最大熵模型的巴基斯坦遗址预测分布研究

针对巴基斯坦的遗址分布问题,该文通过对巴基斯坦的遗址分布进行预测,基于674处遗址点,选取DEM、土壤类型、土地利用类型、离水距等环境变量作为自变量,构建Maxent遗址预测模型,利用Logistic模型验证Maxent模型的精度,并用Kvamme增益值进一步验证两模型的精度。结果表明,印度河流域遗址分布概率高,北部高地地区遗址分布概率低。Maxent模型和Logistic模型都具有较高的准确度,而Maxent模型的增益值远大于Logistic模型,模型精度更高;在局部尺度下,Maxent模型表现更加稳定,其增益值基本保持不变,而Logistic模型在不同的子区域内增益值并不稳定。Maxent模型对于小规模数据集的预测效果也优于Logistic模型。

The study of predictive distribution of Pakistani sites based on Maxent model

Aiming at the problem of the distribution of the ruins in Pakistan,this paper predicted the distribution of the ruins in Pakistan,based on 674 sites,environmental variables such as DEM,soil type,land use type,water separation distance were selected as Variables,Maxent site prediction model was constructed,Logistic model was used to verify the accuracy of Maxent model,and Kvamme gain value was used to further verify the accuracy of the two models.The results showed that the distribution probability of the sites in the Indus Valley was high,and the distribution probability of the sites in the northern highlands was low.Both the Maxent model and the Logistic model had higher accuracy,while the gain value of the Maxent model was much larger than the Logistic model,and the model has higher precision.At the local scale,the Maxent model was more stable and its gain value remained basically unchanged.The logistic model was not stable in different sub-regions.The Maxent model was also superior to the Logistic model in predicting small-scale data sets.