Estimación del contenido de humedad de la vegetacion herbácea en una zona de dehesa a partir de imágenes hiperespectrales adquiridas por el sensor aeroportado CASI
[EN] In this study we analyzed the potential of airborne hyperspectral images acquired by the CASI sensor (Compact Airborne Spectrographic Imager) to estimate water content in the grass layer of a dehesa ecosystem. It was estimated by means of biophysical variables as Canopy Water Content (CWC), Fue...
| Authors: | , , , |
|---|---|
| Format: | article |
| Status: | Published version |
| Publication Date: | 2015 |
| Country: | España |
| Institution: | Consejo Superior de Investigaciones Científicas (CSIC) |
| Repository: | DIGITAL.CSIC. Repositorio Institucional del CSIC |
| OAI Identifier: | oai:dnet:digitalcsic_::640bf411d7460758078351862218c0d9 |
| Online Access: | http://hdl.handle.net/10261/154649 |
| Access Level: | Open access |
| Keyword: | Hyperspectral CASI Water content Herbaceous vegetation Reflectance Spectral índices Sampling Hiperespectral Humedad Vegetación herbácea Reflectividad Correlaciones Índices espectrales Muestreo Soil water |
| Summary: | [EN] In this study we analyzed the potential of airborne hyperspectral images acquired by the CASI sensor (Compact Airborne Spectrographic Imager) to estimate water content in the grass layer of a dehesa ecosystem. It was estimated by means of biophysical variables as Canopy Water Content (CWC), Fuel Moisture Content (FMC) and Equivalent Water Thickness (EWT) calculated from vegetation samples collected simultaneously to image acquisition. Two different field sampling strategies have been tested. First method collects all plants in 25x25cm quadrants and selects a fraction of the sample to be processed in the laboratory. In the second method the worker selects a small sample in the field. This subsample should be representative of the main one and will be completely processed in the laboratory. Afterwards linear relationships between the biophysical variables and twenty spectral indices from literature and from Normalized Difference Indices (NDIs-combination all sensor bands) are analyzed. The results show that the combination of specific bands in the form of NDI performs better than the spectral indices proposed in the literature. The regression models showed maximum correlation values (R2max) of 0.78 in the estimation of water content from NDIs. This value decrease to 0.53 for indices proposed in the literature. Clear differences were found in the correlation values when comparing the different sampling strategies. In general, lower correlation values were found when sampling methods require subjective decision on the selection in the field of plants that composes the sample. Despite of the limited outreach of the method used, the results suggest that this could be successfully applied in circumstances when information is limited and variables are not physically related. |
|---|