| Trunk sap flow measurement is widely used to estimate water consumption by canopy transpiration.However,the current application of calorimetry to sap flow measurement has limitations such as single mode and narrow application range,and there are insufficient innovations in applied research.Therefore,a stem flow measurement system is designed and developed to explore the response relationship of stem flow to air temperature,relative humidity and wind speed,analyze the change of stem flow characteristics under the attack of stem borers,and put forward the application of vector autoregressive VAR model in predicting stem flow,which provides reference for guiding the prevention of tree diseases and insect pests and water-saving irrigation in modern agriculture.The research analyzed the characteristics of heat pulse HPV and heat dissipation TDP methods,and independently developed a dual-mode trunk sap flow detection device through improved heat pulse technology.In order to improve the accuracy,scientificity and reliability of the stalk flow meter monitoring system,a balance weighing method was used to test the sap section,to calibrate the sap flow model parameters,and to compare the accuracy of the measurement results of different diameter probes.The results show that when the liquid flow rate is large,the change trend of the liquid flow rate measured by the HPV method is closer to the actual change trend,and the use of a smaller diameter probe can minimize the measurement error.The research uses a stem flow meter to monitor the dynamic changes of the sap flow of the street tree Sapindus mukorossi,Michelia chapensis,and Ginkgo biloba.The results show that the healthy wood sap flow of various tree species is significantly related to factors such as temperature,humidity,and saturation water vapor pressure deficit,while the change of the damaged wood sap flow is more affected by environmental factors weak.The daily pattern of stem sap flow of wood damaged by stem borers changed significantly,and the peak sap flow was lower than that of healthy wood.The consistency index D between damaged wood and healthy wood was 0.90,0.91,0.94,RMSE was 8.56,4.62,2.59 cm·h-1,and MAE was 5.53,3.15,1.7cm·h-1.The research uses the VAR model to predict the trunk sap flow,and introduces the predicted value of the Elman neural network as an exogenous variable.The results show that the VAR-Elman mixed models have passed the model stability test,and the coefficients of determination are all above 0.97.The VAR-Elman mixed model predicts and evaluates overall obeys 0.75 <NSE ≤ 1 and 0 ≤ RSR ≤ 0.5.Compared with the traditional single model,the VAR-Elman hybrid model makes up for the shortcomings of mid-term forecasting and can improve the overall forecast accuracy.The application of the VAR model to the prediction of tree trunk sap flow provides an explanation basis for the correlation between variables,so that the prediction model can be extended to the study of forest transpiration water consumption within a region. |
