iBenthos Seagrass Mapper
Sign in with your Google account to access the mapping platform.
Sign in with GoogleBy signing in, you agree to the iBenthos terms of service. Your Google account information will be used to identify your session.
Sign in with your Google account to access the mapping platform.
Sign in with GoogleBy signing in, you agree to the iBenthos terms of service. Your Google account information will be used to identify your session.
Give this configuration a name so you can identify it later.
This mapping session isn't saved automatically. Save the configuration or export the probability/label map before you navigate away.
Binary seagrass mapping from M1 presence labels. Produces a probability surface from 0 to 1 and thresholded present/absent layers.
Random Forest regression trained from M2 SG_COVERAGE. Produces
continuous estimated seagrass cover from 0% to 100%.
Uses M2 total cover and M4 morphology fractions. Produces percentage layers for SG_OV, SG_CY, SG_STM, SG_STP_M, SG_STP_N, SG_STP_T, SG_UKN and OT. The eight percentages are non-negative and sum to 100% at each mapped pixel.
Uses the Australian M7 taxonomy, which adds SG_FER to the eight-class set. This task is currently unavailable because M7 has no usable training rows.
Unavailable: M7 training data unavailable.
Corrects the red-band turbidity signal for bottom reflectance in shallow, visible-bottom water using satellite-derived bathymetry, so bright shallow substrate (sand, reef, seagrass) isn't misread as turbid.
Li, J., Carlson, R. R., Knapp, D. E., & Asner, G. P. (2022). Shallow coastal water turbidity monitoring using Planet Dove satellites. Remote Sensing in Ecology and Conservation, 8(4), 521–535. https://doi.org/10.1002/rse2.259
Direct single-band red-reflectance turbidity retrieval calibrated for coastal and estuarine waters, with no bottom-reflectance correction.
Dogliotti, A. I., Ruddick, K. G., Nechad, B., Doxaran, D., & Knaeps, E. (2015). A single algorithm to retrieve turbidity from remotely-sensed data in all coastal and estuarine waters. Remote Sensing of Environment, 156, 157–168. https://doi.org/10.1016/j.rse.2014.09.020
An ensemble of decision trees trained sequentially by gradient boosting: each tree corrects the residual errors of the ensemble so far, shrunk by the learning rate. It outputs a per-pixel seagrass probability and is the default classifier. Key parameters: number of trees, learning rate (shrinkage), and maximum leaf nodes per tree.
Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. The Annals of Statistics, 29(5), 1189–1232. https://doi.org/10.1214/aos/1013203451
An ensemble of decision trees trained independently on bootstrap samples, whose votes are averaged. Key parameters: number of trees, bagging fraction, and minimum samples per leaf.
Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5–32. https://doi.org/10.1023/A:1010933404324
A margin-based classifier (RBF kernel) that separates seagrass from non-seagrass in the spectral/embedding feature space.
Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine Learning, 20(3), 273–297. https://doi.org/10.1007/BF00994018