Effect Evaluation And Algorithm Improvement Of Flux-variance Method For Calculating Sensible Heat Flux On Three Underlying Surfaces

The energy exchange between the earth’s surface and the atmosphere has great effects on climate change,weather evolution and the carbon/water cycle.Sensible heat flux is the main component of energy exchange between the earth’s surface and the atmosphere,also the main way to heat the atmosphere.At present,the eddy covariance technology is been widely used to observe the sensible heat flux between the surface and the atmosphere.While this technology requires expensive instruments and the data post-processing process is complicated,resulting in sparse observation sites and insufficient spatial representation.Therefore new sensible heat flux observation technology with low cost,simple observation principle and accurate calculation algorithm is of urgent need.Based on Monin-Obukhov similarity theory,the flux-variance method only needs temperature standard deviation observation and atmospheric stability parameter to calculate sensible heat flux.So this method has the great potential for large spatial scale observation.However,the calculation performance of this method on different underlying surfaces needs to be evaluated against eddy covariance technology.The selection of temperature standard deviation and the judgment of atmospheric stability need to be determined through field experiments.In this study,high-frequency temperature observations from six kinds of sensors were compared to eddy covariance system on three underlying surfaces:agricultural pond,rice field and bare land.Firstly,the statistical characteristics of atmospheric turbulence,especially the similarity of variance,on the three underlying surfaces were analyzed.Secondly,the performance of six high-frequency temperature sensors for temperature standard deviation observation is quantitatively evaluated.The"flux-variance"method calculation performance is evaluated on three underlying surfaces against eddy covariance observation.Finally,the flux-variance method is improved from the three aspects:calculation algorithm,atmospheric stability judgment and calculation time scale.The main conclusions obtained are summarized as follows:（1）There are obvious differences in the statistical characteristics of turbulence on the three underlying surfaces of aquacultual pond,rice paddy and bare ground.The proportion of atmospheric instability over aquaculture ponds（74%）is higher than that over paddy field（33%）and bare ground（40%）.The instability intensity over bare ground during the daytime is higher than those over other two surfaces.80%of flux footprint area located within the target underlying surfaces.The turbulence spectrum of the three-dimensional wind speeds follow the-2/3 power law in the inertial subrange.The cospectrum of the vertical wind speed and the scalar（temperature,humidity and CO₂concentration）follow the-4/3 power law in the inertial subrange.Therefore,eddy covariance system can be used to observe the flux exchange between the three underlying surfaces and atmosphere.As wind speed increasing,the turbulence intensity and turbulence energy decrease and approach constants under high wind speeds.And turbulence intensity and turbulence energy over paddy field are higher than those over fish pond and bare ground under high wind speed.The flux-variance similarity is applicable to the three underlying surfaces.The normalized standard deviation of the three-dimensional wind speeds on the three underlying surfaces changes with the atmospheric stability follows-1/3 power law.And the relationship is better under unstable conditions,especially for vertical wind speed.The normalized standard deviation of temperature and water vapor density follow-1/3 power law when atmosphere is unstable.The normalized CO₂ density on the paddy field and bare underground surface follow the-1/3 power law under unstable condition.（2）The temperature mean measurement performance of six temperature sensors is comparable.The temperature standard deviation measurement performance of the two-dimensional sonic anemometer is obviously better than that of the thermocouples.The flux-variance method is more sensitive to the parameter c _T1 when calculating the sensible heat flux.A lower value of c _T1 will lead to an overestimation of the sensible heat flux.A 10%increase or decrease in c _T1 will reduce or increase the sensible heat flux by 13.32%,17.12%,respectively.Compared with the default parameters,using the regression parameters obtained by field study will improve the calculation of sensible heat,with root mean square error decreasing by 14.6 W m^-2.With eddy covariance observation as a reference,flux-variance method shows best performance for calculation sensible heat flux over rice fields with correlation coefficient of 0.887,followed by bare land（correlation coefficient of 0.860）and agricultural pond（correlation coefficient of 0.823）.（3）The improved flux-variance iteration algorithm only needs temperature standard derivation and wind speed observations when calculating sensible heat flux.The iteration algorithm shows a little worse performance compared to the original algorithm.On average,the correlation coefficient between the calculation and the observation reduced by 0.056 and the root mean square error increased by 2.9 W m^-2.Compared to the original method,iteration algorithm calculation lost 11.7%,36.4%,and 32.3%data over aquacultural pond,paddy field and bare land,respectively for non-convergence.Using temperature gradient for stability judgment is better than using net radiation.The judgment error for the sensible heat flux direction of the two methods is 12.9%and 34.8%,respectively.Using the original method to calculate the sensible heat flux on the daily scale is more consistent with the eddy covariance observation,while the iterative method is less consistent with the observation due to data lost during iteration.