Predicting relative species composition within mixed conifer forest pixels using zero-inflated models and Landsat imagery
详细信息 查看全文
- 作者：Shannon L. Savagea ; shannon.savage@msu.montana.edu" class="auth_mail" title="E-mail the corresponding author ; Rick L. Lawrencea ; John R. Squiresb
- 关键词：Canopy cover mapping ; Generalized linear model ; Logistic regression ; Operational Land Imager ; randomForest ; Support vector machine ; Thematic Mapper ; Zero-inflated data
- 刊名：Remote Sensing of Environment
- 出版年：2015
- 出版时间：15 December 2015
- 年：2015
- 卷：171
- 期：Complete
- 页码：326-336
- 全文大小：3316 K
文摘
- •
We predicted percent canopy cover by species using zero-inflated models.
- •
We tested RandomForest, support vector machines, and generalized linear methods.
- •
Mean differences ranged from − 3.84% to 2.26%, depending on species.
- •
RMSEs ranged from 11.26% to 22.98%, depending on species.
- •
Zero-inflated models were highly successful at predicting absence (or zero data).