A Comparison Of Spectral Indices For Different Treatments

A COMPARISON OF SPECTRAL INDICES FOR DIFFERENT TREATMENTS Callie Sand

Carleton College Student Airborne Research Program

Overview 

Introduction  California Drought  Massive Agriculture Industry  Maximization of Water Resources



Experiment  Case Study: Paramount Farms  MASTER  Field Data  Indices!



Results  Comparison of fields  Calculated Indices



Conclusions and Future Experiments

The Central Valley and Kern County 



   

Major agricultural region in Central Valley Central Valley: 530,000 acres of almonds $2 billion industry 100% of US’s supply 75% of world’s supply Soil and climate



Water shortages in California  3rd year of drought  Increasing population  Almonds = long-term investments



How to produce more produce with less water?  Central Valley Project: 10% allocation

Constant observation of plant stress levels  Satellite models versus time-consuming ground-based measurements 

Paramount Farms  Nitrogen/Calcium/Potassium

Treatments  Different Water Stress Levels  Part of a larger experiment going on for several years  Previously discovered data: almonds require more water than previously thought for optimal production  Field Health and Crop Indices

MASTER  MODIS/ASTER

airborne simulator  50 spectral bands in four spectral regions Visible through thermal infrared

ER-2 Sensor Coverage

65,000 ft

24 in. LENS 8 n. mi.

6.6 n. mi./min

12 in. LENS

IRIS II Panoramic Camera MAS, MASTER, AOCI, MAMS TMS

4 n. mi 8 n. mi.

6 in. LENS

20 n. mi. 21.4 n. mi. 16 n. mi

2 n. mi. at NADIR

16 n. mi

Image courtesy of Jeff Myers

DC-8 Flight 11,500 ft altitude  Flew over Paramount Farms and Sheely Farms  Field data taken July 20-25  Flight data used from Friday 

Cloud coverage Missed ROI



Rough calibration performed to smooth out image 7 meter resolution

Indices: Methodology Overview  Three

types: greenness/canopy health, water, and foliar chemistry Way of analyzing remote sensing

measurements Multiple ways to analyze various properties of the image Limitations in the spectral band measurements dictated what ones could be calculated ○ Gap from ~1000 to 1600 nm

Indices  Calculate

indices using ENVI software Make each band a different index Test for correlations: r>.9 indicates similar data ○Only use one of the correlated indices ○Pick best choice ○Final indices: NDVI, PI2, WBI

Indices  Comparison

of WBI/PI2/NDVI

 WBI  PI2:

Plant Stress Status

More fluorescence ~ more stress Short-term health

 NDVI:

Ranges from -1 to 1

Closer to 1 indicates denser canopy

Treatments & Indices  Use

GPS points to map out regions corresponding to the different treatments 12 treatments in all Take the average of each treatment area

Comparisons

Conclusions Plots in the best current health overall are treated with UN 275/KTS 120  Best vegetation coverage (indicator of future health) is UN 275/KTS 75, though UN 275/KTS 120 is second-best  However, no strong correlations between data - Wrong map? - Need more indices that are not available through our MASTER data 

Water & Indices  Different

amounts of water applied to different fields  Mask to identify almond fields (different talk) Took 8 fields from those visible Comparison of WBI/NDVI/PI2

Water & Indices

8 1 2 3 5 7 4 6

Conclusions Fields 5, 6, and 3 in the best health  However, Field 6’s trees have the least amount of water in their leaves  Strong correlations between NDVI and PI2  Can’t conclusively say which amount of water is best for trees until irrigation amounts known  Amount of almonds harvested off of each field must be known as well  LAI 

Acknowledgements 

NSERC, NASA, UC Davis and everyone involved in data measurements, Susan Ustin, Shawn Kefauver, entire evapotranspiration team, SARP etc.

References  







Almond history. Almond Board of California, http://www.almond-board.com/index.cfm Glenn, E.; Huete, A.; Nagler, P.; Nelson, S. Relationship between remotely-sensed vegetation indices, canopy attributes and plant physiological process. Sensors 2008, 8, 2136-2160. Peñuelas, J.; Piñol, J.; Ogaya, R.; Filella, I. Estimation of the plant water concentration by the reflectance Water Index WI (R900/R970). Int. J. Remote Sensing 1997, 18:13, 2869-2875. Zarco-Tejada, P.J. et al. Vegetation stress detection through chlorophyll a+b estimation and fluorescence effects on hyperspectral imagery. J. Environ. Qual. 2002, 31, 1433-1441. Petropoulos, G.; Carlson, T.N.; Wooster, M.J.; Islam, S. A review of Ts/VI remote-sensing based methods for the retrieval of land surface energy fluxes and soil surface moisture. Progress in Physical Geography 2009, 33, 224-252:

Questions?