【摘要】 运用遥感数据,以黑龙江省三江国家级湿地自然保护区和挠力河国家级自然保护区为例,建立线性回归模型、多项Logit模型、双重差分模型、倾向得分双重差分模型评估自然保护区的保护效果。通过对模型结果进行对比发现,相较于其他3种模型,倾向得分双重差分模型能够减少选择性偏差,以及直接采用双重差分模型产生的混杂偏误,从而得到更为真实的保护效果。倾向得分双重差分模型的结果表明,建立保护区能够有效保护湿地,且在1%的水平下显著,建立保护区比不建保护区湿地率增加6. 18%。更多还原

【Abstract】 Based on remote sensing data,a linear regression model,multiple Logit models,difference-indifference model,and propensity score matching-difference in difference model are established to evaluate the conservation effectiveness of the Sanjiang National Wetland Nature Reserve and the Naolihe National Nature Reserve in Heilongjiang Province. The results showed that comparing with the other three models,the propensity score matching-difference in difference model can reduce the selection bias and the confounding bias,and obtaining a more realistic result. The result of the propensity score matching-difference in difference model shows that the establishment of nature reserve caused an increase of wetland area,which is significant at the level of 1%,and the wetland rate increased by 6. 18% compared with that without the reserve.更多还原