International Journal of Remote Sensing and Earth Sciences (IJReSES)
Not a member yet
334 research outputs found
Sort by
COMPARISON OF MACHINE LEARNING MODELS FOR LAND COVER CLASSIFICATION
Full text linkLand cover data remain one of crucial information for public use. Â With rapid human-associated land alteration, this information needs to be frequently updated. Remotely-sensed data provide the best option to construct land cover maps with numerous methods available in the literature. While disagreement exists to select the robust one, further exploration should be made to extend the understanding on the behavior of machine learners, in particular, for classification problems. This article discusses performance of pixel-based machine learning algorithms, frequently used in research or implementation. Five popular algorithms were evaluated to distinguish five rural land cover classes, i.e. built-ups, crops, mixed garden, oil palm plantations and rubber estates, from Sentinel-2 data. This research found that the benchmark, classification and regression tree, was unable to differentiate woody vegetation, although the overall accuracy was sufficiently moderate. This suggested that overall accuracy cannot be seen as the only measure for assessing the quality of the thematic output. Meanwhile, support vector machines and random forest competed to yield the highest accuracy and class detection capability, although the latter was in favor with 98% accuracy level. A newly developed model, like extreme gradient boosting, achieved a similar level of accuracy. This research implies that modern machine learning approaches would be invaluable for land cover classification; hence, access to these modeling toolkits is substantial
BIOMASS ESTIMATION MODEL AND CARBON DIOXIDE SEQUESTRATION FOR MANGROVE FOREST USING SENTINEL-2 IN BENOA BAY, BALI
Full text linkRemote sensing technology can be used to find out the potential of mangrove forests information. One of the potentials is to be able to absorb three times more CO2 than other forests. CO2 absorbed during the photosynthesis process, produces organic compounds that are stored in the mangrove forest biomass. Utilization of remote sensing technology is able to detect mangrove forest biomass using the density level of the vegetation index. This study focuses on determining the best AGB model based on the vegetation index and the ability of mangrove forests to absorb CO2. This research was conducted in Benoa Bay, Bali Province, Indonesia. The satellite image used is Sentinel-2. Classification of mangroves and non-mangroves using a multivariate random forest algorithm. Furthermore, the mangrove forest biomass model using a semi-empirical approach, while the estimation of CO2 sequestration using allometric equations. Mean Absolute Error (MAE) is used to evaluate the validation of the model results. The classification results showed that the detected area of Benoa Bay mangrove forest reached 1134 ha (OA: 0.98, kappa: 0.95). The best AGB estimation result is the DVI-based AGB model (MAE: 23,525) with a value range of 0 to 468.38 Mg/ha. DVI-based AGB derivatives are BGB with a value range of 0 to 79.425 Mg/ha, TAB with a value range of 0 to 547.8 Mg/ha, TCS with a value range of 0 to 257.47 Mg/ha, and ACS with a value range of 0 to 944.912 Mg/ha
HAIL IDENTIFICATION BASED ON WEATHER FACTOR ANALYSIS AND HIMAWARI 8 SATELLITE IMAGERY (CASE STUDY OF HAIL ON 2ND MARCH 2021 IN MALANG INDONESIA)
Full text linkA hail phenomenon occurred in Malang, Sumbermanjing Wetan District (8°6’S and 112°24’E) on March 2, 2021. According to the Regional National Disaster Management Agency, it was accompanied by heavy rain and strong winds, which caused several trees to fall, resulting in damage to people's houses (BNPBD, 2021). Hail is precipitation in the form of ice, usually an irregular round shape produced by cumulonimbus convective clouds (AMS, 2019). The research was conducted by examining global, regional, and local weather factors and analysing the cloud characteristics from satellite image data during hail events. Based on the analysis, it was found that ENSO, sea surface temperature anomalies, and MJO had no effect on the incidence of the hail. The streamline map showed the presence of shearlines and tropical cyclones around the Malang area, and the temperature significantly decrease from 07.00 UTC to 08.00 UTC of 4.4°C and from 08.00 UTC to 09.00 UTC of 3.6°C with significant increase in humidity from 07.00 UTC to 08.00 UTC of 10%. The cloud top temperature was analysed to be at the ripe stage at 07.40 UTC and 8.40 UTC, at -68.2°C
AN ENHANCEMENT TO THE QUANTITATIVE PRECIPITATION ESTIMATION USING RADAR-GAUGE MERGING
Full text linkQuantitative Precipitation Estimation (QPE) is quite important information for the hydrology fields and has many advantages for many purposes. Its dense spatial and temporal resolution can be combined with the surface observation to enhance the accuracy of the estimation. This paper presents an enhancement to the QPE product from BMKG weather radar network at Surabaya by adjusting the estimation value form radar to the real data observation from rain gauge. A total of 58 rain gauge is used. The Mean Field Bias (MFB) method used to determine the correction factor through the difference between radar estimation and rain gauge observation value. The correction factor obtained at each gauge points are interpolated to the entire radar grid in a multiplicative adjustment. Radar-gauge merging results a significant improvement revealed by the decreasing of mean absolute error (MAE) about 40% and false alarm ratio (FAR) as well an increasing of possibility of detection (POD) more than 50% at any rain categories (light rain, moderate rain, heavy rain, and very heavy rain). This performance improvement is very beneficial for operational used in BMKG and other hydrological needs
MONITORING MODEL OF LAND COVER CHANGE FOR THE INDICATION OF DEVEGETATION AND REVEGETATION USING SENTINEL-2
Full text linkIInformation on land cover change is very important for various purposes, including the monitoring of changes for environmental sustainability. The objective of this study is to create a monitoring model of land cover change for the indication of devegetation and revegetation usingdata fromSentinel-2 from 2017 to 2018 of the Brantas watershed.This is one of the priority watersheds in Indonesia, so it is necessary to observe changes in its environment, including land cover change. Such change can be detected using remote sensing data. The method used is a hybrid between Normalized Difference Vegetation Index(NDVI) and Normalized Burn Ratio (NBR) which aims to detect land changes with a focus on devegetationand revegetation by determining the threshold value for vegetation index (ΔNDVI) and open land index (ΔNBR).The study found that the best thresholds to detect revegetation were ΔNDVI > 0.0309 and ΔNBR 0.0314.It is concluded that Sentinel-2 data can be used to monitor land changes indicating devegetation and revegetation with established NDVI and NBR threshold conditions
SHORELINE CHANGES AFTER THE SUNDA STRAIT TSUNAMI ON THE COAST OF PANDEGLANG REGENCY, BANTEN
Full text linkThe Sunda Strait tsunami occurred on the coast of west Banten and South Lampung at 22nd December 2018, resulting in 437 deaths, with10 victims missing. The disaster had various impacts on the environment and ecosystem, with this area suffering the greatest effects from the disaster. The utilisation of remote sensing technology enables the monitoring of coastal areas in an effective and low-cost manner. Shoreline extraction using the Google Earth Engine, which is an open-source platform that facilitates the processing of a large number of data quickly. This study used Landsat-8 Surface Reflectance Tier 1 data that was geometrically and radiometrically corrected, with processing using the Modification of Normalized Difference Water Index (MNDWI) algorithm. The results show that 30.1% of the coastline in Pandeglang Regency occurred suffered abrasion, 20.2% suffered accretion,while 40.7% saw no change. The maximum abrasion of 130.2 meters occurred in the village of Tanjung Jaya. Moreover, the maximum shoreline accretion was 43.3 meters in the village of Panimbang Jaya. The average shorelinechange in Pandeglang Regencywas 3.9 meters
HYDRO-METEOROLOGICAL ASPECTS OF THE 2021 SOUTH KALIMANTAN FLOOD: TOPOGRAPHY, TIDES, AND PRECIPITATION
Full text linkThe 2021 South Kalimantan flood was recorded as the most serious ever to have taken place in the province. It occurred due to high-intensity rain during the period 10-19 January, accompanied by a spring tide. This study provides an overview of the disaster, with reference to the hydro-meteorological conditions (topography, tides, and precipitation). The method used was the analysis of the precipitation and its monthly rainfall pattern anomalies using remote sensing data. A Digital Elevation Model (DEM) was also analyzed to indicate the most noticeably flood-affected area. In certain areas, total precipitation during the ten days reached 672.8 mm, with daily precipitation peaking at 255 mm on January 14, greater than the 25-year return period value. The flood coincided with a spring tide, which peaked at 1.21 m on the evening of January 15. Using 20- year GPM data, it was found that ENSO and IOD coexisted with both the highest and lowest anomalies. With a La Niña event at the end of 2020,  a positive precipitation anomaly in 2021 was expected. The extreme precipitation is suspected to be the main driver of the  2021 South Kalimantan flood, whose impact was worsened by the spring tides. This study conducts further research on the correlation between land-use change, rainfall, spring tide and flooding in South Kalimantan. In addition, it is recommended that the government plan flood risk management by prioritizing areas based on vulnerability to climate hazards
ESTIMATION OF ABOVEGROUND CARBON STOCK USING SAR SENTINEL-1 IMAGERY IN SAMARINDA CITY
Full text linkEstimation of aboveground carbon stock on stands vegetation, especially in green open space, has become an urgent issue in the effort to calculate, monitor, manage, and evaluate carbon stocks, especially in a massive urban area such as Samarinda City, Kalimantan Timur Province, Indonesia. The use of Sentinel-1 imagery was maximised to accommodate the weaknesses in its optical imagery, and combined with its ability to produce cloud-free imagery and minimal atmospheric influence. The study aims to test the accuracy of the estimated model of above-ground carbon stocks, to ascertain the total carbon stock, and to map the spatial distribution of carbon stocks on stands vegetation in Samarinda City. The methods used included empirical modelling of carbon stocks and statistical analysis comparing backscatter values and actual carbon stocks in the field using VV and VH polarisation. Model accuracy tests were performed using the standard error of estimate in independent accuracy test samples. The results show that Samarinda Utara subdistrict had the highest carbon stock of 3,765,255.9 tons in the VH exponential model. Total carbon stocks in the exponential VH models were 6,489,478.1 tons, with the highest maximum accuracy of 87.6 %, and an estimated error of 0.57 tons/pixel
ANALYSIS OF POTENTIAL FISHING ZONES IN COASTAL WATERS: A CASE STUDY OF NIAS ISLAND WATERS
Full text linkThe need for information on potential fishing zones based on remote sensing satellite data (ZPPI) in coastal waters is increasing. This study aims to create an information model of such zones in coastal waters (coastal ZPPI). The image data used include GHRSST, SNPP-VIIRS and MODIS-Aqua images acquired from September 1st-30th, 2018 and September 1st-30th, 2019, together with other supporting data. The coastal ZPPI information is based on the results of thermal front SST detection and overlaying this with chlorophyll-a. The method of determining the thermal front sea surface temperature (SST) used Single Image Edge Detection (SIED). The chlorophyll-a range used was in the mesotropic area (0.2-0.5 mg/m3). Coastal ZPPI coordinates were determined using the polygon centre of mass, while the coastal ZPPI information generated was only for coastal areas with a radius of between 4-12 nautical miles and was divided into two criteria, namely High Potential (HP) and Low Potential (LP). The results show that the coastal ZPPI models were suitable to determine fishing locations around Nias Island. The percentage of coastal ZPPI information generated was around 90% information monthly. In September 2018, 27 days of information were produced, consisting of 11 HP sets of coastal ZPPI information and 16 sets of LP information, while in September 2019 it was possible to produce 29 days of such information, comprising 11 sets of HP coastal ZPPI information and 18 LP sets. The use of SST parameters of GHRSST images and the addition of chlorophyll-a parameters to MODIS-Aqua images are very effective and efficient ways of supporting the provision of coastal ZPPI information in the waters of Nias Island and its surroundings