比较建模图层之间的差异

## 注意这里尽量是相同空间序列的不同时间图层；
pa1 <- raster(ass )
pa1[] <- ifelse(ass[] >= 377, 1, 0)
pa2 <- raster(asg)
pa2[] <- ifelse(asg[] >= 377, 1, 0)
chp <- pa2 - pa1
plot(chp,col=c('red','gray','blue'))

SDM分类阈值可视化

order(as.data.frame(aucs$Training.AUC),decreasing=T)
ped_sa <- predict(mx_sa@models[[1]], envs_sa)

## 基于ENMTOOLS计算生态位宽度：

raster.breadth(ped_sa)

plot(ped_sa)
## 查看评估结果：
mean(aucs$Maximum.training.sensitivity.plus.specificity.Cloglog.threshold)

ef_sa <- reclassify(ped_sa,rcl= c(0,0.27,NA,0.27,Inf,1))

## 重分类：

setwd("C:/Users/Administrator/Desktop/")
bg_enm_sa <- dismo::randomPoints(ef_sa,3000,p=xh_sa[,2:3])
bg_enm_sa <- as.data.frame(bg_enm_sa)

write.csv(bg_enm_sa,"./sa_enm_bg.csv")

writeRaster(ped_sa,"./sa_sdm.tif",format="GTiff")


## 改进的二分类方法：
library(rgdal)
library(rgeos )
asrange <- readOGR('E:/环境数据/亚洲shp/euisa.shp')
plot(asrange)

pa1 <- raster(ass )
pa1[] <- ifelse(ass[] >= 377, 1, 0)
pa2 <- raster(asg)
pa2[] <- ifelse(asg[] >= 377, 1, 0)
chp <- pa2 - pa1
tiff(file = "C:/Users/admin/Desktop/as.tiff", res = 300, width = 3000, height =3000, compression = "lzw")

plot(chp,col=c('green','white','yellow'))
# plot(map, border="red", lwd=3)
plot(asrange,add=T,border="#33330033", lwd=0.5)
dev.off()