DMC 22m Sensors for Supertemporal Land Cover Monitoring Gary Holmes DMC International Imaging Ltd June 2014
DMC 2 nd Generation Satellites UK-DMC2 and Deimos-1 launched 29 th July 2009 650km swath width 22m GSD Vis/NIR 3 bands Calibrated within 1% of Landsat Up to 1,500km along-track at full swath New operational modes, NRT Direct Downlink Service
UK-DMC DMC-2, 2009 California Forest Fires
UK-DMC DMC-2, 2010 Amsterdam, Netherlands
UK-DMC DMC-2, 2010 Desert Agriculture, Libya
DMC timeliness DMC 600 x 560 kmkm Image LANDSAT 185 x 185 Images 2-yearfrom mosaic single image UK-DMC -24ofscenes from different - Most England captured within months/seasons 1 minute Impossible to apply consistent - -Ideal for consistent classification classification method approach Near real-time real-time precision precision -- Near agriculturefully not supported possible agriculture - Multitemporal coverage often - Reliable multi-temporal unavailable for per a given growing coverage season season - Many fewer images to process than Landsat
Reliable multitemporal coverage (UK monthly sequence example capturing crop phenology changes) Feb Mar Apr May Jul Jul Aug Sep Oct
Reliable multitemporal coverage Feb Mar Apr May Jul Jul Aug Sep Oct
Reliable multitemporal coverage Feb Mar Apr May Jul Jul Aug Sep Oct
Reliable multitemporal coverage Feb Mar Apr May Jul Jul Aug Sep Oct
Reliable multitemporal coverage Feb Mar Apr May Jul Jul Aug Sep Oct
Reliable multitemporal coverage Feb Mar Apr May Jul Jul Aug Sep Oct
Reliable multitemporal coverage Feb Mar Apr May Jul Jul Aug Sep Oct
Reliable multitemporal coverage Feb Mar Apr May Jul Jul Aug Sep Oct
Reliable multitemporal coverage Feb Mar Apr May Jul Jul Aug Sep Oct
Multitemporal Classification Product
Case Studies: Land Cover, Agriculture and Forest Monitoring
USA Multitemporal Monitoring For US Department of Agriculture since 2011 Satellites used: Deimos-1 and UK-DMC2 Complete coverage of lower 48 States every 15-days 12 coverages April October (average ~90% cloud-free) 150 Million km 2 of cloud-free imagery delivered per year Image courtesy of Astrium / Deimos Imaging
USA Multitemporal Monitoring 2011-12-13-14 Image courtesy of Astrium / Deimos Imaging 19
USDA Cropland Data Layer ~ 9 billion pixels! Image Courtesy of USDA
National Cropland Data Layer Legend
Improved Crop Classification Results through multitemporal DMC acquisitions: North Carolina Example Landsat& DMC Landsat Only DMC Only ALL CROPS Producer 81.7% User Corn 94.3% 92.1% Cotton 96.3% 94.4% Soybeans 84.7% 84.0% WW/Soy 93.9% 87.9% Pasture 66.5% 68.1% Peanuts 87.1% 92.8% ALL CROPS Producer 69.0% User Corn 83.3% 81.9% Cotton 91.0% 85.3% Soybeans 62.9% 71.1% WW/Soy 79.9% 82.0% Pasture 58.5% 44.0% Peanuts 71.4% 90.0% ALL CROPS Producer 81.6% User Corn 94.6% 92.1% Cotton 95.9% 94.4% Soybeans 84.3% 83.3% WW/Soy 93.8% 87.7% Pasture 67.2% 68.9% Peanuts 82.1% 91.0% 118 bands 73 bands 66 bands Statistics courtesy of USDA
Crop Classification in Emilia-Romagna, Italy Classification work performed by Regional Government For water balance, irrigation monitoring and statistics Small field parcels Multitemporal DMC-based classification (November-June) A fully operational system Courtesy of ARPA Emilia-Romagna
Netherlands Systematic Coverage Procured by NSO (Dutch Space Office) DMC 22m coverage of Netherlands 3x per week March-October; 2x per week November-February Entire country imaged 140x per yr Continuity secured for 2012-2016 Data licensed for use by multiple organisations within the Netherlands
Fieldlook Weekly updates using multiple data sources >10 quantitative parameters Growth biomass production (kg/ha) CO2 intake (kg/ha) leaf area index LAI (m2 leaf/m2 ground) vegetation index NDVI Moisture evaporation shortage (mm/week) current evaporation (mm/week) surplus rain (mm/2 weeks) reference evaporation Minerals Nitrogen content in the top leaf layer (kg/ha) Nitrogen content in all leafs (kg/ha) Yield Potatoes, Wheat, Maize, Sugar beet Dry matter content Sugar Yield Protein content
Green Monitor Alterra Wageningen
Applications: Precision Agriculture
Precision Agriculture: The Imagery Acquisition Challenge for Satellites Highly demanding application for remote sensing: Very large areas to be covered for operational services Multitemporal coverage needed Timeliness is vital (cloud-free data at specific growth stages) High enough resolution to map within-field variability Rapid processing & delivery (for NRT services) Must be low cost per acre DMC constellation matches these demands well wide swath sensors on multiple small satellites DMC imagery now used for Precision Ag in: USA, Canada, Argentina, Brazil, UK, France, Netherlands, Lithuania, Ukraine, Russia, Bulgaria, South Africa, Japan
UK coverage: a successful and growing service SOYLSensehas used DMC data since 2005for Precision Farming Service in UK 3.6 million acres of maps delivered to UK farmers in 2012 DMCii delivers imagery to Geosys, France Geosys: data hosting, basic processing, web portal SOYL: calibration, recommendation maps, delivery to farmers, variable rate applications DMCiiperforms high frequency of acquisitions on an area of more than 125,000 km² January to May : over 30 full coveragesof most of UK, from England to Scotland Typically 7-12 cloud-free images per field 29
From image to application map SOYLSensemaps delivered by SOYL typically less than 3 days after DMCii image acquisition Leaf Area Index (LAI) is estimated from the imagery and field samples Fertilization models bring LAI data together with UK fertilizer guidelines Customers receive a reportof field variability and Nitrogen Variable Rate Application map for each of their fields 30
Profitability A GEOSYS Yield = 12 T/ha N = 206 kg/ha B Standard Yield = 11.4 T/ha N = 225 kg/ha 87 kg/ha 50 kg/ha 69 kg/ha Yield = +147 /ha -19 kg N/ha +15% efficiency N/kg fert UK 31
Applications: Forest Monitoring
Brazilian Amazon Annual DMC campaigns for INPE: 2005 2006 2007 2008 2009 2010 2011 Image Courtesy of INPE, Brazil
Forest change detection- Brazil
UK-DMC2 Direct Downlink Service 2012+ Near real-time direct downlink to Cuiaba High data volume every month Upgrade from 250m MODIS pixel to 22m pixel for DETER program Imagery will be openly accessible through INPE website along with CBERS and Landsat
Contentious Issues / Questions Copernicus DWH very patchy IRS/Rapideye coverages at high cost in 2009-13. A new ITT is coming up in 2014 and S2A won t be operational until late 2015 and possibly later. S2B not until 2017/18. Danger of another patchy DWH European dataset? Sentinel 2 what do we expect it to deliver for UK? Probably not a panacea. S2A 10-day revisit slightly better than Landsat reliable coverage every ~ 6 months? S2A+S2B 5- day revisit coverage ~ 3 months? We are always looking to future sensors but we can easily cover the whole of Europe every month NOW with 22m Vis/NIR data with 1-2 day revisit. Would this be useful for many applications as a safety net? Looks like the DWH ITT will again exclude DMC as a non-compliant data source. Is this justified or an accident of the reqs/procurement process? UK national data supply program (like NL). DMCii has already proposed open access to UK wall to wall monthly 22m Vis/NIR coverage via NERC and now Catapult. Is there an appetite for this?
g.holmes@dmcii.com www.dmcii.com