5. RESULTS AND DISCUSSION 
5.1 Characteristics of LULC NDVI profiles 
The mean NDVI profile for each land-use class was presented in 
Figure 2. There were distinctions between cropping patterns 
throughout the year. The single-cropped rice was practiced once 
a year. The peak (heading date) exhibited once a year by the end 
of December. Double-cropped rice has two peaks, indicating two 
rice crops being practiced per year. However, the peaks were 
shifted depending on the region attributed to the local climatic 
conditions. Similarly three peaks were observed for 
4 4 
0.8 
0.6 
0.4 
NDVI 
0.2 
  
Z $^» Double-cropped rain-fed rice 
«== Triple-cropped irrigated rice 
224. | 7^ Field crops 
». Forests/orchards/fruit trees 
^t Built-up areas 
-0.4 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B3, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
     
  
«p= Gingle-cropped rain-fed rice 
7ii- Double-cropped irrigated rice 
triple-cropped rice because three crops were practiced per year. 
Farmland characterized by field crops (e.g., sugarcane, cassava, 
and maize) revealed one peak in the NDVI profile by the end of 
the crop because field crops were usually planted once a year in 
the rainy season. There was an increase in vegetation activity in 
the field crop NDVI profile before the cultivating season due to 
the growth of vegetation. Built-up areas including high albedo 
surfaces (e.g., concrete, metal proof, parking lots, and roads) had 
low NDVI values and were stable fluctuation throughout the 
year. 
n a sende 
  
TT TF FT _ _— _ _ _— _ _ _ _ peer rase aqua sn 
1 25 49 73 97 121 145 169 193 217 241 265 289 313 337 361 
DOYs (2010) 
Figure 2. Mean NDVI profiles of land-use types. 
5.2 Classification results 
The 2010 classification results were presented in Figure 3. In 
general, single-crop rice was commonly practiced in the 
lowlands of Thailand and areas surrounding the Cambodian 
Tonle Sap Lake. Double-cropped rice was mainly concentrated 
in three deltas: RRD, MRD, and CRD. Triple rice cropping was 
especially observed in MRD and CRD that received irrigation 
water around the year. Field crops occupied the uplands, 
especially in the upper parts of Thailand. Forests were generally 
found in slope and mountainous areas where the land was not 
suitable for rice agriculture due to major soil constraints. 
The error matrix of classification accuracy assessment was 
shown in Table 1. Of 12,000 pixels extracted from the ground 
reference image for accuracy assessment of the 2010 classified 
map, the results indicated that the overall accuracy was 80.3% 
and the Kappa coefficient was 0.76. Of the total of 2,000 pixels 
checked to measure the accuracy in each class, the two classes 
with the highest producer accuracy levels were single-cropped 
rain-fed rice (88.7%) and double-cropped irrigated rice (91.9%). 
The lowest producer accuracy was observed for field crops class 
(67.5%) or 32.5% of pixels incorrectly classified into other 
classes. This could be explained that it was more difficult to 
accurately predict this class because the size of upland crop fields 
was general small and they were scattered. Moreover, the 
temporal confusion caused difficulties in the discrimination of 
this class from other classes. 
   
   
    
    
     
   
  
     
   
   
  
  
  
  
   
   
         
    
  
   
  
   
   
   
   
   
   
   
Grot 
Sing 
Dou 
Trip 
Fiek 
Fore 
Buil 
| Tot 
Proc 
Usei 
Ove