Correlation of Total suspended Sediments and Reflectance of Landsat TM in Laguna De Bay, Philippines
Geographic Information system (GIS) Phase
The highest TSS and band reflectance were selected for further processing using ARC/INFO GIS software. The lake boundary, tributaries and fish pens were added into the digitized maps. Landsat TM image was then overlays with the digitized map. Universal Transverse Mercator(UTM) system, to ensure that the satellite image and the digitized map have a common reference point on the earth surface.
Clipping, which separates the land and water features of the lake, was done to avoid mixing of the lake and land pixels. The clipped overlayed image had pixel values ranging from a to 255. Reclassification into four user-defined classes for Band 2 and TSS correlation was used. This process was done by grouping the extreme values (high and low tails) in the normal curve as one group. Followed by computer-generated grouping of pixels in the normal curve area where the highest aggregation of pixels are clustered. The resulting image contains contiguous areas (a connected group of cells ) that have the same value (Aranoff, 1993).
Results And Discussion
The February 1992 period
The correlation coefficients for the linear models showed that the relationship between TSS and the reflection for February 1992 are not significant at 5% level. This therefore leads to the acceptance of the null hypothesis (Table 1).
Table 1. Correlation Coefficients between landsat TM Scenes and TSS for Feb 1992 and April 1993.
| |
Band1 |
Band2 |
Band3 |
Band4 |
Band5 |
Band7 |
| TSS 1992 |
.03 |
-.01 |
-.04 |
.00 |
-.46 |
.51 |
| TSS1993 |
.88 |
.99 |
.96* |
.89 |
.90 |
.94 |
*significant at the 5% level
The non significant correlation may be due to the environmental conditions (Table 2) during sampling which could have affected the image.
Table 2. Detas of Landsat Scenes and Field Data and Environmental Conditons.
Landsat
TM (date)1 |
Field data
(date)2 |
Surface
Temp.(oC)3 |
Precipitation
(mm)3 |
Influencing
wind regimes4 |
NHCS
(Status)5 |
| 11Feb92 |
11Feb92 |
26.36 |
4Feb
Trace amounts |
Northeasterlies |
all gates
closed |
| 02Apr93 |
23Mar93 |
28.15 |
19-22Mar
Trace amounts |
Easterlies |
all gates opended |
Source:
1NAMRIA;
2LLDA;
3PAG-ASA;
4de las Alas et. Al, 1982;
5DPWH
The prevailing meteorological conditions (Table 3) in the PAG-ASA stations dated 11 Feb 1992 show that the maximum wind velocity in the stations were stronger compared to 2 April 1993. Wind forces affect the lake through the formation of waves which then affect reflectance by making the surface of the water rough and may adopt diffuse or lambertian reflectance that causes the energy to be reflected in all directions. This, could hamper the complete capture of the reflected energy by the satellite. Furthermore, the mean cloudiness on 11 Feb 1992 ranged from 1 to 7 octas while on the latter data it was 2 to 3 octas.
The apparent difference between the two images illustrate the operational restrictions on the use of passive remote sensors such as landsat TM due to environmental conditions. However, there are already new technologies and image correction techniques that answer this need. Unfortunately, the cost, availability of data, technical capability and support of the concerned sectors may slow down the acquisition and implementation.
Table 3. Surface meteorological conditions of the Landsat TM Scenes
| Landsat TM (date) |
PAS-ASA (stations) |
Average Wind Velocity (mps) |
Maximum Speed (mps) |
Mean Cloudiness |
Presence of Smoke/haze |
| 11Feb92 |
Ambulong |
1 |
-nil- |
2 |
+ |
| Sta.Cruz |
-nil- |
-nil- |
7 |
+ |
| Anso-Naia |
3 |
8 |
1 |
+ |
| Science Garden |
1 |
10 |
2 |
+ |
| Agromet |
-nil- |
9 |
3 |
+ |
| 02Apr93 |
Ambulong |
1 |
-nil- |
3 |
+ |
| Sta.Cruz |
-nil- |
-nil- |
2 |
+ |
| Anso-Naia |
2 |
3 |
2 |
+ |
| Science Garden |
2 |
-nil- |
2 |
+ |
| Agromet. |
-nil- |
-nil- |
2 |
+ |
| Source: PAG-ASA,; (nil-not in the list); (+) - present; |
The April 1993 period
For the period correlation test revealted that TSS has shown a highly significant correlation with reflectance. These results have led to the rejection of the null hypothesis at 5% level of significance. High correlation coefficients (Table 1) between TSS and Band 2 (0.56-.0.60mm) was noted (r=.99). the same relationship was noted between TSS and Band 3 (0.63-0.69 mm) which is (r=.96).
Several studies in the past have shown similar results in the use especially of Bands 2 and 3. Ritchie (1990) found that the highest correlation between suspended sediment and reflectance wee in Bands 3 and 4. In another study done by lathrop (1991), the simple exponential model that be used ws the Band3/Band1 ratio and it was fund to be a useful index for estimating TSS more of the incoming solr radiation was being reflected and measured by the sensor. the strength of this relationship falls under the region of shorter wavelengths, specifically in the visible region where Bands 1,2,3 belong.
The average reflectance of the pixels of the sampling points as represented by the TSS could be mostly made up of vegetation and soil (Harrison, 1989; ERDAS, 1994). According to Newspaper reports from March 17 up to may 9, 1993, a "fishkill" phenomenon (Bartolome, 1993; Mogato, 1993; Philippine Times Journal, 1993; Magsino, 1993; Nocum, 1993). Which seems to support the idea that algal bloom is one factor that contributes to the reflectance of suspended solids. In addition, the fish pens/cages may also affect reflectance. Since they are made of bamboo or stumps in general, it is probable that the sensor is treating it is totality as vegetation.
Another possible explanation for these results is the concentration of TSS. It has been established by the ADAGWQ (1997), that TSS is comprised of organic and mineral particles that are transported in the water column coming from land erosion, erosion of river channels, and a wide range of industrial and agricultural chemicals. TSS Concentration is also affected by factors such as the actual circulation pattern of the lake and the possible influence of the opening and closing of NHCS.
