Landscape Aesthetic Evaluation And Management Technology Research Of Triadica Cochinchinensis Loureiro Forest

Sapium discolor(Triadica cochinchinensis Loureiro)has high landscape value as a native tree species in China,but little research has been done on its landscape aesthetic evaluation,and it is impossible to propose targeted management transformation techniques.A landscape aesthetic evaluation of Sapium discolor forest based on the Scenic Beauty Estimation(SBE)method was conducted to determine the relationship between different landscape elements and the landscape aesthetic quality of Sapium discolor forest.Based on the feature factors obtained from the Random Forest(RF)importance screening method,a beauty estimation model was constructed by combining Multiple Linear Regression(MLR)and RF Regression,and the accuracy was evaluated based on the actual beauty score values.The measured data of crown width and Diameter at Breast Height(DBH)of 2086 plants per tree on 40 standard plots of Sapium discolor were analyzed,and the relationship between crown width and DBH ratio of Sapium discolor in the experimental area was obtained.The main outcomes of the study were as follows:(1)The in-forest landscape evaluation model based on the SBE method showed that the landscape elements that contributed most to the landscape beauty of Sapium discolor forests were,in order,depression(23.7%),average tree height(19.2%),cover(17.2%),shrub layer height(12.1%),plant density(10.2%),species composition(7.3%),crown width(5.6%),and branch height(4.2%).Depression of Sapium discolor stands was the most influential factor on public preference,and stands with moderate depression were preferred by more people.Sapium discolor stands with good growth conditions,dense foliage and vigorous growth were more ornamental.Increasing the cover of the stand,reducing the height of the shrub layer,and maintaining a moderate plant density and dominant species proportion will result in greater SBE values and more popularity among the public.(2)MLR model and RF regression model were constructed based on the actual rating value of beauty degree and the feature factors screened by RF importance,respectively,to obtain the estimated value of beauty degree.The accuracy analysis was conducted by combining the actual rating value of beauty degree.The results show that the RF’s Root Mean Square Error(RMSE)=13.11,which is 4.02%lower than the MLR model’s RMSE=13.66,i.e.,the random importance ranking method can effectively screen the feature factors and achieve high accuracy estimation.The coefficient of determination of the RF regression model,R2=0.94,increased by 2.17%compared with the coefficient of determination of the MLR model,R2=0.92,and the mean value of its beauty estimation increased by 0.83,i.e.the RF regression model combined with the importance screening of the feature factors has better fitting effect and estimation accuracy.The analysis combined with the actual scores of beauty degree shows that the RF regression model has stronger stability compared with the MLR model.(3)DBH is the easiest factor to measure and can best reflect the growth condition of the stand.The relationship between plant density and scenic beauty obtained from the in-forest landscape evaluation model was combined with the results showing that the highest scenic beauty value was achieved when the average diameter at breast height of Sapium discolor forest was equal to 20 cm,the density of plant density was around 800 plants/hm2 with a denseness of 0.8.(4)The current situation of Xishan state-owned forest was analyzed,and the overall landscape effect was poor due to the problems of too homogeneous levels of plants,poor plant growth,monotonous colors,and sparse understory vegetation in some areas.Targeted management transformation measures were proposed to address the existing problems,i.e.,artificial pruning,understory ground cover replanting,and nurturing interval.