Overview Carbon Water Waste Ecosystems Appendix 63 Enabling systems change Investing in AI Measuring biodiversity Catalyzing Our AI for Earth grantees show how AI can drive 92B In April 2020, we announced a new AI for Earth exponential scale. A few that have pushed boundaries collaboration with the Group on Earth Observations this year include the following: Microsoft Azure and Biodiversity Observation Network (GEO BON). This innovation and Closing the gap to real-time biodiversity surveys high-resolution satellite $1 million Essential Biodiversity Variables (EBVs) on the NOAA΄s Alaska Fisheries Science Center and UW- imagery was used to cloud grant program was launched to help advance CICOES are working with Microsoft on training AI tools count 92 billion trees efforts to monitor Earth’s biodiversity and create useful collective action to distinguish a beluga whale’s call from a dredging in the US. measurements required for the study, reporting, and machine’s squeak, to better understand how human management of biodiversity change that inform noise might be related to the lack of recovery for this conservation decisions across the globe. In October 2020, We are focused on new ways to endangered and declining population. This type of data we announced five winners that will receive financial allows scientists to devise strategies to reduce negative support, as well as access to resources from Microsoft and harness the power of technology, effects from increased human activity in the whales΄ Esri. These projects include: mapping Belize’s coastal and partnerships, investments, and policy habitat. Machine learning is achieving more than 96 marine ecosystems, detecting change in essential to drive impact at scale and pace to percent accuracy in classifying detections compared to biodiversity for citizen science initiatives, improving a more traditional approach, and even picking up things mapping accuracy of mountainous ecosystems, creating help the world protect ecosystems human analysts miss. The model can also process a large open source tools to analyze bioacoustics data, and and biodiversity. amount of data very quickly—moving datasets from two combining data sources to assess EBVs. weeks of processing and manual classification to Advancing forestry and sequestration measurement completing a batch of six months of sound recordings Accurate forest inventories are an essential tool for in just two days. Distinguishing whale calls conservationists, governments, and landowners who seek with AI. to promote ecological, social, and economic health. Microsoft has worked with SilviaTerra to accelerate their cutting-edge approach to precision forestry, and radically increase the scope, efficiency, and resolution of forest inventories. They produced the first ever comprehensive forest map of the US, covering every acre of the contiguous US and accounting for almost 92 billion trees. SilviaTerra applies this annually-updated data to predict carbon sequestration on each forest parcel, bringing to market a new gigaton-scale supply of forest carbon credits. They are also working to explicitly measure co-benefits such as wildlife habitat, water yield, and fire risk reduction. Microsoft is also supporting SilviaTerra΄s work by purchasing carbon credits generated by their pilot project.