Data Source Regression Analysis

Based on the regression result, the spatial distribution and the estimated area of rice field in the research area can be identified by using formula: Rice Area Y = Area Group D + Group F + Group G + Group H + Group J By summarizing the EVI Group EVI Group D, Group F, Group G, Group H, and Group J, the estimation of rice field can be measured and compared with the existing data from BPS for validation see table 4-3 and figure 4-5. A bias between estimated area and BPS data can occurred in this process because MODIS EVI data has the spatial resolution of 250m by 250m pixel size, meanwhile the real condition of rice field can be smaller or bigger than 250m by 250m. Table 4-3. Rice field area estimation GROUP Area Ha D 19.603,9805 F 73.686,4218 G 86.752,9034 H 65.791,9433 J 38.516,5273 Total Rice Field Area 284.351,7763 The harvested areas from BPS data then were compared with the area estimation result from image data. The data from BPS seems to be larger from the estimation from image data because BPS data measured the harvested area which means that when the rice plantations were harvested twice a year then the area also multiplied by two most of rice variety planted in the research area were harvested twice a year. To compare the area from BPS data with the result from image data, the estimation area need to be multiplied by 2 times of harvesting which gives the result of 568.703,5526 Ha of estimated harvested area. The error was under estimated compared to BPS data with the difference of 7,38 for rice field area. Figure 4-5. Rice crop spatial distribution Once the area of rice field estimated based on statistic regression, to estimate the rice production information on the rice variety that being planted in the research area are needed. Because most of the rice field in the research area were own by individual and farmers are allowed to plant any type of rice variety on their field, this research only use the dominant rice variety that being planted as the parameter to estimate rice production. The selection of rice variety that being planted by farmers usually coming from the ability of certain rice variety to have high productivity and short growing seasons and also stands for certain pest. The dominant rice variety that being planted in the research area based on information from Balai Besar Padi – Ministry of Agriculture are Ciherang, IR64, and Cilamaya Munjul. The variety of rice and its productivity become the base in estimating the rice productivity. The productivity of Ciherang, IR64, and Cilamaya Munjul are ranging from 5 to 8 tonsHa. To estimate the total production of rice in the research area, the total rice field area then multiplied by the average of rice productivity in this case is 6,5 tonsHa and harvesting time in one year which shown by two bell shape phenology pattern in each EVI groups. The estimation of rice production can be seen in the calculation below: Rice Production Ton = 284.351,7763 Ha 6,5 TonsHa 2 time harvesting = 3.696.573,09 Tons To validate the rice field area and rice production estimation, the result was compared to the existing data of rice production of 2008-2010 collected from BPS see table 4-4. The statistic data for Karawang Regency was not complete because there was no data about Karawang Regency in Figures year 2010, so the comparison was done using 2009 data. Based on the comparison, the estimation result was under estimate compare to the crop statistic. The error of the estimation are 10,36 for rice production. Table 4-4. Crop statistic from BPS REGENCY 2008 2009 2010 HARVESTED AREA Ha PRODUCTION TON HARVESTED AREA Ha PRODUCTION TON HARVESTED AREA Ha PRODUCTION TON KARAWANG 194,536.00 1,255,118.00 195,670.00 1,362,357.00 NO DATA NO DATA SUBANG 172,447.00 1,091,612.00 185,209.00 1,128,353.00 174,337.00 959,533.00 INDRAMAYU 190,090.00 1,229,476.75 229,784.00 1,588,866.12 239,698.00 1,557,552.30 TOTAL 557,073.00 3,576,206.75 610,663.00 4,079,576.12 414,035.00 2,517,085.30 The estimation results from the calculation were under estimate for both area and production of rice. This condition was caused by the resolution of MODIS imagery that covers 250m by 250m for one pixel. The coarse resolution of MODIS imagery cannot identify rice field smaller than 250m by 250m area and causing that some of the rice fields were not detected and gives less result compared with BPS data. On the other hands, the error value for rice field area were less than 10 which means that more than 90 pixel were able to identify rice field. The high error on production estimation which gives value of 10,36 were mainly caused by the identification of rice variety that being planted in the research area. This research only considers three types of rice variety which is dominant in the research area, meanwhile in the field farmers not only planted Ciherang, Cilamaya Munjul, and IR64 rice variety.

4.4 Rice Planting Rotation

The temporal EVI images show the difference of planting date in research location. From the EVI temporal pattern, the planting and harvesting date were rotating starting from the south area towards north of the area see figure 4-6. Figure 4-6 shows the EVI value in research area where the white color represent the panicle initiation phase of rice plantation where rice leaf is at the greenest color phase. Through the time white color are shifting from south area towards north area or in other word that the planting rotation were started from the south and then shifting to the north. In figure 4-7, rice ages in research area were able to be determined based on EVI value through time. Rice age were identified by growing phases of rice plantation that consist of pre-flooding phase, germination phase, tillering phase, panicle initiation phase, flowering phase, and harvesting phase. The figure also describes the planting rotation of rice through time shifting from south towards north. The type of rice field in the research location were dominated by irrigated rice field which causing that the planting and harvesting date moving towards north through time were the irrigation water started from the south region. Each EVI classes then were examined to determine the growing season of rice for each class as a base for generating a crop calendar. Growing seasons of rice were varied according MODIS EVI temporal pattern. The variations ranging from 96-128 days and for each class there is a rotating days of planting pattern ranging from 16-48 days where the changes are moving backwards according to number of classes see figure 4-8 and table 4-5. Figure 4-6. Time series EVI images showing the movement of planting and harvesting time red line showing the boundary of rice field and non-rice field Figure 4-7. Time Series of Rice Age Based on EVI Value Figure 4-8. Phenology pattern of rice in 2008-2010 Table 4-5. Growing Seasons of Rice of MODIS EVI Classes G roup Growing Season GS Days 2008 GS 1 2008 GS 2 2009 GS 1 2009 GS 2 2010 GS 1 2010 GS 2 D 18 Feb – 9 Jun 12 Aug – 16 Nov 4 Feb – 12 Jun 15 Aug – 28 Nov 8 Mar – 15 Jun 2 Aug – 8 Dec F 1 Jan – 8 May 9 Jun – 13 Sep 19 Jan – 25 Apr 12 Jun – 2 Oct 6 Jan – 14 May 15 Jun – 22 Nov G 2007 – 22 Apr 24 May – 12 Aug 18 Dec 08 – 24 Mar 11 May – 15 Aug 6 Jan – 28 Apr 15 Jun – 3 Sep H 2007 – 21 Mar 22 Apr – 27 Jul 16 Nov 08 - 8 Mar 25 Apr – 30 Jul 21 Dec 09 – 12 Apr 14 May – 2 Aug J 2007 – 5 Mar 6 Apr – 11 Jul 2 Dec 08 – 8 Mar 9 Apr – 28 Jun 6 Jan – 27 Mar 28 Apr – 17 Jul The planting rotation pattern from south to north area is caused by the water irrigation system that irrigates the rice field in Karawang, Subang, and Indramayu Regency. The water flow of irrigation systems in the research area were started from the south of research area where two dams located and become the water source for irrigation. The research location were irrigated by two river system; Citarum river system with Jatiluhur Dam that irrigate Karawang, Subang, and west part of Indramayu Regency, and Cimanuk-Cisanggarug river system with Jatigede Dam that irrigate most of Indramayu Regency see figure 4-9. The irrigation systems in Karawang, Subang, and Indramayu Regency become the most important variable in the growing of rice and related to rice crop production. Most of the rice variety that planted in the northern coast of Java Island is very dependent to water irrigation. That is why if there is a problem within water distribution for rice, the production will decrease because of many rice fields cannot be harvested. Figure 4-9. Irrigation Region of Karawang, Subang, and Indramayu Regency From the growing season table 4-5 and the spatial distribution of rice field shows that Indramayu Regency has early planting rotation for each growing season showed by Group J. Meanwhile Karawang and Subang Regency started 16-48 days later which is showed by Group H. Based on the pattern of temporal MODIS EVI images for Group D and Group F, the growing season of rice shows normal pattern of rice growth since 2008 through 2010. Different pattern showed by Group G, Group H, and group J. The last three classes shows normal growing pattern in 2008 and 2009, but in 2010 show disturbed pattern with many variation of EVI value through time see figure 4-10 and figure 4-11. The disturbed phenology pattern shows in time range of day 945 until day 1089 or in calendar date starting August 2010 until December 2010. Figure 4-10. MODIS EVI Pattern of Group D and Group F Figure 4-11. MODIS EVI Pattern of Group G, Group H, and Group J The crop statistic from BPS in table 4-4 shows that rice productions in 2010 were decreasing compare to 2009 production see figure 4-12. In Subang Regency the rice production decreases from 1.128.353 Tons in 2009 becoming 959.533 Tons in 2010. While in Indramayu Regency, rice production decreases from 1.588.866,12 Tons in 2009 becoming 1.557.552,3 Tons in 2010.