Abstract Small stands of mangrove trees are difficult to detect and monitor using satellite remote sensing because the width of the narrow strips of vegetation are typically much smaller than the spatial resolution of the imagery. Every mangrove pixel also contains water and bare soil reflectance. Linear spectral unmixing, which estimates the fractional presence of specific land cover types per pixel, was performed on Landsat 8 imagery to detect mangroves on the eastern shoreline of the Bay of La Paz on the Baja California Peninsula of Mexico. Low-altitude aerial imagery collected from a DJI Mavic Pro drone was used as ground-reference data in the accuracy assessment. Continuous fractional presence of mangroves was detected with 80% accuracy and 85% of mangrove area was found. Future work will use linear spectral unmixing to systematically monitor mangrove extent and health in the region relative to expected growth in tourism development.

中文翻译：

---

使用线性光谱解混技术监测墨西哥南下加利福尼亚州的狭窄红树林

摘要 小红树林林分难以使用卫星遥感进行检测和监测，因为窄带植被的宽度通常远小于图像的空间分辨率。每个红树林像素还包含水和裸土反射率。在 Landsat 8 图像上执行线性光谱分解，估计每个像素特定土地覆盖类型的部分存在，以检测墨西哥下加利福尼亚半岛拉巴斯湾东部海岸线上的红树林。从 DJI Mavic Pro 无人机收集的低空航拍图像被用作精度评估中的地面参考数据。以 80% 的准确率检测到连续部分存在的红树林，并发现了 85% 的红树林面积。