Granite Geospatial Land Surface Temperature

--- license: apache-2.0 library_name: terratorch pipeline_tag: image-feature-extraction --- # Model Card for granite-geospatial-land-surface-temperature <p align="center" width="100%"> <img src="Johannesburg_summer_lst_animation.gif" width="800"> </p> [<b><i>>>Try it on Colab<<</i></b>](https://colab.research.google.com/github/ibm-granite/granite-geospatial-land-surface-temperature/blob/main/notebooks/1_getting_started.ipynb) The granite-geospatial-land-surface-temperature model is a fine-tuned geospatial foundation model for predicting the land surface temperature (LST) using satellite imagery along with climate statistics. Excessive urban heat has been shown to have adverse effects across a range of dimensions, including increased energy demand, severe heat stress on human and non-human populations, and worse air and water quality. As global cities become more populous with increasing rates of urbanization, it is crucial to model and understand urban temperature dynamics and its impacts. Characterizing and mitigating Urban Heat Island (UHI) effects is dependent on the availability of high-resolution (spatial and temporal) LST data. This model is fine-tuned using a combination of Harmonised Landsat Sentinel-2 [(HLS L30)](https://hls.gsfc.nasa.gov/products-description/l30/) and ECMWF Reanalysis v5 [(ERA5-Land)](https://cds.climate.copernicus.eu/datasets/reanalysis-era5-land?tab=overview) 2m near-surface air temperature (T2m) datasets across 28 global cities from varying hydroclimatic zones for the period 2013-2023. <p align="center" width="100%"> <img src="cities_map2.png" width="800"> </p> ## How to Get Started with the Model This model was trained using [Terratorch](https://github.com/IBM/terratorch). We make the weights as well as the configuration file that defines it available. You can use it easily with Terratorch through: ```python from terratorch.cli_tools import LightningInferenceModel ckpt_path = hf_hub_download(repo_id="ibm-granite/granite-geospatial-land-surface-temperature", filename="LST_model.ckpt") config_path = hf_hub_download(repo_id="ibm-granite/granite-geospatial-land-surface-temperature", filename="config.yaml") model = LightningInferenceModel.from_config(config_path, ckpt_path) inference_results, input_file_names = model.inference_on_dir(<input_directory>) ``` For more details, check out the tutorials below which guide t...