![projects/global-pasture-watch/assets/gsvh-30m/v1/short-veg-height_m](https://developers.google.com/earth-engine/datasets/images/global-pasture-watch/projects_global-pasture-watch_assets_gsvh-30m_v1_short-veg-height_m_sample.png) info


This dataset is part of a Publisher Catalog, and not managed by Google Earth Engine.

Contact [Land \& Carbon Lab](https://landcarbonlab.org/subscribe)

for bugs or [view more datasets](https://developers.google.com/earth-engine/datasets/publisher/global-pasture-watch.md.txt)
from the Global Pasture Watch Catalog. [Learn more about Publisher datasets](https://developers.google.com/earth-engine/datasets/publisher).
[![Catalog Owner Logo](https://developers.google.com/static/earth-engine/datasets/logos/global-pasture-watch_logo.png)](https://landcarbonlab.org/data/global-grassland-and-livestock-monitoring)

Catalog Owner
:   Global Pasture Watch

Dataset Availability
:   2000-01-01T00:00:00Z--2025-01-01T00:00:00Z

Dataset Producer
:


    [Land and Carbon Lab Global Pasture Watch](https://landcarbonlab.org/data/global-grassland-and-livestock-monitoring)

Contact
:   [Land \& Carbon Lab](https://landcarbonlab.org/subscribe)

Cadence
:   1 Year

Tags
:
    [canopy](https://developers.google.com/earth-engine/datasets/tags/canopy.md.txt) [global](https://developers.google.com/earth-engine/datasets/tags/global.md.txt) [global-pasture-watch](https://developers.google.com/earth-engine/datasets/tags/global-pasture-watch.md.txt) [land](https://developers.google.com/earth-engine/datasets/tags/land.md.txt) [landcover](https://developers.google.com/earth-engine/datasets/tags/landcover.md.txt) [plant-productivity](https://developers.google.com/earth-engine/datasets/tags/plant-productivity.md.txt) [publisher-dataset](https://developers.google.com/earth-engine/datasets/tags/publisher-dataset.md.txt) [vegetation](https://developers.google.com/earth-engine/datasets/tags/vegetation.md.txt)

#### Description

This dataset provides global median vegetation height
from 2000 at 30-m spatial resolution.
Produced by the Land \& Carbon Lab's Global Pasture Watch initiative, this dataset provides **median vegetation
height** values (50th percentile) globally at 30-m spatial resolution from 2000 onwards. The dataset is based on
**ICESat-2 ATL08** median height of vegetation returns and is modeled via machine learning (ensemble Gradient Boosted Trees)
using the [**GLAD Landsat ARD** (collection-2)](https://glad.umd.edu/ard/home)
aggregated every two months (see [Consoli et al., 2024](https://peerj.com/articles/18585/)) and combined with
additional covariates, including **terrain elevation** ([GEDTM30](https://doi.org/10.7717/peerj.19673)),
**geometric mean temperature** , and **MODIS long-term temperature** and **water vapor**.

Originally designed to support the monitoring of **open ecosystems** (grasslands, open shrublands, savannas, tundra), the
dataset offers wall-to-wall coverage of all terrestrial ecosystems. Thus, median height values of forest ecosystems
should not be interpreted as top of canopy. For comparison with other existing canopy height products access the app
[GPW Height Comparison Toolkit (GPW-HCT)](https://gpw-lapig.projects.earthengine.app/view/hct).

Estimates of **90% prediction interval values** (5th \& 95th percentiles) are available in [OpenLandMap STAC](https://stac.openlandmap.org/openlandmap/collections/gpw_gsvh-30m_go).

**Per-pixel trend analysis** is computed on-the-fly using [GEE App](https://global-pasture-watch.projects.earthengine.app/view/gsvh-30m).

**Limitations:**

- **Vegetation heterogeneity**: The median height is sensitive to shrubs and trees within landscapes dominated by herbaceous cover.
  Users focused on herbaceous biomass should consider using fractional cover maps to address sub-pixel mixtures. Future versions
  may incorporate ICESat-2's vertical structure information for a better separation of woody cover.

- **Sub-annual and seasonal patterns**: Current ICESat-2 data density is insufficient to map seasonal height changes common
  in grassy ecosystems. Increased Lidar acquisitions or denser sampling from ongoing measurement are needed to capture
  more frequent snapshots, which would improve monitoring of intra-annual changes and disturbances, including grazing, fire, and harvesting.

- **Data restrictions and uncertainty**: The training data was restricted to strong-beam, night-time measurements to minimize
  signal noise, and additional filters were applied, but this does not fully account for all sources of data uncertainty
  in the ICESat-2 mission. Furthermore, cloud cover and smoke hinder signal penetration, increasing uncertainty in some regions.
  Specifically for 2019, the model presented slightly poorer performance compared to more recent years, which might be
  related to differences in data availability and quality at the beginning of the ICESat-2 mission.

- **Difficulty detecting very short vegetation**: Lidar sensors, including ICESat-2, struggle to accurately measure very
  low canopies, especially with sparse cover or partial ground occlusion. The training data does not identify vegetation
  less than 50 cm from the ground surface, which may lead to overestimation of height in the shortest or most sparse land
  covers, such as wetland saltpans.

- **Underestimation of maximum heights** : The trained machine learning model exhibited a tendency toward the mean, which
  resulted in overly narrow and optimistic [prediction intervals](https://stac.openlandmap.org/openlandmap/collections/gpw_gsvh-30m_go).
  Due to the underestimation of vegetation height in the upper extremes, care should be taken when interpreting absolute
  values in regions with very tall forests, but relative patterns and trends are still informative.

- **Limited independent validation**: While validated against ICESat-2 data (testing set), broader validation using airborne
  or drone Lidar would be necessary to fully assess the quality of the dataset at regional level.

**For more information see [Hunter et al., 2025](https://doi.org/10.1038/s41597-025-05739-6),
[Zenodo](https://doi.org/10.5281/zenodo.15198654) and
[Global Pasture Watch GitHub site](https://github.com/wri/global-pasture-watch)**

### Bands

**Bands**

Pixel size: 27.83 meters (all bands)

| Name | Units | Min | Max | Scale | Pixel Size | Description |
|---|---|---|---|---|---|---|
| `height` | m | 0 | 10 | 0.1 | 27.83 meters | Median vegetation height |

### Image Properties

**Image Properties**

| Name | Type | Description |
|---|---|---|
| version | INT | Product version |

### Terms of Use

**Terms of Use**

[CC-BY-4.0](https://spdx.org/licenses/CC-BY-4.0.html)

### Citations

Citations:

- Parente, L., Hunter, M., Ho, Y., Bonannella, C. el al. (2025).
  Global Pasture Watch - Annual short vegetation height maps at
  30-m spatial resolution (2000---2022) (Version v1)
  \[Data set\]. Zenodo. [doi:https://doi.org/10.5281/zenodo.15198654](https://doi.org/10.5281/zenodo.15198654)
- Hunter, M.O., Parente, L., Ho, Yf. et al. (2025)
  Global 30-m annual median vegetation height maps (2000---2022)
  based on ICESat-2 data and Machine Learning. Scientific Data 12, 1470.
  [doi: https://doi.org/10.1038/s41597-025-05739-6](https://doi.org/10.1038/s41597-025-05739-6)

### DOIs

- <https://doi.org/10.1038/s41597-025-05739-6>
- <https://doi.org/10.5281/zenodo.15198654>

### Explore with Earth Engine

> [!NOTE]
> **Important:** Earth Engine is a platform for petabyte-scale scientific analysis and visualization of geospatial datasets, both for public benefit and for business and government users. Earth Engine is free to use for research, education, and nonprofit use. To get started, please [register for Earth Engine access.](https://console.cloud.google.com/earth-engine)

### Code Editor (JavaScript)

```javascript
Map.setCenter(-49.265188, -16.602052, 4);

var short_veg_height_m = ee.ImageCollection("projects/global-pasture-watch/assets/gsvh-30m/v1/short-veg-height_m"),
    style = {"opacity":1,"bands":["height"],"min":0,"max":10,"palette":["#fcffa4","#f98d0a","#bb3755","#57106e","#000004"]};

var SCALE_FACTOR = 0.1
var svhVis = {min: 0, max: 10, palette: "fcffa4,f98d0a,bb3755,57106e,000004"}
var svh = ee.ImageCollection(
    "projects/global-pasture-watch/assets/gsvh-30m/v1/short-veg-height_m"
)

var svh2024 = svh.filterDate('2024-01-01', '2025-01-01').first().multiply(SCALE_FACTOR);
Map.addLayer(svh2024, svhVis, 'Median vegetation height (2024)');

var svh2000 = svh.filterDate('2000-01-01', '2001-01-01').first().multiply(SCALE_FACTOR);
Map.addLayer(svh2000, svhVis, 'Median vegetation height (2000)');
```
[Open in Code Editor](https://code.earthengine.google.com/?scriptPath=Examples:Datasets/global-pasture-watch/projects_global-pasture-watch_assets_gsvh-30m_v1_short-veg-height_m) [GPW Annual short vegetation height v1](https://developers.google.com/earth-engine/datasets/catalog/projects_global-pasture-watch_assets_gsvh-30m_v1_short-veg-height_m.md.txt) This dataset provides global median vegetation height from 2000 at 30-m spatial resolution. Produced by the Land \& Carbon Lab's Global Pasture Watch initiative, this dataset provides median vegetation height values (50th percentile) globally at 30-m spatial resolution from 2000 onwards. The dataset is based on ICESat-2 ATL08 median height ... projects/global-pasture-watch/assets/gsvh-30m/v1/short-veg-height_m, canopy,global,global-pasture-watch,land,landcover,plant-productivity,publisher-dataset,vegetation 2000-01-01T00:00:00Z/2025-01-01T00:00:00Z -90 -180 90 180 Google Earth Engine https://developers.google.com/earth-engine/datasets

- [https://doi.org/10.5281/zenodo.15198654](https://doi.org/https://landcarbonlab.org/data/global-grassland-and-livestock-monitoring)
- [https://doi.org/10.5281/zenodo.15198654](https://doi.org/https://developers.google.com/earth-engine/datasets/catalog/projects_global-pasture-watch_assets_gsvh-30m_v1_short-veg-height_m)