使用Apache Spark™和H3地理空间索引系统的卫星图像数据处理.pdf

1、Satellite Imagery Data Processing Using H3 Geospatial IndexNaresh YegireddiStaff Data Engineer,IndigoAGDatabricks2023AgricultureQuick overview on crop lifecycle1AgriculturePrinciples to Improve Soil Healthsource:2AgricultureFew Agronomy Field Events that effect Organic Soil CarbonTillagePlantingCove

2、r CropsChemical Spray3PreambleTracking field events can be challengingHigher field volumes(900 million acres of crop land alone in USA)Field events are not time boundField boundaries can change over timeTracking Field events manually requires more resources and not a cost-effective approach.Agricult

3、ure datasets are disparate in nature.Finding a standard unit of analysis by mapping all the disparate data sets is the key challenge and it can be addressed using Satellite Imagery and H3.4Satellite Imagery ProcessingAn automated approach to track field eventsSource Data:Leverage on satellite imager

4、y available for publicExample:Sentinel-2 data hosted on a public cloud storage.Data refreshes once in every 10 daysImage data are stored as tiffs,cloud optimized geotiffs(COGS)Big Data in size5Remote SensingFinding field events by processing Satellite Imagery6H3Hexagonal Hierarchical Geospatial Inde

5、xSource:https:/h3geo.org/docs/highlights/indexing7H3Hexagonal Hierarchical Geospatial IndexSource:https:/ level data flow9DemoLow resolution tile aggregation Aggregate data across space for use in visualization or other analysis.Partition space for processing.Example-the unit of processing for a satellite imagery batch job10DemoDetect a change in field boundary11Q&A