|
|
SOIL AND VEGETATION OPTICAL PROPERTIESBrigitte Leblon, Ph.D.Remote Sensing and GIS laboratory,
Faculty of Forestry and Environmental Management
University of New Brunswick, Fredericton (NB), Canada, E3B 6C2 Phone: (506) 453-4924; Fax: (506) 453-3538; E-mail: bleblon @ unb.ca |
Solutions to Exercises
Exercise #1
1.1. Which bands are the most correlated? Red and near-infrared in each case
Verify using (i) all the data and (ii) each data series. In each case, present your results on the form of a correlation matrix.
Series #1 (Clear days) (N = 9)
|
Green |
Red |
Near-infrared |
|
|
Green |
1 |
||
|
Red |
0.9772 |
1 |
|
|
Near-infrared |
0.9386 |
0.9842 |
1 |
Series #2 (Cloudy days) (N = 3)
|
Green |
Red |
Near-infrared |
|
|
Green |
1 |
||
|
Red |
-0.1798 |
1 |
|
|
Near-infrared |
-0.2527 |
0.9972 |
1 |
Series #1 + #2 (N=12)
|
Green |
Red |
Near-infrared |
|
|
Green |
1 |
||
|
Red |
0.9216 |
1 |
|
|
Near-infrared |
0.8941 |
0.9859 |
1 |
Draw the scatterplot between the most correlated bands, by using data from both series and define a possible method to distinguish among both data series.
Figure 5. Scatterplot of data from both series between red and near-infrared bands.
A possible method will be to draw the scatterplot between data related to the most correlated bands, i.e., red and near-infrared. Data of cloudy days are grouped.
1.2. By calculating a relative reflectance difference between the most contrasted cases, determine on which band the influence of the soil color is the less important for series #1? Use mean reflectance values for a given object, if necessary. Present your results in a table.
Soil color:
|
Objects |
Green |
Red |
Near-infrared |
|
Dry sand soil |
17.2 |
17.8 |
23.3 |
|
Dark brown wet soil |
6.7 |
7.1 |
11.9 |
|
Relative reflectance difference (%) |
61.0 |
60.1 |
48.9 |
This table shows that soil color has the lowest effect on the near-infrared reflectance
Determine on which band the influence of the soil type is the less important for series #2? Present your results in a table.
Soil type:
|
Objects |
Green |
Red |
Near-infrared |
|
Sand |
9.0 |
5.9 |
10.3 |
|
Gravel |
10.2 |
4.3 |
7.4 |
|
Relative reflectance difference (%) |
-13.3 |
27.1 |
28.2 |
This table shows that soil type has the lowest effect on the green reflectance.
1.3. Calculate and draw the soil line, using (i) all the data and (ii) each data series.
Figure 6. Soil line for clear days (series#1)
Figure 7. Soil line for cloudy days (series#2)
Figure 8. Soil line for both kind of days (series#1+2)
In each case, do wet and dry soils belong to the same line? Yes
Where are they positionned on each line?
Dry soils are on the highest part of the soil line, whereas wet soils are on the lowest part of the soil line
Exercise #2
2.1. Verify which bands are the most correlated (i) for each data series and (ii) for both series together. In each case, present your results on the form of a correlation matrix. In each case, do not use data acquired on water, because we are looking for correlations related to vegetated targets.
Series #1 (Clear days) (N = 7)
|
Green |
Red |
Near-infrared |
|
|
Green |
1 |
||
|
Red |
0.9354 |
1 |
|
|
Near-infrared |
-0.1211 |
-0.3040 |
1 |
Series #2 (Cloudy days) (N = 7)
|
Green |
Red |
Near-infrared |
|
|
Green |
1 |
||
|
Red |
0.9682 |
1 |
|
|
Near-infrared |
0.7113 |
0.6069 |
1 |
Series #1 + #2 (N=14)
|
Green |
Red |
Near-infrared |
|
|
Green |
1 |
||
|
Red |
0.9576 |
1 |
|
|
Near-infrared |
0.1997 |
0.0759 |
1 |
In all cases, the most correlated bands are the green and the red ones.
Try to explain the difference with the results of Question 1.1 of Exercise#1.
For soil objects, the most correlated bands were the red and near-infrared bands (Question 1.1 of Exercise#1), whereas for vegetated objects, they are the green and red bands. Indeed, in the case of soils, reflectance increases almost linearly with wavelength, from blue to near-infrareed bands (Figure 1), whereas for vegetated targets, because of the red absorption band due to the chlorophyll (Figure 3), refectance do not more increase linearly with wavelength. Also, in the first case (soil), near-infrared and red bands are closer than red and green bands.
Draw the scatterplot between the most correlated bands, by using both data series together.
Figure 9. Scatterplot of data of both series between red and green bands.
2.2. Add the data of Exercise#1. Let us consider all these data as typical spectral signatures of classes (water, vegetation, soil, ...). Which 2D-scatterplot cannot be used to discriminate these classes?
The scatterplot which cannot be used for discriminating between vegetation, soil and water classes is the one built with the green and red reflectances.
Figure 10. Scatterplot of reflectances acquired in the red and green bands.
2.3. Calculate the following vegetation indices (RVI, NDVI, GEMI, PVI, TSAVI) for each object and each series. For PVI and TSAVI, use adequate soil line parameters with regard to the series number. Present your results in a table (one per series).
Series#1
|
Object |
RVI |
NDVI |
GEMI |
PVI |
TSAVI |
|
Dry sandy soil |
1.31 |
0.13 |
-95.05 |
0.18 |
0.01 |
|
Half-wet soil |
1.43 |
0.18 |
-16.26 |
-0.23 |
-0.02 |
|
Brownish wet soil |
1.46 |
0.19 |
-41.43 |
0.52 |
0.03 |
|
Wet soil |
2.48 |
0.43 |
-16.96 |
1.13 |
0.21 |
|
Dark brown wet soil (with high content of organic matter |
1.68 |
0.25 |
-28.31 |
0.81 |
0.08 |
|
Gray wet soil |
1.22 |
0.10 |
-3.81 |
-1.30 |
-0.17 |
|
Yellowish-green grass |
6.34 |
0.73 |
-1262.82 |
20.94 |
0.70 |
|
Green grass |
9.72 |
0.81 |
-1566.40 |
23.96 |
0.80 |
|
Pine needles on grass |
3.45 |
0.55 |
-811.76 |
15.29 |
0.50 |
|
Beech leaves in fall on grass |
3.04 |
0.50 |
-694.99 |
13.61 |
0.45 |
|
Yellow grass |
3.10 |
0.51 |
-810.72 |
14.86 |
0.46 |
|
Yellow leaves on 95% grass |
11.31 |
0.84 |
-2984.87 |
33.63 |
0.83 |
|
Yellow leaves on 50% grass |
3.69 |
0.57 |
-1589.04 |
22.13 |
0.54 |
|
Water |
0.73 |
-0.16 |
2.02 |
-2.72 |
-0.94 |
Series#2
|
Object |
RVI |
NDVI |
GEMI |
PVI |
TSAVI |
|
Wet soil |
1.77 |
0.28 |
-13.37 |
0.06 |
0.01 |
|
Sand |
1.75 |
0.27 |
-21.45 |
-0.02 |
0.00 |
|
Gravel |
1.72 |
0.26 |
-8.45 |
-0.04 |
-0.01 |
|
Yellowish-green grass |
5.99 |
0.71 |
-1274.39 |
14.89 |
0.64 |
|
Green grass |
9.90 |
0.82 |
-1358.09 |
16.77 |
0.77 |
|
Pine needles on grass |
4.64 |
0.65 |
-742.71 |
10.57 |
0.55 |
|
Beech leaves in fall on grass |
3.47 |
0.55 |
-631.13 |
8.34 |
0.42 |
|
Yellow leaves on grass |
2.47 |
0.42 |
-958.66 |
6.66 |
0.23 |
|
Water |
1.06 |
0.03 |
2.14 |
-1.15 |
-0.39 |
For each series, which index is the best to distinguish vegetation from soil-type objects, vegetation from water-type objects as well as water from soil-type objects. You may use relative difference between mean values of vegetation indices to solve this question. Present your results in a table (one per series).
Series#1
|
RVI |
NDVI |
GEMI |
PVI |
TSAVI |
|
|
SOIL |
1.50 |
0.19 |
-37.53 |
0.00 |
0.00 |
|
VEGETATION |
5.81 |
0.65 |
-1388.66 |
20.63 |
0.61 |
|
WATER |
0.79 |
-0.12 |
2.35 |
-2.32 |
-1.50 |
|
SOIL/ |
-2.88 |
-2.48 |
-36.00 |
6.07E+08 |
-8008.85 |
|
VEGETATION/WATER |
0.86 |
1.18 |
1.00 |
1.11 |
3.46 |
|
SOIL/WATER |
0.47 |
1.64 |
1.06 |
-6.84E+07 |
19693.24 |
Series#2
|
RVI |
NDVI |
GEMI |
PVI |
TSAVI |
|
|
SOIL |
1.75 |
0.27 |
-14.43 |
0.00 |
0.00 |
|
VEGETATION |
4.60 |
0.59 |
-935.67 |
10.32 |
0.46 |
|
WATER |
1.01 |
0.01 |
2.21 |
-1.25 |
-0.43 |
|
SOIL/ |
-1.64 |
-1.16 |
-63.86 |
-1.77E+05 |
3580.95 |
|
VEGETATION/WATER |
0.78 |
0.99 |
1.00 |
1.12 |
1.92 |
|
SOIL/WATER |
0.42 |
0.97 |
1.15 |
2.13E+04 |
-3306.36 |
For both series, the best index to differentiate soil from vegetation or from water is PVI. The best distinction between vegetation and water is done with TSAVI.
2.4. Demonstrate mathematically that NDVI = tan (a - 45°), if RVI = tan (a)
[i) consider the definition of NDVI and RVI;
ii) determine the relationship between NDVI and RVI;
iii) calculate trigonometric transformations; and iv) Eurêka! you are done!;
Be careful: tan (a - 45°) ≠ tan (a) - tan (45°)]
(9)