Results

Earth Engine & Cloud Computation in Remote Sensing

At it’s core, EE allows user to tap into Google’s massive computation capability to apply remote sensing research to general public. Their first massive project, Global Forest Fire, was done in conjunction with the University of Maryland, to demonstrate this capability. “Google Earth Engine is a massively parallel technology for high-performance processing of geospatial data, and houses a copy of the entire Landsat image catalog […] What would have taken a single computer 15 years to perform was completed in a matter of days using Google Earth Engine computing [2]. This project have already been cited in recent papers for its potential in helping developing areas obtain information which may otherwise be unavailable due to resource. In his survey of cloud computation in remote sensing, Kshetri utilizes EE’s deforestation data as an example in which cloud computation “can help address environmental issues”, thereby offsetting the environmental impact caused by large server farms [3]. In fact, EE was first unveiled during a United Nations climate talk in Mexico, whereby it was positioned as a resource ““for measuring, reporting and verifying anthropogenic forest-related emissions” [4] .

EE provides several other processed products available for browsing to the public on its website. They provide specific examples for area study through application of temporal Landsat data such as: “Saudi Arabia Irrigation” “Drying of Lake Urmia, Iran”, “Columbia glacier Retreat,” “Dubai Coastal Expansion”, and “Drying of the Aral Sea,” as well as products done through computation such as “Global roadless Areas,” “water Mask of central Africa,” “NDFI over the Amazon,” and “Landsat 7 L1T Coverage” [2].

Another major contribution to the field is Exelis’ Service Engine. Service Engine is a buyable product that follows “the concept of master and worker nodes,” whereby consumers would load their product on a centralized servers which contains the computation power, as well as the data, which the consumer can then access through thin clients remotely, in other words allows ENVI analytics in the cloud [5]. Examples provided include “environmental responder[s] getting real-time updates on rescue efforts while in the field, a deployed soldier getting updates on enemy troop movements, severe weather warnings going out to disaster response teams, or even an assessment of civil unrest within a region” [5]. On top of existing data on servers, Service Engine enables information uploads through clients on the ground to provide more real-time results. They also tout interoperability, being adherent to Open Geospatial Consortium and the Esri Geo Services REST specification, and accessible via the IDL programming language [5]. Such interoperability is demonstrated with their example of working in conjunction with Milcord on dPlan which optimizes UAV routing and analysis [5]. In fact, Service Engine is largely only a component in Exelis’ cloud strategy, since it depends on interfaces for users to use. One such example is Exelis’ Jagwire, which is a “web-based software system that is specifically designed for ingest, storage, management, discovery, and delivery of geospatial full motion video (FMV), imagery, and derived products with near real-time access,” though they promote “custom web pages designed to enable those capabilities for the end-user. “

Applying to Google Earth Engine

Without applying to EE, users can test drive basic features of EE. For example, they may add data from the Data Catalog and experiment with how the data is viewed (e.g. bands allocation, gain, palette, range, and year). However, more advanced remote sensing analysis is not possible without registration.

We applied to Earth Engine by submitting a request to the Earth Engine Beta Signup, which can be obtain through this form [6]. Earth Engine Beta applicants must have a Gmail account, but other than that there does not seem to be any restrictions. An example of our request looks like, “Some colleagues and I are working on remote sensing application and analysis, at UCLA, were hoping to be permitted access to the Earth Engine Beta. Our work has previously utilized programs such as ENVI and Earth Explorer. In addition we are interested to see how Earth Engine might fit into the academic sphere.” Within roughly 24-hours we all go emails outlining how to access the EE workspace. Provided full beta access you are able to do a variety of things,  which we will discuss later in in the paper.

Data Catalog Availability

Table 1. Earth Explorer sans ASTER, MODIS, LANDSAT

Product Name Coverage Availability on EE Notes
Aerial Regional L NAIP: National Agriculture Imagery Program
Cal/Val- Referece Sites Regional N
Declassified (Old-Military) N
Digital Elevation L
Digital Line Graphs Regional N
Digital Maps Regional N
USGS Group on Earth Observations (GEO) Global Agricultural Monitoring (GLAM) Regional N
Global Fiducials Library (GFL) N
Global Forest Observation Initiative N
Global Land Survey Y
Heat Capacity Mapping Mission Regional N
Lidar Regional N
AVHRR 1km global L EE has surface temperature from AVHRR
AVHRR Composites (NDVI) L EE has NDVI from other satellites
IKONOS-2 Regional N
Declassified Military Regional N USGS has declassified military aerial photos from Corona, Argon, and Lanyard, originally for recon.
EO-1 N
JECAMM Regional L Joint Experiment of Crop Assessment and Monitoring. data may intersect with NAIP, although it seems its for Canada only
Orbview L Orbview itself is not listed, but maybe it is a part of EE’s base map
National Land Cover Data Regional Y
NASA LPDACC Y
GEO-Eye N

Y means product category is available, N means no, L means limited availability

Table 2. Landsat Data

Satellites Availability on EE
Landsat 1 Available as part of Land Survey 1975
Landsat 2 Available as part of Land Survey 1975
Landsat 3 Available as part of Land Survey 1975
Landsat 4 Yes
Landsat 5 Yes
Landsat 7 Yes
Landsat 8 Yes

 

Table 3. NOAA Data

Name of Data Availability
VIIRS (Night fire / infrared spectral) No
DMSP (Night time lights) Yes
Nightsat (Moderate resolution human settlements sprawl) No
GOES (Continuous atmosphere monitoring) No

 

Table 4. MEaSUREs Data Products Availability in Google Earth Explorer

Short Name Collection MEaSUREs Data Product Spatial Resolution Available in EE
SRTMGL1 SRTM SRTM Global 1 arc second 1 arc-second No
SRTMGL1N SRTM SRTM Global 1 arc second number 1 arc-second No
SRTMGL3 SRTM SRTM Global 3 arc second 3 arc-second No
SRTMGL30 SRTM SRTM Global 30 arc second 30 arc-second No
SRTMGL3N SRTM SRTM Global 3 arc second number 3 arc-second No
SRTMGL3S SRTM SRTM Global 3 arc second sub-sampled 3 arc-second No
SRTMSWBD SRTM SRTM Water Body Data Shapefiles & Raster Files 1 arc-second No
SRTMUS1 SRTM SRTM US 1 arc second 1 arc-second No
SRTMUS1N SRTM SRTM US 1 arc second number 1 arc-second No
WELDAKLL WELD WELD Alaska Lat/Longs 30 m No
WELDAKMO WELD WELD Alaska Monthly 30 m No
WELDAKSE WELD WELD Alaska Seasonal 30 m No
WELDAKWK WELD WELD Alaska Weekly 30 m No
WELDAKYR WELD WELD Alaska Annual 30 m No
WELDUSLL WELD WELD CONUS Lat/Long 30 m No
WELDUSMO WELD WELD CONUS Monthly 30 m No
WELDUSSE WELD WELD CONUS Seasonal 30 m No
WELDUSWK WELD WELD CONUS Weekly 30 m No
WELDUSYR WELD WELD CONUS Annual 30 m No

 

Table 5. ASTER Data Products Availability in Google Earth Explorer

Shortname Level Aster Data Product Resolution Available in EE Alternative in EE
AST_L1BE 1B Registered Radiance at the Sensor – Expedited 15, 30, 90 No
AST_L1AE 1A Reconstructed Unprocessed Instrument Data – Expedited 15, 30, 90 No
AST_07 2 Surface Reflectance – VNIR & SWIR 15, 30 No Landsat 5 Surface Reflectance
AST_07XT 2 Surface Reflectance – VNIR & Crosstalk Corrected SWIR 15, 30 No Landsat 5 Surface Reflectance
AST_09 2 Surface Radiance – VNIR & SWIR 15, 30 No
AST_09XT 2 Surface Radiance – VNIR & Crosstalk Corrected SWIR 15, 30 No
AST_09T 2 Surface Radiance TIR 90 No
AST_08 2 Surface Kinetic Temperature 90 No MOD11A1 Land Surface Temperature and Emissivity Daily Global 1 km Grid SIN
AST_05 2 Surface Emissivity 90 No MOD11A1 Land Surface Temperature and Emissivity Daily Global 1 km Grid SIN
AST14OTH 3 Registered Radiance at the Sensor – Orthorectified 15, 30, 90 No
AST_L1B 1B Registered Radiance at the Sensor 15, 30, 90 No
AST14DMO 3 Digital Elevation Model & Registered Radiance at the Sensor – Orthorectified 15, 30, 90 No SRTM Digital Elevation Data Version 4
AST_L1A 1A Reconstructed Unprocessed Instrument Data 15, 30, 90 No
AST14DEM 3 Digital Elevation Model 30 No SRTM Digital Elevation Data Version 4
ASTGTM 3 ASTER Global Digital Elevation Model 30 No SRTM Digital Elevation Data Version 4

 

Table 6. MODIS Data Products Availability in Google Earth Explorer

Short Name Platform MODIS Data Product Raster Type Res (m) Temporal Granularity Available in EE Alternative in EE
MCD12C1 Combined Land Cover Type CMG 5600m Yearly No MCD12Q1 Land Cover Type Yearly Global 500m
MCD12Q1 Combined Land Cover Type Tile 500m Yearly Yes
MCD12Q2 Combined Land Cover Dynamics Tile 500m Yearly No
MCD15A2 Combined Leaf Area Index – FPAR Tile 1000m 8 day No MCD12Q1-3 LAI/fPAR
MCD15A3 Combined Leaf Area Index – FPAR Tile 1000m 4 day No MCD12Q1-3 LAI/fPAR
MCD43A1 Combined BRDF-Albedo Model Parameters Tile 500m 16 day Yes
MCD43A2 Combined BRDF-Albedo Quality Tile 500m 16 day Yes BRDF-Albedo Model Parameters 16-Day L3 Global 500m
MCD43A3 Combined Albedo Tile 500m 16 day No
MCD43A4 Combined Nadir BRDF-Adjusted Reflectance Tile 500m 16 day Yes
MCD43B1 Combined BRDF-Albedo Model Parameters Tile 1000m 16 day No BRDF-Albedo Model Parameters 16-Day L3 Global 500m
MCD43B2 Combined BRDF-Albedo Quality Tile 1000m 16 day No MCD43A2 BRDF-Albedo Quality 16-Day Global 500m
MCD43B3 Combined Albedo Tile 1000m 16 day Yes
MCD43B4 Combined Nadir BRDF-Adjusted Reflectance Tile 1000m 16 day No MCD43A4 BRDF-Adjusted Reflectance 16-Day Global 500m
MCD43C1 Combined BRDF-Albedo Model Parameters CMG 5600m 16 day No BRDF-Albedo Model Parameters 16-Day L3 Global 500m
MCD43C2 Combined BRDF-Albedo Snow-free Quality CMG 5600m 16 day No
MCD43C3 Combined Albedo CMG 5600m 16 day No
MCD43C4 Combined Nadir BRDF-Adjusted Reflectance CMG 5600m 16 day No MCD43A4 BRDF-Adjusted Reflectance 16-Day Global 500m
MCD45A1 Combined Thermal Anomalies & Fire Tile 500m Monthly No N/A
MOD09A1 Terra Surface Reflectance Bands 1–7 Tile 500m 8 day Yes
MOD09CMG Terra Surface Reflectance Bands 1–7 CMG 5600m Daily No
MOD09GA Terra Surface Reflectance Bands 1–7 Tile 500/1000m Daily Yes
MOD09GQ Terra Surface Reflectance Bands 1–2 Tile 250m Daily Yes
MOD09Q1 Terra Surface Reflectance Bands 1–2 Tile 250m 8 day No MOD09GQ Surface Reflectance Daily L2G Global 250m
MOD11A1 Terra Land Surface Temperature & Emissivity Tile 1000m Daily Yes
MOD11A2 Terra Land Surface Temperature & Emissivity Tile 1000m 8 day Yes
MOD11B1 Terra Land Surface Temperature & Emissivity Tile 5600m Daily No MOD11A2 Land Surface Temperature and Emissivity 8-Day Global 1km
MOD11C1 Terra Land Surface Temperature & Emissivity CMG 5600m Daily No MOD11A2 Land Surface Temperature and Emissivity 8-Day Global 1km
MOD11C2 Terra Land Surface Temperature & Emissivity CMG 5600m 8 day No MOD11A2 Land Surface Temperature and Emissivity 8-Day Global 1km
MOD11C3 Terra Land Surface Temperature & Emissivity CMG 5600m Monthly No MOD11A2 Land Surface Temperature and Emissivity 8-Day Global 1km
MOD11_L2 Terra Land Surface Temperature & Emissivity Swath 1000m 5 min No MOD11A2 Land Surface Temperature and Emissivity 8-Day Global 1km
MOD13A1 Terra Vegetation Indices Tile 500m 16 day Yes
MOD13A2 Terra Vegetation Indices Tile 1000m 16 day No MOD13A1 Vegetation Indices 16-Day L3 Global 500m
MOD13A3 Terra Vegetation Indices Tile 1000m Monthly No MOD13A1 Vegetation Indices 16-Day L3 Global 500m
MOD13C1 Terra Vegetation Indices CMG 5600m 16 day No MOD13A1 Vegetation Indices 16-Day L3 Global 500m
MOD13C2 Terra Vegetation Indices CMG 5600m Monthly No MOD13A1 Vegetation Indices 16-Day L3 Global 500m
MOD13Q1 Terra Vegetation Indices Tile 250m 16 day Yes
MOD14 Terra Thermal Anomalies & Fire Swath 1000m 5 min No N/A
MOD14A1 Terra Thermal Anomalies & Fire Tile 1000m Daily No N/A
MOD14A2 Terra Thermal Anomalies & Fire Tile 1000m 8 day No N/A
MOD15A2 Terra Leaf Area Index – FPAR Tile 1000m 8 day No N/A
MOD17A2 Terra Gross Primary Productivity Tile 1000m 8 day No N/A
MOD17A3 Terra Net Primary Productivity Tile 1000m Yearly No MCD12Q1-4 NPP
MOD44A Terra Vegetation Continuous Cover Tile 250m 96 day No N/A
MOD44B Terra Vegetation Continuous Fields Tile 250m Yearly Yes
MOD44W Terra Land Water Mask Derived Tile 250m None Yes
MYD09A1 Aqua Surface Reflectance Bands 1–7 Tile 500m 8 day Yes
MYD09CMG Aqua Surface Reflectance Bands 1–7 CMG 5600m Daily No MYD09GA Surface Reflectance Daily L2G Global 1km and 500m
MYD09GA Aqua Surface Reflectance Bands 1–7 Tile 500/1000m Daily Yes
MYD09GQ Aqua Surface Reflectance Bands 1–2 Tile 250m Daily Yes
MYD09Q1 Aqua Surface Reflectance Bands 1–2 Tile 250m 8 day No MYD09GQ Surface Reflectance Daily L2G Global 250m
MYD11A1 Aqua Land Surface Temperature & Emissivity Tile 1000m Daily Yes
MYD11A2 Aqua Land Surface Temperature & Emissivity Tile 1000m 8 day Yes
MYD11B1 Aqua Land Surface Temperature & Emissivity Tile 5600m Daily No MYD11A1 Land Surface Temperature and Emissivity Daily Global 1 km Grid SIN
MYD11C1 Aqua Land Surface Temperature & Emissivity CMG 5600m Daily No MYD11A1 Land Surface Temperature and Emissivity Daily Global 1 km Grid SIN
MYD11C2 Aqua Land Surface Temperature & Emissivity CMG 5600m 8 day No MYD11A1 Land Surface Temperature and Emissivity Daily Global 1 km Grid SIN
MYD11C3 Aqua Land Surface Temperature & Emissivity CMG 5600m Monthly No MYD11A1 Land Surface Temperature and Emissivity Daily Global 1 km Grid SIN
MYD11_L2 Aqua Land Surface Temperature & Emissivity Swath 1000m 5 min No MYD11A1 Land Surface Temperature and Emissivity Daily Global 1 km Grid SIN
MYD13A1 Aqua Vegetation Indices Tile 500m 16 day Yes
MYD13A2 Aqua Vegetation Indices Tile 1000m 16 day No MYD13A1 Vegetation Indices 16-Day L3 Global 500m
MYD13A3 Aqua Vegetation Indices Tile 1000m Monthly No MYD13A1 Vegetation Indices 16-Day L3 Global 500m
MYD13C1 Aqua Vegetation Indices CMG 5600m 16 day No MYD13A1 Vegetation Indices 16-Day L3 Global 500m
MYD13C2 Aqua Vegetation Indices CMG 5600m Monthly No MYD13A1 Vegetation Indices 16-Day L3 Global 500m
MYD13Q1 Aqua Vegetation Indices Tile 250m 16 day Yes
MYD14 Aqua Thermal Anomalies & Fire Swath 1000m 5 min No N/A
MYD14A1 Aqua Thermal Anomalies & Fire Tile 1000m Daily No N/A
MYD14A2 Aqua Thermal Anomalies & Fire Tile 1000m 8 day No N/A
MYD15A2 Aqua Leaf Area Index – FPAR Tile 1000m 8 day No MCD12Q1-3 LAI/fPAR
MYD17A2 Aqua Gross Primary Productivity Tile 1000m 8 day No N/A

 

GUI & Programming

Both EE and alternative software, such as ENVI have their benefits and drawbacks when it comes to UI . However, the breadth of manipulation in EE is highly dependant on the users programming capabilities (either java or python).

Generally speaking microscale analysis is much more difficult in EE than alternative, proprietary platforms. It should be mentioned that this is largely with regard to non-classified data. Imagery and analysis that has been published in EE allows quite intuitive manipulation. Furthermore, not having to download imagery via platforms such as Earth Explorer and then upload it into the desired program can be very efficient.

Macro or global scale analysis of precomputed data sets including NDWI, NDVI, etc. facilitate tremendously powerful computation processed with Google’s serves. If one wanted to do the similar scale analysis on their own machine, not only would it take a tremendous amount of time but would surely strain your computer (if even possible). Additional benefits of EE’s platform and data repository include pre-written java templates for image classification. Again the downside of this is needing Java programming experience and the disorganization of resource material.

To the same end, working with imagery that is not included in the EE catalog requires laborious processes. Simple functions that can be completed in ENVI such as color mapping, are much more complex to mimic using EE palette classification (Figure 4). The ability to obtain a cursor location/value in ENVI is something that does not seem possible in EE, with or without programming abilities (Figure 3). It would be nice to have some basic tools, and UI function in EE similar to those offered in the standalone platforms. Below are some screenshots demonstrating or displaying the comments mentioned above.

Envi Features & Functionality

am1
Figure 1. Thermal (4,5,7) Band Selection-ENVI
am2
Figure 2. Color Mapping with Color Tables-ENVI
am3
Figure 3. Cursor Location/Value- ENVI

 

Earth Engine GUI & Functionality

am4
Figure 4. Color Palette (without programming & Clouds)
am5
Figure 5. Landsat 8 32-Day NDWI- Precomputed (EE)
am6
Figure 6. Landsat 8-Annual NDVI- Precomputed (EE)
am7
Figure 7. Thermal Band Selection(4,5,7)- EE

Survey of Google Earth Engine for academia and in comparison with existing platforms and publicly available data.