Bima City, located in the eastern coast of the island of Sumbawa, West Nusa Tenggara (NTB), Indonesia was hit by huge flood in the late 2016 to early 2017.
From the rapid assessment of space-based dataset of Landsat OLI 8, we found that during 20 years (1997-2016) span, there are uncontrolled land use land cover transformation.
Area in the upper stream of DAS (Watershed) Sari of NTB completely transformed and lost many of its vegetation covers, while in the down stream, agricultural areas was transformed into man-made structures. Furthermore, river flood area was gone and changed into agriculture or man-made structures.
These transformations lead to huge volume of run-off when heavy rain event happened. The water will follow the physic’s rule, and flow to the lower elevation profile of land.
By the end of last month, flood was hit Garut City, where located in the center of Cimanuk watershed area. Many believed that flood event this time was due to mismanagement of upper stream area.
To have better understanding, we need to employ space-based image analysis to identify the most recent LULC (Land Use Land Cover) of Cimanuk Watershed.
Based-on Landsat OLI 8 acquired on August 10, 2016. The result are as shown in Figure 1, below
I used supervised neural netwok (NN) algorithm to extract the above figure. Based-on my evaluation, supervised NN is better than maximum likelihood or minimum distance algorithms. The class is defined in very smooth visualization.
The man-made structures (red color) are appeared in the upperstream, while agriculture (light green) and industrial forest (dark green) are dominant in this region. No doubt, when huge amount of rainfall drop, there will huge run-off be generated. This run-off will bring huge amount of materials.
Affected locations are highlighted in yellow (Sukakarya, Paminggir, Muarasanding, Pakuwon, Haurpanggung, and Sumantri). As we could see, this location is place when main streams ‘meet each other’. These streams loaded with huge amount of material, not only water but also suspended sediments (i.e., soil, mud, rock, logs, etc)
Multispectral data is very hard to be analyzed for high percentage of cloud cover. Fortunately, we have anothe space-based image of radar data of SENTINEL-1 to assess the same area. Data was acquired on October 4, 2016. The following is the result for man-made structures of upper stream of Cimanuk Watershed.
Sentinel-1 data enhanced the visualization, and we could confirmed, that there are extensive land use transformation in the upper stream.
The Sentinel-1 data has been calibrated and speckle filtering has been employed. Geometric correction also has been done based-on SRTM 3Sec data.
During the last 6 months I have been writing another scientific paper. It is about classification technique for mangrove forests assessment. And for the accuracy assessment, I am using not only data derived from field work but also data from Giri et al 2011 (Full paper available at http://onlinelibrary.wiley.com/doi/10.1111/j.1466-8238.2010.00584.x/abstract).
For everyone who need the data for another research activity, please do not hesitate to download from the following url
On of the important thing in land use land cover mapping is cloud cover assessment. Fortunately, free open source software for imagery processing such as GRASS provides us with the Automatic Cloud Cover Assessment (ACCA) module; i.landsat.acca.
i.landsat.acca implements the ACCA Algorithm from Irish (2000) with the constant values for pass filter one from Irish et al. (2006). To do this, it needs Landsat band numbers 2, 3, 4, 5, and 6 (or band 61 for Landsat 7 ETM+) which have already been processed from DN into reflectance and band-6 temperature with i.landsat.toar.
In this case we will assess the cloud cover over Surabaya City, East Java Province, Indonesia. Data used in this cloud assessment is Landsat 7 ETM+, it was acquired on December 27, 2013. Figure 1 shows the study area with the cloud cover in the northern part, and some part with the cloud shadow and haze. We will remove this cloud cover, by employ the i.landsat.acc module in GRASS.
Note: I always working with command console (Figure 2) instead of GUI, it is more convenient, so please learn how to get used with it.
This image shows the “flame” on the coastline environment of Banda Aceh, Nanggroe Aceh Darussalam Province, Indonesia.
In fact, it should be sedimentation process captured by Landsat 8 OLI sensor on October 11, 2013. As we can see from the image, there are two main streams empties in the Strait of Bengal.
There are many variables involve to create this “flame”, such as wind speed, pressure, sediments source, and bathymetry. To produce the image above, I employed MNF transformation and used 765 MNF band output for RGB color composite.
Another unique phenomenon
There is another interesting earth surface phenomenon on the northern part of the image. Large size of wave!
At first I thought this was related to an earthquake events, because in the northwestern part of the image there is a plate boundary (ridge in Andaman Sea). Then I was looking at the USGS earthquake database for an earthquake event with scale ranging from Mw 1, in accordance with the date of the acquisition, however I could not find any related earthquake event.
Indonesia is situated in the “Pacific Ring of Fire”, which is surrounded by volcanoes and vulnerable from volcanic eruptions. Since the population development, the volcanic environment then occupied by settlements or agricultural fields.
It is important to provide the peoples and decision makers the hazardous map or rapid assessment map for new safety zone.
With the availability of free open source software GRASS and free download-able DEM (e.g. ASTER GDEM), it is possible to generate new base-line of hazardous map.
In this map, we assume that the main river will provides the new base-line for new safety zone of volcanic eruptions. Study site Mount Sinabung, Karo Regency, North Sumatera Indonesia.
Originated from Pleistocene-to-Holocene geological ages, this stratovolcano located 40 km from Lake Toba, which was “born” from Toba Supervolcano. Recently, Mount Sinabung was erupted in January and February 2014. The other eruptions were September and November 2013, 29 August 2010, the year 1912, and the year 1600.
The main rivers was generated using Horton Stream Order. Further analysis of new safety zone should begin from this base-line. Our assumption is when the debris flow from the summit, it will flow through the streams and follow the topographic condition.
The main streams is indicated with the highest number (No. 4), while the other streams are No. 1, 2, and 3. The main streams is situated surrounding the Mount Sinabung with radius around 3 km from the summit in the north side, to about 5 km from the summit in the southern part. Combination of radial and annular pattern of streams were identified in the Mount Sinabung environment.
This method can provide zonation with minimum data requirements. However, further analysis should be conducted to provide the peoples with the integrated zonation of hazardous map such as susceptibility mapping of debris flows and other gravitational hazards (e.g. Landslide)