1 Background

In this exercise, we will generate a wall-to-wall canopy height map for a selected AOI by combining ICESat-2 reference canopy height observations (RH98) with AlphaEarth embedding predictors and terrain covariates from Google Earth Engine (GEE). The workflow demonstrates how sparse spaceborne LiDAR samples can be used to train a machine-learning model and extrapolate canopy height continuously across space.

2 Objective

The primary aim of this hands-on exercise is to produce a wall-to-wall canopy height raster for a selected Area of Interest (AOI) using ICESat-2 as reference data and AlphaEarth embeddings + terrain predictors for spatial extrapolation.

Participants will learn how to: - Use ICESat2VegR to work with ICESat-2 data (ATL03 + ATL08). - Create canopy height reference data by computing RH98 at 20 m segments. - Integrate ICESat-2 reference data with GEE predictors (AlphaEarth embeddings + terrain). - Train and validate a Random Forest model in R. - Produce a wall-to-wall canopy height prediction in GEE and export a GeoTIFF. - This end-to-end workflow is implemented in R and relies on open tools, enabling reproducible canopy height mapping for regional applications.

3 Methodology

The exercise follows a sequential model-based mapping workflow. First, ICESat-2 observations are used to derive reference canopy height information (RH98) at 20 m segments by combining the ATL03 photon data with the ATL08 land and vegetation classifications. These segments represent sparse but accurate measurements of canopy height along satellite tracks. Next, environmental predictor variables are obtained from Google Earth Engine, including AlphaEarth embedding layers and terrain attributes. Predictor values are extracted at the ICESat-2 segment locations to build a training dataset linking canopy height observations with explanatory variables. A Random Forest regression model is then trained and validated using the sampled ICESat-2 segments. Finally, the trained model is applied to the predictor mosaic to generate a continuous, wall-to-wall canopy height map across the entire study area. Together, these steps demonstrate how sparse spaceborne LiDAR measurements can be combined with environmental predictors and machine learning to produce spatially continuous forest structure maps.

4 Input data

Area of Interest polygon (GeoJSON, EPSG:4326)
ICESat-2 ATL03-ATL08 data products
GEE access: terrain and AlphaEarth embedding dataset

5 Expected output

ICESat-2 canopy height reference dataset (20 m segments, RH98) clipped to the AOI
Random Forest model performance report (training vs testing statistics)
Wall-to-wall canopy height raster (GeoTIFF) representing predicted canopy height across the study area
Optional interactive visualization map (HTML)

6 R packages

ICESat2VegR – Tools for discovering, downloading, joining, and processing ICESat-2 ATL03 and ATL08 data and computing canopy height metrics.
rgee - Tool for accesssing geospatial dataset
sf – Handles vector spatial data and AOI geometry operations.
terra – Spatial raster/vector operations and extraction of predictor values.
data.table – Efficient manipulation of large ICESat-2 segment tables.
dplyr – Data manipulation and preparation of modeling datasets.
reticulate – Interface between R and Python required for Earth Engine access.
rgee – R interface to Google Earth Engine for predictor generation and mapping.
randomForest – Random Forest regression modeling.
caret – Data partitioning and model evaluation utilities.
ggplot2 – Visualization of model diagnostics and variable importance.
openxlsx – Exporting model accuracy tables to Excel.
htmlwidgets – Exporting optional interactive maps.

7 Workflow

This document starts at Step 1 of your workflow:

Load GEE + modeling libraries
Initialize Earth Engine
Load existing ICESat-2 training segments (GeoJSON)
Build AlphaEarth embedding + terrain predictor stack
Train Random Forest to predict RH98
Create wall-to-wall canopy height map in GEE
Export interactive HTML map (optional)
Export GeoTIFF via Google Drive (ee_as_rast)

8 Step 1 — Loading R packages

In R Markdown, rm(list = ls()) can break the document if used mid-way.
Since this Rmd starts here, we can safely clean at the beginning.

suppressPackageStartupMessages({
  library(reticulate)
  library(rgee)
  library(sf)
  library(terra)
  library(data.table)
  library(randomForest)
  library(dplyr)
  library(ggplot2)
  library(caret)
  library(openxlsx)
  library(ICESat2VegR)
  library(htmlwidgets)
})

print("Verify the python configuration ...")

## [1] "Verify the python configuration ..."

py_config()

## python:         C:/Users/calvi/OneDrive/Documentos/.virtualenvs/r-reticulate/Scripts/python.exe
## libpython:      C:/Users/calvi/AppData/Local/r-reticulate/r-reticulate/pyenv/pyenv-win/versions/3.9.13/python39.dll
## pythonhome:     C:/Users/calvi/OneDrive/Documentos/.virtualenvs/r-reticulate
## version:        3.9.13 (tags/v3.9.13:6de2ca5, May 17 2022, 16:36:42) [MSC v.1929 64 bit (AMD64)]
## Architecture:   64bit
## numpy:          C:/Users/calvi/OneDrive/Documentos/.virtualenvs/r-reticulate/Lib/site-packages/numpy
## numpy_version:  1.24.3
## ee:             C:\Users\calvi\OneDrive\DOCUME~1\VIRTUA~1\R-RETI~1\lib\site-packages\ee
## 
## NOTE: Python version was forced by VIRTUAL_ENV

9 Step 2 — Authenticate/initialize Google Earth Engine + build AOI geometry

# Load your helper functions (must define ee_build_AlphaEarth_embedding_terrain_stack)
#if (!file.exists(BASE_R_PATH)) stop("BASE_R_PATH not found: ", BASE_R_PATH)
source("C:\\Users\\calvi\\OneDrive\\Desktop\\Workshop_ICESat2VegR\\base.R")

if (!exists("ee_build_AlphaEarth_embedding_terrain_stack")) {
  stop("Missing ee_build_AlphaEarth_embedding_terrain_stack() after sourcing base.R")
}

print("Initializing Google Earth Engine...")

## [1] "Initializing Google Earth Engine..."

 reticulate::py_run_string("import ee")
reticulate::py_run_string("ee.Authenticate()")
reticulate::py_run_string('ee.Initialize(project="ee-calvites1990")')


# Load AOI
boundary <- st_read("C:\\Users\\calvi\\OneDrive\\Desktop\\Workshop_ICESat2VegR\\01_InputData\\aoi_4326.geojson")

## Reading layer `aoi_4326' from data source 
##   `C:\Users\calvi\OneDrive\Desktop\Workshop_ICESat2VegR\01_InputData\aoi_4326.geojson' 
##   using driver `GeoJSON'
## Simple feature collection with 1 feature and 2 fields
## Geometry type: MULTIPOLYGON
## Dimension:     XY
## Bounding box:  xmin: -84.71409 ymin: 30.12047 xmax: -84.6435 ymax: 30.17786
## Geodetic CRS:  WGS 84

# Clean AOI geometry for EE
boundary_raw2_valid <- st_make_valid(boundary)
boundary_union <- st_union(boundary_raw2_valid)
boundary_simple <- st_simplify(boundary_union, dTolerance = 0.001)

region <- sf_as_ee(boundary_simple)
boundary_ee <- sf_as_ee(boundary_simple)
geom <- ee$FeatureCollection(boundary_ee)$geometry()

# Output dir
outdir <- "C:\\Users\\calvi\\OneDrive\\Desktop\\Workshop_ICESat2VegR\\02_OutputData"
dir.create(file.path(outdir, "01_STATS"), showWarnings = FALSE, recursive = TRUE)

cat("AOI loaded and EE geometry built.\n")

## AOI loaded and EE geometry built.

10 Step 3 — Load ICESat2 datset and ee_build_AlphaEarth_embedding_terrain_stack

# Load existing ICESat-2 training segments
data_raw <- read_sf("C:\\Users\\calvi\\OneDrive\\Desktop\\Workshop_ICESat2VegR\\02_OutputData\\02_SEG\\ATL03_ATL08_20250826092226_20.geojson")
class(data_raw)

## [1] "sf"         "tbl_df"     "tbl"        "data.frame"

set.seed(1)
df <- dplyr::slice_sample(data_raw, n = 4000, replace = TRUE)

i <- 1
yr <- 2025

# Build predictor stack and extract predictors at 20 m segments
predictors2 <- ee_build_AlphaEarth_embedding_terrain_stack(geom, yr, yr)
sampling_df <- ICESat2VegR::seg_ancillary_extract(predictors2, terra::vect(df), 20)

# Quick check
head(sampling_df)

##           A00         A01        A02          A03         A04         A05
##         <num>       <num>      <num>        <num>       <num>       <num>
## 1: 0.07972318 -0.11909266 -0.1137409 -0.108512111 -0.18608228 -0.33685506
## 2: 0.14173010 -0.10340638 -0.2069358  0.024605921 -0.13588620  0.07111111
## 3: 0.09356401 -0.06299116 -0.1358862 -0.013840830 -0.21413303 -0.26795848
## 4: 0.10851211 -0.15378700 -0.1600000  0.048227605 -0.14173010  0.00615148
## 5: 0.13588620 -0.01777778 -0.1034064  0.032541330 -0.09842368  0.03543253
## 6: 0.09356401 -0.08421376 -0.1245675  0.007443291 -0.21413303 -0.29287197
##            A06          A07         A08          A09         A10          A11
##          <num>        <num>       <num>        <num>       <num>        <num>
## 1:  0.05536332 -0.051733948 -0.11909266  0.035432526  0.12456747 -0.051733948
## 2:  0.01384083  0.084213764  0.04158401 -0.022206844  0.03254133  0.103406382
## 3:  0.07972318  0.022206844 -0.06299116  0.013840830  0.08421376  0.008858131
## 4: -0.07972318 -0.015747789  0.02712803 -0.071111111 -0.02220684  0.103406382
## 5: -0.03844675 -0.008858131  0.08421376 -0.048227605 -0.04158401  0.141730104
## 6:  0.05173395 -0.006151480 -0.07535563  0.003014225  0.10340638 -0.015747789
##           A12         A13           A14         A15        A16         A17
##         <num>       <num>         <num>       <num>      <num>       <num>
## 1: -0.1085121 -0.07972318 -0.1245674740 -0.07111111 -0.1663360 -0.12456747
## 2: -0.1417301  0.17937716  0.0797231834 -0.04822760 -0.1793772 -0.10340638
## 3: -0.1476970 -0.02712803 -0.0005536332 -0.08421376 -0.2441522 -0.11374087
## 4: -0.1476970  0.07972318  0.0629911572 -0.05536332 -0.2141330 -0.08882737
## 5: -0.1190927  0.15378700  0.0482276048 -0.02977316 -0.1929104 -0.09842368
## 6: -0.1537870 -0.05911572 -0.0448442907 -0.04822760 -0.2288966 -0.15378700
##             A18          A19         A20          A21         A22         A23
##           <num>        <num>       <num>        <num>       <num>       <num>
## 1: -0.153787005 -0.019930796 -0.10340638  0.113740869 -0.17937716  0.13016532
## 2:  0.186082276 -0.012056901 -0.04158401 -0.075355632  0.05173395 -0.05911572
## 3: -0.003936947 -0.029773164 -0.07972318  0.041584006 -0.11909266  0.02977316
## 4:  0.153787005 -0.029773164 -0.07111111 -0.024605921  0.01384083 -0.04822760
## 5:  0.108512111 -0.079723183 -0.14173010 -0.004982699 -0.01205690  0.00615148
## 6: -0.062991157 -0.008858131 -0.11909266  0.066989619 -0.15378700  0.04822760
##            A24       A25         A26        A27         A28       A29
##          <num>     <num>       <num>      <num>       <num>     <num>
## 1:  0.04822760 0.2214533 -0.06299116 0.05173395  0.05173395 0.1537870
## 2: -0.04822760 0.2844444 -0.02460592 0.06698962 -0.02460592 0.1663360
## 3:  0.03844675 0.2599000 -0.02460592 0.05536332  0.01384083 0.1476970
## 4: -0.17937716 0.3188927  0.01574779 0.08882737  0.04822760 0.2141330
## 5: -0.19291042 0.2679585 -0.04158401 0.07535563  0.06299116 0.2679585
## 6:  0.02977316 0.2441522 -0.05173395 0.04484429  0.02977316 0.1929104
##           A30         A31        A32        A33         A34         A35
##         <num>       <num>      <num>      <num>       <num>       <num>
## 1: 0.08882737 -0.06698962 -0.1793772 0.01384083  0.00615148 -0.17937716
## 2: 0.20693579  0.07972318 -0.1663360 0.03844675 -0.15378700  0.03844675
## 3: 0.17279508 -0.01039600 -0.1929104 0.02460592 -0.10340638 -0.07111111
## 4: 0.17937716  0.04158401 -0.1537870 0.07111111 -0.11374087  0.01039600
## 5: 0.18608228  0.09842368 -0.1190927 0.05911572 -0.14769704  0.10851211
## 6: 0.11909266  0.01574779 -0.1727951 0.05536332 -0.07535563 -0.09356401
##             A36         A37       A38         A39         A40        A41
##           <num>       <num>     <num>       <num>       <num>      <num>
## 1: -0.214133026 -0.03543253 0.1301653 -0.30142253 -0.09356401 0.04484429
## 2: -0.199861592 -0.01777778 0.2214533 -0.04158401 -0.15378700 0.07111111
## 3: -0.179377163 -0.06299116 0.1301653 -0.25990004 -0.14769704 0.09842368
## 4:  0.004982699 -0.02460592 0.1727951 -0.03543253 -0.04484429 0.10851211
## 5: -0.051733948  0.03844675 0.1600000 -0.05173395  0.01777778 0.18608228
## 6: -0.172795079 -0.04158401 0.1137409 -0.27613995 -0.13588620 0.11374087
##          A42          A43       A44          A45          A46        A47
##        <num>        <num>     <num>        <num>        <num>      <num>
## 1: 0.1663360 -0.088827374 0.1137409 -0.004982699  0.108512111 -0.2364629
## 2: 0.2141330 -0.002214533 0.1137409  0.055363322 -0.003014225 -0.1085121
## 3: 0.1727951 -0.035432526 0.1476970  0.006151480  0.113740869 -0.1793772
## 4: 0.2761399  0.075355632 0.1537870  0.055363322 -0.062991157 -0.1301653
## 5: 0.2288966 -0.006151480 0.2069358  0.051733948 -0.098423683 -0.1929104
## 6: 0.1417301 -0.032541330 0.1417301  0.004982699  0.119092657 -0.2364629
##          A48         A49        A50        A51         A52          A53
##        <num>       <num>      <num>      <num>       <num>        <num>
## 1: 0.2141330 -0.01993080 -0.1537870 -0.1358862 -0.04822760  0.038446751
## 2: 0.1034064 -0.13016532 -0.1663360 -0.2519646 -0.04158401  0.062991157
## 3: 0.2679585 -0.11909266 -0.1929104 -0.2214533 -0.03254133  0.088827374
## 4: 0.1929104 -0.19291042 -0.2141330 -0.2441522 -0.05911572  0.022206844
## 5: 0.1301653 -0.23646290 -0.2069358 -0.1417301 -0.02712803 -0.002214533
## 6: 0.2519646 -0.07535563 -0.1998616 -0.1860823 -0.07535563  0.088827374
##           A54         A55         A56          A57        A58         A59
##         <num>       <num>       <num>        <num>      <num>       <num>
## 1: 0.08882737 0.093564014  0.03254133 -0.003936947 0.06698962 -0.05173395
## 2: 0.16633602 0.088827374 -0.13588620  0.236462899 0.14769704 -0.17937716
## 3: 0.09842368 0.119092657 -0.01039600  0.088827374 0.07972318 -0.12456747
## 4: 0.19291042 0.003014225 -0.10851211  0.172795079 0.13588620 -0.19291042
## 5: 0.14769704 0.055363322 -0.13588620  0.172795079 0.20693579 -0.19986159
## 6: 0.08882737 0.088827374  0.01039600  0.041584006 0.10340638 -0.09842368
##             A60         A61         A62        A63    aspect   beam elevation
##           <num>       <num>       <num>      <num>     <num> <char>     <num>
## 1:  0.032541330 -0.11909266 -0.06299116 0.10851211  936.1740   gt1l  2.963774
## 2:  0.088827374 -0.05536332  0.11374087 0.03254133 2343.6733   gt1r  6.954786
## 3: -0.013840830 -0.17937716 -0.03543253 0.09356401 1385.0271   gt1r  3.566396
## 4:  0.075355632 -0.03844675  0.02460592 0.06698962 3296.8274   gt1l 12.364920
## 5: -0.002214533 -0.01777778  0.07111111 0.04822760  258.9939   gt1l 12.045389
## 6: -0.006151480 -0.16633602 -0.05911572 0.09356401  672.9902   gt1l  3.080051
##         lat latitude       lon longitude mean_solar n_canopy_total n_ground
##       <num>    <num>     <num>     <num>      <num>          <int>    <int>
## 1: 30.15402 30.15402 -84.67834 -84.67831  -21.63491             11        0
## 2: 30.14054 30.14063 -84.68068 -84.68071  -21.64105             21        0
## 3: 30.14755 30.14746 -84.67996 -84.67995  -21.63772             27        0
## 4: 30.12294 30.12298 -84.68176 -84.68178  -21.65006              3        0
## 5: 30.12869 30.12861 -84.68122 -84.68115  -21.64730              1        0
## 6: 30.14360 30.14360 -84.67942 -84.67947  -21.64000              3        0
##    n_mid_canopy n_top_canopy night_flag2       rh98 segment_id     slope  year
##           <int>        <int>       <int>      <num>      <int>     <num> <int>
## 1:           10            1           1 14.7535931     163810  2.439643  2025
## 2:           14            7           1 34.2744095     332590  4.166642  2025
## 3:           21            6           1 21.3566180     332552 10.272169  2025
## 4:            3            0           1 14.3197423     163983  5.123228  2025
## 5:            1            0           1  0.5029621     163952  2.330704  2025
## 6:            2            1           1 16.9825489     163868  4.446283  2025

names(sampling_df)

##  [1] "A00"            "A01"            "A02"            "A03"           
##  [5] "A04"            "A05"            "A06"            "A07"           
##  [9] "A08"            "A09"            "A10"            "A11"           
## [13] "A12"            "A13"            "A14"            "A15"           
## [17] "A16"            "A17"            "A18"            "A19"           
## [21] "A20"            "A21"            "A22"            "A23"           
## [25] "A24"            "A25"            "A26"            "A27"           
## [29] "A28"            "A29"            "A30"            "A31"           
## [33] "A32"            "A33"            "A34"            "A35"           
## [37] "A36"            "A37"            "A38"            "A39"           
## [41] "A40"            "A41"            "A42"            "A43"           
## [45] "A44"            "A45"            "A46"            "A47"           
## [49] "A48"            "A49"            "A50"            "A51"           
## [53] "A52"            "A53"            "A54"            "A55"           
## [57] "A56"            "A57"            "A58"            "A59"           
## [61] "A60"            "A61"            "A62"            "A63"           
## [65] "aspect"         "beam"           "elevation"      "lat"           
## [69] "latitude"       "lon"            "longitude"      "mean_solar"    
## [73] "n_canopy_total" "n_ground"       "n_mid_canopy"   "n_top_canopy"  
## [77] "night_flag2"    "rh98"           "segment_id"     "slope"         
## [81] "year"

11 Step 3 — Variable selection and choose best predictors

x <- sampling_df %>% dplyr::select(starts_with("A"), "slope", "aspect", "elevation") %>% data.frame()
y <- sampling_df %>% dplyr::select("rh98") %>% data.frame()

sel_rf_rfe <- ICESat2VegR::varSel(x, y$rh98)

imp <- sel_rf_rfe$importance %>%
  mutate(selected = parameter %in% sel_rf_rfe$selvars) %>%
  arrange(importance) %>%
  mutate(parameter = factor(parameter, levels = parameter))

ggplot(imp, aes(x = importance, y = parameter, shape = selected)) +
  geom_point(size = 3) +
  labs(
    x = paste0("Importance (", sel_rf_rfe$scaling, " scale)"),
    y = NULL,
    title = "Variable importance (scaled)",
    subtitle = "Selected variables highlighted"
  ) +
  theme_minimal(base_size = 12)

best_metrics <- sel_rf_rfe$importance$parameter[sel_rf_rfe$importance$importance >= 0.2]
length(best_metrics)

## [1] 21

12 Step 4 — Data splitting and Random Forest fitting

x <- dplyr::select(sampling_df, dplyr::any_of(best_metrics))
y <- sampling_df$rh98
data_i <- data.frame(y, x, check.names = FALSE)

idx_train <- createDataPartition(y = data_i$y, p = 0.7, list = FALSE)
trainData <- data_i[idx_train, , drop = FALSE]
testData  <- data_i[-idx_train, , drop = FALSE]

x_train <- dplyr::select(trainData, dplyr::any_of(best_metrics))
y_train <- trainData$y
x_test  <- dplyr::select(testData,  dplyr::any_of(best_metrics))
y_test  <- testData$y

fit_rf <- ICESat2VegR::fit_model(
  x = x_train,
  y = y_train,
  rf_args = list(ntree = 100)
)

rf_model <- fit_rf$model

pred_train <- predict(rf_model, newdata = x_train)
pred_test  <- predict(rf_model, newdata = x_test)

results_train <- compute_stats(y_train, as.numeric(pred_train))
results_test  <- compute_stats(y_test,  as.numeric(pred_test))

dataset_train <- data.frame(part = "train", obs = y_train, pred = as.numeric(pred_train))
dataset_test  <- data.frame(part = "test",  obs = y_test,  pred = as.numeric(pred_test))
preds <- rbind(dataset_train, dataset_test)

out_list <- list(
  results_train = results_train,
  results_test  = results_test,
  preds         = preds
)
OUTDIR      <- "C:/Users/calvi/OneDrive/Desktop/Workshop_ICESat2VegR/02_OutputData"
outdir_STATS <- file.path(OUTDIR, "01_STATS")
  dir.create(outdir_STATS, showWarnings = FALSE, recursive = TRUE)
  
  write.xlsx(
    out_list,
    file = file.path(outdir_STATS, "rf_accuracy.xlsx"),
    overwrite = TRUE
  )
cat("Accuracy exported to:", outdir_STATS, "\n")

## Accuracy exported to: C:/Users/calvi/OneDrive/Desktop/Workshop_ICESat2VegR/02_OutputData/01_STATS

results_test

##          r2     rmse rmse_pct
## 1 0.9556868 1.298237 6.234385

13 Step 5 — Wall-to-wall mapping and stats

gee_rf_model <- suppressMessages(ICESat2VegR:::build_ee_forest(rf_model))

yr <- 2025
mosaic <- ee_build_AlphaEarth_embedding_terrain_stack(geom, yr, yr)

ch_map <- mosaic$classify(gee_rf_model)$rename("CH")$unmask(0)

mm <- ch_map$reduceRegion(
  reducer = ee$Reducer$minMax(),
  geometry = geom,
  scale = 25,
  maxPixels = 1e9
)$getInfo()

print(mm)

## $CH_max
## [1] 34.65489
## 
## $CH_min
## [1] 1.094095

14 Step 6 — Wall-to-wall mapping visualization

  ch_palette <- c(
    "#440154FF","#482878FF","#3E4989FF","#31688EFF","#26828EFF",
    "#1F9E89FF","#35B779FF","#6DCD59FF","#B4DE2CFF","#FDE725FF"
  )

  m <- ICESat2VegR::map_view(
    layers = list(
      list(
        type   = "ee_image",
        x      = ch_map,
        bands  = "CH",
        aoi    = geom,
        group  = "Canopy Height",
        min_value = 0,
        max_value = 40,
        palette = ch_palette,
        legend = list(title = "Canopy Height (m)", auto = "quantile")
      )
    ),
    base_tiles = "Carto.Light",
    add_layers_control = TRUE,
    add_opacity_controls = TRUE,
    fit_to = geom
  )
  m

OUTDIR      <- "C:/Users/calvi/OneDrive/Desktop/Workshop_ICESat2VegR/02_OutputData"
MAP_HTML      <- file.path(OUTDIR, "canopy_height_map.html")
htmlwidgets::saveWidget(m, MAP_HTML, selfcontained = TRUE)
cat("Interactive HTML saved:", MAP_HTML, "\n")

## Interactive HTML saved: C:/Users/calvi/OneDrive/Desktop/Workshop_ICESat2VegR/02_OutputData/canopy_height_map.html

15 Step 7 — Export map to GeoTIFF (via Google Drive)

#  map_download(
#    ch_map,
#    method = c("drive"))
#  
cat("GeoTIFF export started to Google Drive: ...", "\n")

## GeoTIFF export started to Google Drive: ...

16 End

17 Key takeaways

ICESat2VegR is a novel and powerful tool for handling ICESat-2 data.
ICESat-2 provides accurate but spatially sparse measurements of canopy height. 3 Artificial intelligence enables high-resolution mapping. 4 Cloud-based predictors are essential for wall-to-wall mapping.

ForestSAT2026 Workshop Exercise – ICESat-2 Canopy Height Mapping

Cesar Alvites1 & Caio Hamamura2 - 1) Forest Biometrics, Remote Sensing and Artificial Intelligence Lab (SilvaLab), University of Florida, 2) Federal Institute of Education, Science and Technology of São Paulo, Avenida Doutor Ênio Pires de Camargo, Capivari 13365-010, SP, Brazil

2026-02-23