MULTI-TEMPORAL VEGETATION INDEX ANALYSIS ON PEAT LANDS IN MUARO JAMBI REGENCY, JAMBI PROVINCE, INDONESIA

Transcription

1 MULTI-TEMPORAL VEGETATION INDEX ANALYSIS ON PEAT LANDS IN MUARO JAMBI REGENCY, JAMBI PROVINCE, INDONESIA Annisa Nurdiana, Yudi Setiawan Center for Environmental Research, Bogor Agricultural University, Bogor 16680, Indonesia, KEY WORDS: EVI, land use/land cover change, MODIS, NDWI, temporal analysis ABSTRACT: Peat land in Indonesia is being damaged severely by land development and fire, including peat lands in Muaro Jambi Regency, Jambi Province, Indonesia. Deforestation and drainage put pressure in peat lands in Muaro Jambi. The purpose of this study is to detect and monitor changes in peat lands in period of using multi-temporal EVI and NDWI. By knowing the temporal patterns of vegetation index (EVI and NDWI) obtained from MODIS data, we can see the significant changes in the patterns that describes land use/land cover changes in an area. Wavelet transform applied to reduce residual noises of the values. Results of the analysis show that there is a slight gap value between EVI and NDWI, however both index have a similar pattern characteristic. Most of the changes detected by the index occurred in August-October with low rainfall, high temperature and high hotspot number. These conditions help people to burn faster and easier, but also worsen its effects. 1. INTRODUCTION Earth resource satellites data are very applicable and useful for land use/cover change detection studies (Yuan et al., 2005; Brondizio et al., 1994). Application of remote sensing data made possible to study the changes in land cover in less time, at low cost and with better accuracy (Kachhwala, 1985). According to Setiawan and Yoshino (2014); Setiawan et al., (2013, 2014), temporal characterization of vegetation dynamics in a long period can be used to detect changes in Earth s surface, whether the change is gradual due to extreme climatic variations, or changes dramatically due to human or natural (e.g, land use change and forest fires). Population growth push peat land conversion to agricultural land in order to support food security, meet the paper industry raw materials, meet the plantations and the development of bioenergy. Use of peat lands in Indonesia at this time is the conversion for pulp plantations and oil palm plantations. One of the main trigger is because of the government program to look for alternative energy by utilizing palm oil as a renewable alternative energy. Human activities in the peat ecosystem has increased rapidly in the last two decades so that many areas of peat land become degraded and easier to burn (Miettinen and Liew, 2010). Jambi province is one of the largest carbon storage in Indonesia, particularly in Sumatra. Based on the Ministry of Forestry data, in 2011 Jambi Province has 676,341 hectares of peat land area. By this fact, Jambi Province placed in the seventh largest peat land in Indonesia. This area is about 10% of the total national peat lands. With data of the area, it can be estimated that there will be a huge potential losses if the land is not managed properly. The distribution of peat lands in Jambi province are in the downstream area. Most are part of a cluster of east coast of Sumatra. Sequentially, the largest peat land spread in East Tanjung Jabung (46%), Muaro Jambi (30%), and West Tanjung Jabung (20%). Muaro Jambi is one of four Green Prosperity starter regencies selected based on the potential to achieve significant reductions in poverty based on economic development opportunities, renewable energy potential, and environmental factors such as peat land degradation. Muaro Jambi Regency was formerly part of Batanghari Regency and became Muaro Jambi Regency in 1999, along with three other new regencies in Jambi. The regions encircle the city of Jambi. Like most coastal and downstream river areas, the regency s population varies with the number of immigrants. Muaro Jambi is home to the Berbak National Park, a protected area in Indonesia s Ramsar Site and an internationally recognized waterfowl habitat. The national park has a total peat land area of about 110,000 hectares while a forest park (Taman Hutan Raya or Tahura) covers 60,000 hectares. Land use in the regency is dominated by dry agricultural land (293,256 hectares) followed by oil palm plantations (87,992 hectares) and wet agricultural land suitable for rice (17,000 hectares). Agriculture and mining (largely petroleum mining) are the main economic sectors, respectively contributing 30% and 26% of gross regional economic product (Statistics Indonesia, 2012). Oil palm plantations are the largest contributor of GDP in the agriculture sector. High returns from oil palm and other agricultural uses put pressure on the forests in Muaro Jambi. Other factors in forest degradation are: a legacy of ineffective land use licensing prior to the establishment of the new regency; confusing forest boundary demarcation; land disputes; illegal logging; and peat fires due to drainage.

2 The objective of this research is to monitor the land use/land cover changes on peat land using multi-temporal data of EVI and NDWI during a known period in Muaro Jambi Regency, Jambi Province, Indonesia. Hotspot and climate data are also used as supporting data so we can find out that the changes detected by vegetation index related to hotspot occurrence and climate. 2. METHODOLOGY 2.1 Study Area Muaro Jambi Regency is geographically located at E E and 1 15 S S, it is surrounded by Batang Hari and West Tanjung Jabung Regency in the west side, East Tanjung Jabung Regency in the east and north side, and South Sumatera in the south side. Muaro Jambi has an area of 5,326 km 2 and is divided into 153 village and 11 districts, which are district of Jambi Luar Kota, Sekernan, Kumpeh Ilir, Maro Sebo, Mestong, Kumpeh Ulu, Sungai Bahar, and Sungai Gelam. Figure 1. Study area Based on the observation in Jambi Province Climatology Station in Sungai Duren, the average air temperature in Muaro Jambi in 2013 is 26.8 C and the highest temperature occurs in June 2013 is 33.3 C. Average air humidity in Muaro Jambi in 2013 is 85.8%, an average rainfall is mm, and most rainy days are 24 days in May and December (BPS, 2014). The regency is home to the Berbak National Park. Berbak National Park is a protected area of wetlands that are important in Southeast Asia since it is pointed as an area of Ramsar (International Wetlands). Aside from Muaro Jambi, the park administratively located in East Tanjung Jabung Regency. 2.2 Satellite Imagery EVI was developed to optimize the vegetation signal with improved sensitivity in high biomass regions through a decoupling of the canopy background signal and a reduction in atmospheric influences (Huete et al., 2002). EVI also

3 minimizes atmospheric influences with the "aerosol resistance" term which uses the blue band to correct aerosols influence in band red (Huete et al. 1997). In this study, we used the time series MODIS EVI (embedded in the MOD13Q1 product), which was filtered by wavelet transforms (Setiawan et al., 2011), and MOD09A1 product to derive NDWI estimation value. The datasets were acquired from January 2001 to December MOD13Q1 captured in a 299 time series with an interval of 16 days, while MOD09A1 captured in a 598 time series with an interval of 8 days. Information on the datasets was obtained from which is maintained by the NASA Land Processes Distributed Active Archive Center (LP DAAC) at the USGS/Earth Resources Observation and Science (EROS) Center, Sioux Falls, South Dakota (2009). The MODIS LAND Discipline Group (MODLAND) developed the EVI with this following equation: EVI G nir C nir 1 red red C 2 blue (1 L) L where, nir, red and blue are the remote sensing reflectances in the NIR, red and blue, respectively, L is a soil adjustment factor, and C 1 and C 2 describe the use of the blue band in correction of the red band for atmospheric aerosol scattering. The coefficients, C 1, C 2, and L, are empirically determined as 6.0, 7.5, and 1.0, respectively. G is a gain factor set to 2.5 (Huete et al. 1997). The Normalized Difference Water Index (NDWI) (Gao, 1996) is a satellite-derived index from the Near-Infrared (NIR) and Short Wave Infrared (SWIR) channels. The SWIR reflectance reflects changes in both the vegetation water content and the spongy mesophyll structure in vegetation canopies, while the NIR reflectance is affected by leaf internal structure and leaf dry matter content but not by water content. The combination of the NIR with the SWIR removes variations induced by leaf internal structure and leaf dry matter content, improving the accuracy in retrieving the vegetation water content (Ceccato et al. 2001). Since it is useful for drought monitoring, NDWI expected to also detect land use/land cover change associated with land clearing by burning. The study area is covered by MODIS tiles h28v09. The images were then clipped to cover the Muaro Jambi peat lands and sequentially stacked to produce the EVI and NDWI time-series datasets. Accordingly, peat land area of Muaro Jambi can be characterized by regular EVI sequence at 299 time series with the interval time 16 days and by NDWI sequence at 598 time series with the interval time 8 days. Besides, we also used hotspot data from MODIS (obtained from rainfall and air temperature from Indonesian Agency for Meteorology, Climatology and Geophysics (BMKG) (obtained from Methods The MODIS data provides information on inter-annual variations of land surface, however, the datasets contain disturbances caused by some errors; such as atmospheric variability (Huete and Liu, 1994), aerosol scattering (Xiao et al., 2003) and some residual errors (Lu et al., 2007). These errors showed as residual noise in the EVI and NDWI composited datasets. To deal with this issue, we used wavelet transform so the inevitable disturbance in the profiles could be reduced. But, we only applied the wavelet transform on EVI datasets, so we can compare between the pattern of the original data of NDWI and wavelet-filtered pattern of EVI. After obtaining the patterns of EVI and NDWI, we used hotspot data derived from MODIS and daily-monthly rainfall data from BMKG as supporting data to help us analyzing whether land use/land cover changes occurred related to hotspot existence and the weather. Image processing in this study was designed as illustrated in Figure RESULTS AND DISCUSSION EVI and NDWI that are stacked for 13 years (2001 to 2013) result patterns that indicate a change at certain times. Both of index patterns have a similar characteristic. When there is a disturbance in an area of vegetation, such as logging, drought, or burning, the vegetation index will decline significantly, while weeding and replanting cause a positive change of patterns. When vegetation is in normal condition, the pattern will move within the normal range. Normal range can be seen from the pattern moving constantly in a certain range in a long time. However, when vegetation get disturbed, the pattern will go down away from the normal range. And when the vegetation has restored or replanted, there are two possibilities: the pattern will go back into its normal range or the pattern will form a new normal range. Change of normal range indicates changes in land use/land cover that is no longer the same with its former condition before the disturbance.

4 MODIS Data 8-day composite for 13 years ( ) MOD09A1 MOD13Q1 MODIS Data 16-day composite for 13 years ( ) Channel 2 & 5 extracted from MOD09A1 (captured 598 time series data) Channel 2 (NIR) Channel 5 (SWIR) EVI value EVI value extracted from MOD13Q1 (captured 299 time series data) Layer-stacking NDWI calculated from channel 2 & channel 5 NDWI value EVI value for 299 time series data Layer-stacking Wavelet transform NDWI value for 598 time series data De-noise EVI values Hotspot Hotspot data from MODIS for 13 years ( ) Analyze Rainfall Rainfall data from BMKG for 13 years ( ) Figure 2. Flow chart. NDWI produces a denser pattern than EVI because NDWI obtained from MOD09A1 data which is 8-day composite while the EVI from MOD13Q1 data is 16-day composite. NDWI pattern still contains residual noises because it is not through the wavelet transformation as is done in the EVI. Even so, the pattern of EVI and NDWI has characteristics that are not much different. The pattern in Figure 1 is the spectral profile of a pixel changes in the time range. The pixel value of the index has decreased significantly as much as 6 times within a period of 13 years. Some other sample points used are also shows that if an area has ever been disturbed, that area would be more vulnerable against another disturbance. As in the case of change of primary forests to pulp or oil palm plantations, negative changing patterns of vegetation indexes would be repeated at specific time period. Oil palm harvesting would occur in every few years and it showed in the patterns. Besides the repeating changes in the patterns, most of the pixels which was originally primary or secondary forest and its vegetation index decline significantly, the index would not rise again into the normal range but went down away the normal range (see Figure 4). In another word, the pixel represented changes in the condition of cover or land use.

5 It is very rare to find primary or secondary forest that has been disturbed and return to the forest as before, unless it is actually carried out by weeding or replanting by the related parties. However, to restore the initial condition of the forest will take a very long time. Figure 3. EVI and NDWI time series data for 13 years ( ) (a) (b) (c) (d) Figure 4. EVI and NDWI time series patterns by different 4 sample points According to four sample points used, the pattern changes coincided with hotspot occurrences. Hotspot data that sourced from MODIS satellite is a pixel with a temperature above 320 K. Hotspots in Figure 4a coincided with a decrease in vegetation index occurred in August 2003 where the number of hotspot in peat land in Muaro Jambi reached 438 out of total 849 in The number of hotspot in August 2004 (Figure 4b) reached 691 out of total 1493 in 2004, in October 2006 (Figure 4b) reached 663 out of total 1179 in 2006, in October 2008 (Figure 4c) reached 69 out of total 127 in 2008, and in September 2011 (Figure 4d) reached 330 out of total 570 in During 2001 and

6 2013, as many as 7179 hotspots appear in Muaro Jambi and 87.35% of them, or 6271 hotspots, located on peat lands. This high number proves that at least in the period of peat lands are more vulnerable to fires. This is very unfortunate to happen considering the great role of peat lands for the environment. Table 1. Monthly air temperature and rainfall according to land changes showed in Figure 4 Month Fig. 4a (2003) Fig. 4b (2004) Fig. 4b (2006) Fig. 4c (2008) Fig. 4d (2011) Air temp rainfall Air temp rainfall Air temp rainfall ( C) (mm) ( C) (mm) ( C) (mm) Air temp rainfall Air temp rainfall ( C) (mm) ( C) (mm) Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Many people assume that the weather or climate can cause forest and land fires. Basically, humans are the main driving factors on fire incidents. When doing burning activities, humans tend to not care about the surrounding environment, so that often found the fire hard to extinguish or spread to the other land and become uncontrollable. These events have been severely aggravated by improper burning time. Based on Table 1, land use/land cover changes that is detected by the EVI and NDWI patterns occurred in months with low rainfall and high air temperature. People who burns may think low rainfall and high temperature as a help so that the burning activities can be faster and easier. But on the other side, it also causes the effect worse. Rainfall is commonly used for assessing forest fire vulnerability because it affects the water content of vegetation. High temperature in Earth s atmosphere which is supported by the low rainfall conditions in the region will cause the fuel to burn more easily. These conditions can aggravate the fires, such as difficulties in extinguishing fires. Moreover, fires in peat land that is occurred below the ground surface will be more difficult to extinguish. Peat land fires occur almost every year and has become an annual agenda in Indonesia. Peat land is a large carbon storage. When peat is drained and or deforested, carbon released to the Earth's atmosphere. Apart from CO emissions, developments are also a threat to the remaining biodiversity in SE Asia as the peat lands are an important habitat for many endangered species. Furthermore, the peat fires cause regional haze (smog) problems that affect public health, activities and economies in the SE Asian region. If the stakeholders do not immediately take an action to prevent and cope with this issue, there will be more problems to be faced by the country. 4. CONCLUSION The EVI and NDWI patterns have a similar characteristic and both are able to detect land use/land cover changes in peat lands in Muaro Jambi Regency. Most of changes detected occurred in August-October, when the air temperature was high, rainfall was low, and the number of hotspot was also high. Hotspot occurrences in Muaro Jambi Regency during were higher in the peat lands area than in the non-peat lands. This proves that peat lands way more vulnerable to fire and need immediate yet serious action from stakeholders to prevent and deal with this country s annual issue.

7 5. ACKNOWNLEDGEMENT We would like to thank the Center for Environmental Research, Bogor Agricultural University (PPLH-IPB), Indonesia, for giving us opportunity to get the research funding. This research was funded by Indonesian Directorate General of Higher Education (DIKTI) for fiscal year REFERENCES Brondizio, E.S., Moran, E.F., Wu, Y., Land use change in the Amazon estuary: patterns of Caboclo settlement and landscape management. Hum. Ecol., 22 (3), pp Ceccato, P., Flasse, S., Tarantola, S., Jacquemond, S., and Gregoire, J.M., Detecting vegetation water content using reflectance in the optical domain. Remote Sensing of Environment, 77, pp Gao, B.C., NDWI - A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sensing of Environment, 58, pp Huete, A.R., Liu, H.Q., Batchily, K., and Van Leeuwen, W., A comparison of vegetation indices over a global set of TM images for EOS-MODIS. Remote Sensing of Environment, 59, pp Huete, A., Didan, K., Miura, T., Rodriguez, E.P., Gao, X., and Ferreira, L.G., Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment, 83, pp Huete, A.R., and Liu, H.Q., An error and sensitivity analysis of the atmospheric and soil correcting variants of the NDVI for the MODIS-EOS. IEEE Transactions on Geoscience and Remote Sensing, 32, pp Kachwala, T.S., Temporal monitoring of forest land for change detection and forest cover mapping through satellite remote sensing. In: Proceedings of the 6th Asian Conference on Remote Sensing. National Remote Sensing Agency, Hyderabad, pp Lu, X., Liu, R., and Liang, S., Removal of noise by wavelet method to generate high quality temporal data of terrestrial MODIS products. Photogrammetric Engineering and Remote Sensing, 73, pp Miettinen, J., and Liew, S.C., Status of peatland degradation and development in Sumatra and Kalimantan. Ambio, 39, pp NASA LP DAAC, MOD13Q1. USGS/Earth Resources Observation and Science (EROS) Center, Sioux Falls, South Dakota. Setiawan, Y., and Yoshino, K., Land use change detection by characterizing the vegetation dynamics: case study of Java Island, Indonesia. Journal of the Japan Society of Photogrammetry and Remote Sensing, 50(2), pp Setiawan, Y., and Yoshino, K., Detecting land use change from seasonal vegetation dynamics in regional scale with MODIS EVI 250 m time-series imagery. Journal of Land Use Science, 9 (3) pp Setiawan, Y., Yoshino, K., and Philpot, W. D Characterizing temporal vegetation dynamics of land use in regional scale of Java Island, Indonesia. Journal of Land Use Science, 8 (1) pp Xiao, X.M., Braswell, B., Zhang, Q.Y., Boles, S., and Frolking, S., Sensitivity of vegetation indices to atmospheric aerosols: continental-scale observations in northern Asia. Remote Sensing of Environment, 84, pp Statistics Indonesia [BPS], Muaro Jambi dalam Angka Tahun Statistics Indonesia of Muaro Jambi, Jambi. Statistics Indonesia [BPS], Muaro Jambi dalam Angka Tahun Statistics Indonesia of Muaro Jambi, Jambi. Yuan, F., Sawaya, K.E., Loeffelholz, B., Bauer, M.E., Land cover classification and change analysis of the Twin Cities (Minnesota) metropolitan area by multitemporal Landsat remote sensing. Rem. Sen. Envi., 98,