digitalearthafrica/ndvi-anomalies

Production of Landsat resolution NDVI anomalies for the African continent # Digital Earth Africa's Continental Standardised NDVI Anomalies ## Background Standardised NDVI Anomalies provide a measure of vegetation health relative to long term average conditions by measuring the departure, in units of standard devaiations, away from the long-term average. These indices can be used to monitor areas where vegetation may be stressed, and as a proxy to detect potential drought. Negative values represent a reduction from normal NDVI, while positive values represent an increase from normal. ## Description The Standardized NDVI Anomaly will have the following specifications: * NDVI climatologies are developed using harmonized Landsat 5,7,and 8 satellite imagery from the years 1984 to 2020 * Anomalies will have monthly temporal frequency and include images from Landsat 8, Landsat 9, and Sentinel-2 * All datasets will have a native spatial resolution of 30 metres ## Updating the pip requirements Fix any requirement versions in `docker/fixed-requirements.txt` Install `pip-tools` and then run `pip-compile --output-file=docker/requirements.txt production/ndvi_tools/setup.py docker/fixed-requirements.txt`. ## Testing notes Using the dev Sandbox, so that we have a full index of datasets, clone the NDVI repo and then install using pip `pip install --editable ndvi-anomalies/production/ndvi_tools`. Dump out a DB for a single month: `odc-stats save-tasks --temporal-range=2021-08--P1M --grid=africa_30 --usgs-collection-category=T1 ls8_sr-ls9_sr-s2_l2a --tiles=160:171,80:91` or a subset: `odc-stats save-tasks --temporal-range=2021-08--P1M --grid=africa_30 --usgs-collection-category=T1 ls8_sr-ls9_sr-s2_l2a --tiles 170:181,80:91` Run one tile: `odc-stats run --config=production/ndvi_tools/config/ndvi_anomaly.yaml --location=file:///home/jovyan/ndvi ls8_sr-ls9_sr-s2_l2a_2021-08--P1M.db 2021-08--P1M/178/088` ## Additional information **License:** The code in this notebook is licensed under the Apache License, Version 2.0. Digital Earth Africa dat …