Land-atmosphere Interaction Processes Over The Rice-wheat Rotation Area Of The Huaihe River Basin

The Huaihe River Basin（HRB）,located in the south-north climate transition zone of China with abundant water,heat and light resources,is one of the most important rice-wheat rotation cropland areas in China.Recent anomalous climate changes and air pollutions have simultaneously increased the complexity of land surface processes over the rice-wheat rotation croplands in the HRB.To cope with these changes,this paper established a high-quality four-dimensional water,heat and carbon database by integration and assimilation with data gap-filling methods,quality control technologies and a variety of mathematical statistics methods,to reveal the spatio-temporal characteristics of carbon budget,energy balance and water cycle together with their driving forces in the HRB.This paper provides convincing evidence that the dataset has potential for improving boundary layer parameterization schemes.Present work aims to further reveal the characteristics of land-atmosphere interaction processes,water-heat balance,as well as their environmental effect mechanisms over the rice-wheat rotation croplands in the HRB.Key findings are summarized as follows:（1）A high precision and high spatial-temporal resolution Gross Primary Productivity（GPP）upscaled method and a long-time carbon flux dataset were developed by integrating machine learning algorithms,in situ observations and multi-source satellite remote sensing data.Meanwhile,the Moderate Resolution Imaging Spectroradiometer（MODIS）GPP product for the rice and wheat rotation cropland in eastern China were calibrated.On this basis,the main drivers of carbon flux over the rice-wheat rotation cropland in eastern China were revealed in terms of vegetation phenology,meteorological conditions,and land-air interactions.Compared to support vector machine models,decision tree regression models,deep belief network models and artificial neural network models,the Random Forest（RF）model showed better simulation performance(with a coefficient of determination of 0.88 and a root-mean-square error of 3.04g C m^-2 d^-1).Therefore,it was deemed reliable to upscale eddy covarianced based GPP(GPP_EC)to regional scales through the RF model.The upscaled cumulative seasonal GPP was higher for rice growing season(924 g C m^-2)than that for wheat growing season(532 g C m^-2).In addition,RF model could well identify the feature importance of the predictor variables.Vegetation phenology was the dominant factor regulating the seasonal dynamics of the GPP over the rice-wheat cropland.The comparison between GPP predicted by MODIS(GPP_MOD)and GPP_ECshowed that GPP_MOD performed better during the crop rotation periods but severely underestimated during the rice/wheat active growing seasons.Therefore,GPP_MOD was calibrated by RF based GPP(GPP_RF),and the error range ofΔGPP(GPP_RF minus GPP_MOD)was found to be 2.5～3.25 g C m^-2 d^-1 for rice growing season and 0.75～1.25 g C m^-2 d^-1 for wheat growing season.The upscaled GPP_RF products demonstrate the potential to be applied in accurately assessing MODIS-based agroecosystem carbon balance on a regional or even global scales.（2）The multi-temporal characteristics of atmospheric CO₂ concentration over the rice-wheat rotation cropland of eastern China was quantitatively assessed.Meanwhile,the influence of meteorological conditions,air pollutants,as well as regional transport on the atmospheric CO₂ concentrations were further revealed.Atmospheric CO₂ concentrations showed significant annual,monthly,and diurnal patterns.The annual mean atmospheric CO₂ concentration decreased by about 2%in 2018 when compared to that in 2015.The atmospheric CO₂concentrations exhibited monthly dynamics,peaking in February（444 ppm）and reaching its lowest value in July（363 ppm）.Nocturnal maximum CO₂ concentrations and Mid-afternoon minimum CO₂ concentrations was mainly modulated by photosynthesis and respiration of crops,atmospheric boundary layer height and the intensity of turbulent mixing.In addition,meteorological conditions（wind,temperature,humidity）and atmospheric pollutants(O₃,NOx and PM_2.5)play key roles in the seasonal variations of atmospheric CO₂concentrations.Temperature is the dominant meteorological factor regulating atmospheric CO₂ concentration in spring and autumn,while relative humidity is the main meteorological factor influencing atmospheric CO₂ concentration in summer and winter.Atmospheric CO₂ has a significant positive correlation with O₃(NOx and PM_2.5)in winter（spring,summer,and autumn）.High atmospheric CO₂ concentrations are significantly influenced by long-distance transports in spring and by short-distance transports in other seasons.（3）The seasonal and interannual variabilities of energy and water exchanges were quantitatively analyzed under two different crop environments-flooded and aerobic soil conditions-using three years（December 2014 to November 2017）of eddy covariance observations over a rice-wheat rotation in the HRB.Daytime latent heat flux（LE）and nocturnal surface soil heat flux（G₀）were the main consumers of net radiation（R_n）across the rice-wheat rotation.Averages over the three cropping seasons suggested that rice paddies had 52%more LE than wheat fields,whereas wheat fields had 73%more H than rice paddies.This resulted in a lower Bowen ratio in the rice paddies（0.14）than that in the wheat fields（0.4）.As eddy covariance observations of turbulent heat fluxes are less than the available energy（R_n-G₀）,energy balance closure（EBC）therefore does not occur.EBC varied between 0.78-0.97 over the rice paddies,while for wheat fields it varied 0.59-0.95.Across the whole rice-wheat rotation,the average daily ET were 3.6 mm d^-1 and 2.4 mm d^-1 for the rice paddies and wheat fields,respectively.The average seasonal ET over the rice paddies and wheat fields were 473and 387 mm,indicating a higher water consumption for rice.Water use efficiency（WUE=GPP/ET）couples the carbon and water cycles,with overall WUE slightly higher in wheat fields than that in rice paddies(2.3 vs 2.2 g C kg^-1H₂O).（4）Surface energy partitioning,carbon budget and water cycle under different PM_2.5pollution levels were compared and analyzed over the rice-wheat rotation cropland of the HRB in Spring.The land-atmosphere heat exchange process significantly weakened during the pollution period,i.e.,compared to clean days,daytime R_n on heavily（lightly）polluted days decreased by 16.1%（4.4%）,LE decreased by 45%（19%）,H decreased by 29%（25%）,and soil heat flux at 5 cm depth decreased by 31%（28%）.As PM_2.5 pollution could weaken solar radiation and further reduce vegetation photosynthesis,net ecosystem exchange increased by57.5%（47.5%）on heavily（lightly）polluted days,vegetation respiration increased by 36.2%（32.6%）,GPP decreased by 39.3%（20.7%）,ET decreased by 50%（30%）,and WUE increased by 24%（3.4%）.