balance the effects of T_a,PAR,VPD_l and g_s through leaf water consumption and carbon assimilation process on leaf scale WUE.At the ecosystem scale,adding VPD(especially VPD^0.5)to GPP,could significantly reduce the lag relationship between GPP and ET at a half-hour scale.Therefore,GPP*VPD^0.5 possessed a stronger correlation with ET at the half-hour scale（r=0.99,p<0.01）.The path analysis showed that R_g,T_a,U₂ and VPD were the dominant climatic factors affecting WUE at the ecosystem scale,and all of them mainly affected WUE by affecting the process of water consumption（ET）rather than carbon assimilation（GPP）.R_g,T_aand VPD were the dominant factors that affected the light use efficiency（LUE）by affecting the light absorption process,while the water vapor density（CH₂O）and LAI affected LUE by affecting the carbon assimilation process.The study also found that ET and GPP responded differently to direct PAR(PAR_dir),resulting in a significant decrease in e WUE with the increased clear sky index（K_t）（p<0.01）,while the significant decrease in LUE with the increased K_t was due to the rapid decrease in direct light use efficiency(GPP/PAR_dir).（3）The simulation accuracy of ET and its components（T and E）in kiwifruit orchard were compared with five classical evapotranspiration models.It was found that different models could simulate the seasonal variation of ET well,while there were great differences as far as the ET component（E or T）among different models.In this paper,considering the horizontal stratification characteristics of kiwifruit canopy and the process of canopy light interception between exposed and sunlit leaves,a module was added to the CL model to calculate the canopy resistance and available energy of shade and sunny leaves of kiwifruit,respectively,to construct two-layer block（T-T）model.The simulated ET by the T-T model was in good agreement with the measured value of EC system,the R² between the two was above 0.92 during the growing seasons of 2018-2020.The slope between the simulated values of T and the measured values of the sap flow was 0.851,0.847,1.00,the R² between the two was0.914,0.887 and 0.927,respectively,and the slope between the simulated values of E and the measured values of the micro-lysimeter was 1.06,0.972,0.979,the R²between the two was 0.925,0.871,0.934 during the growing seasons of 2018-2020,respectively.Compared with the traditional model,the T-T model constructed in this paper could better simulate the water-heat transfer process of the kiwifruit orchard.（4）Seven classical light use efficiency（LUE）based gross primary productivity assessment（LUE-GPP）models（also called big-leaf（BL）model）were compared with the EC system(GPP_EC).It was found that different BL models could well simulate the seasonal variation of GPP_EC,with the R² varied from 0.659 to 0.729.The performed differences between the BL models were mainly due to the different quantification of temperature limitation（T_s）and humidity limitation（W_s）,also the component of the two.The results showed that the accuracy of the BL model was different under different weather conditions,and it tended to overestimate GPP on sunny days and underestimate GPP on cloudy days.The two-leaf（TL）LUE-GPP model can improve the GPP estimates accuracy in different weather conditions due to considering the difference of photosynthetic capacity between the canopy leaves（R²=0.721,p<0.01）.In this paper,the clear sky index（K_t）was introduced as a variable in the BL and TL models,the dynamic LUE-GPP assessment methods were construct based on K_t and two function fitting methods（power function and exponential function）,results showed that the constructed model improved the accuracy of GPP estimates in different weather conditions（R²=0.762）.（5）For the orchard-grass composite ecosystem,according to the water-carbon coupling relationship and the K_t-based dynamic LUE-GPP model,the ET and its components were correlated with each conductance by assuming that each conductance was independent and the water flux through each conductance was consistent,and the ET partitioning method of the orchard-grass composite ecosystem based on underlying surface conductance was proposed.Results showed that the seasonal variation of the kiwi canopy transpiration(T_Tree)estimated by the underlying surface conductance method was consistent with that measured value by the sap flow,the R² between the two was 0.881,0.842 and 0.908 during the growing seasons of2018-2020,respectively.The seasonal variation of transpiration of grass(T_Grass)and the evaporation of bare soil（E_s）were consistent with surface evaporation measured by the micro-lysimeter(E_ly),and the R² between the two was 0.779,0.805 and 0.857during the growing seasons of 2018-2020,respectively.The seasonal variation of transpiration on the underlying surface(T_Gc)（including canopy transpiration and grass transpiration of kiwifruit）estimated by the underlying surface conductance method was consistent with that of the total transpiration estimated by the underlying water use efficiency method(T_{u WUE}),and the values of R² between the two was 0.975,0.978and 0.967 during the growing seasons of 2018-2020,respectively.In conclusion,this paper provided a new way to distinguish the total ET of the tree-grass composite system.