Earthsys 144: Fundamentals of GIScience
Remote Sensing
where is your data? gis.stanford.edu
Remotely-sensed data
aerial imagery
aerial imagery
Gaspar Felix Tournachon, more commonly known as “Nadar,” is credited with taking the first successful aerial photograph in 1858 from a hot air balloon tethered 262 feet over Petit-Bicêtre (now Petit-Clamart), just outside Paris; his original photos have been lost.
aerial imagery
“Boston, as the Eagle and the Wild Goose See It”
October 13, 1860
by
James Wallace Black is the earliest surviving aerial photograph.
Credit: James Wallace Black, The Metropolitan Museum of Art.
aerial imagery
Photograph of San Francisco in ruins from Lawrence Captive Airship, 2000 feet above San Francisco Bay overlooking water front. Sunset over Golden Gate circa 1906
George Lawrence—Prints & Photographs Divison/Library of Congress
aerial imagery
satellite imagery
First image of Earth from outer space, taken by the V-2 No. 13 suborbital spaceflight.
October 24, 1946
The first crude image taken by a satellite, Explorer 6, shows a sunlit area of the Central Pacific Ocean and its cloud cover. The photo was taken when the satellite was about 17,000 mi (27,000 km) above the surface of the earth on August 14, 1959. At the time, the satellite was crossing Mexico.
Corona Program
Mission 13
August 1960
In one mission, Corona imaged 1.5 million square miles of Soviet territory. It revealed 64 new Soviet airfields and 26 missile sites.
Corona satellites would go on to fly more than 130 missions, taking more than 800,000 photographs of the entire Soviet Union and much of the world.
Landsat
the value of remote sensing data
broad coverage
spectral range
geometric accuracy
permanent recording
Dallas - Fort Worth area of Texas on July 25, 1972
permanent recording
Imaging Systems
passive systems
image scanners
active systems
geophysical prospecting
imagery properties
Radiometric Resolution
Spectral Resolution
spectral curves
Cadence (Revisit Rates)
- Landsat - 16 days
- Sentinel 2 - 10 days at the equator, 2-3 at mid-latitudes
- Planetscope Doves - Daily
- MAXAR/SkySat/Pelican - varies, tasked
image enhancement
stretching & clipping
rendering methods
Nearest Neighbor
Bilinear Interpolation
Cubic Convolution
pan-sharpening
imagery interpretation
Band Combinations
visible
infrared
8-bit color
Band Combinations
Band Combinations
Bands 3,2,1
Bands 4,3,2
Bands 7,4,2
Band Math (Indices)
Normalized Difference Vegetation Index (NDVI) = (NIR - R) / (NIR + R)
Panel Index Purpose Interpretation
1. Natural Composite |
R-G-B (True Color) |
Human-eye view |
Good for general orientation, but limited in analytical value. |
2. NDVI(Normalized Difference Vegetation Index) |
(NIR − R) / (NIR + R) |
Plant vigor and density |
Bright areas = healthy vegetation; dark = sparse or urban. |
3. SAVI(Soil-Adjusted Vegetation Index) |
1.5 × (NIR − R)/(NIR + R + 0.5) |
NDVI + soil brightness correction |
Improves vegetation contrast in low-canopy or bare-soil zones. |
Panel Index Purpose Interpretation
| 4. EVI(Enhanced Vegetation Index) | 2.5 × (NIR − R) / (NIR + 6R − 7.5B + 1) | Reduces atmospheric & soil noise | More nuanced vegetation health; sensitive to canopy structure. |
| 5. NDRE(Normalized Difference Red Edge) | (NIR − RE) / (NIR + RE) | Chlorophyll concentration | Excellent for later-stage crop monitoring and stress detection. |
| 6. NGRDI(Normalized Green-Red Difference Index) | (G − R) / (G + R) | Quick vegetation index using only visible bands | Useful when NIR isn’t available—more sensitive to chlorosis. |
Panel Index Purpose Interpretation
| 7. MNDWI(Modified Normalized Difference Water Index) | (G − SWIR1) / (G + SWIR1) | Water delineation | High = open water, dark = land; vegetation & soil are suppressed. |
| 8. NDBI(Normalized Difference Built-up Index) | (SWIR − NIR) / (SWIR + NIR) | Urban and impervious surface detection | Red areas = built-up zones; vegetation appears dark. |
| 9. NDSI(Normalized Difference Soil Index) | (SWIR2 − B) / (SWIR2 + B) | Bare soil & sand mapping | High values = bare earth or dry land; useful in ag and arid zones. |
Landcover Classification
Landcover Classification
Commonly used algorithms for supervised classification in remote sensing, include:
random forest (RF)
artificial neural networks (ANN)
support vector machines (SVM)
change detection
2001
2006
2011
