May 19 E108.5,N49.2 51346 38 1883 2054 1686 8 12
Mean value 11 13
The HJ-CCD burned area pixels at the same time and region were used as the actual HJ-IRS value to test the accuracy of fire pixel detection. The IRSCCD deviation was used to describe the actual accuracy of firedetectionusingHJ-IRS. Table 2 shows that the IRSCCD deviation ranges from 6% to 14%, with a mean value of 11%. The change is relatively stable, indicating that the improvedalgorithm based on HJ- IRS is highly accurate. The MODIS fire products at the same time and region were used to test the validity of the firedetection based on HJ-IRS. The IRSMODIS deviation was used to describe the referenced accuracy of firedetectionusingHJ-IRS. The IRSMODIS deviation ranges from 2% to 19%, with a mean value of 13%. The change is relatively large and unstable, and is attributed to the inherent uncertainties in the MODIS fire products. The 13% mean value of the IRSMODIS deviation also shows that the improvedalgorithm has high universality and portability. The number of HJ-IRS fire pixels, HJ- CCD burned area pixels, and MODIS fire pixels are shown in Fig. 3. The number of HJ-IRS fire pixels is 23.6 times that of the HJ-CCD fire burned area pixels, which is nearly 25 times that of the theoretically converted value according to the spatial resolution. The degree of similarity, which is defined as the percentage of the two values, is 94%. The number of fire pixels detected by HJ-IRS is 49.9 times that of the number of MODIS fire pixels, which is very close to 44.4 times of the theoretically converted value according to the spatial resolution; the degree of similarity is 89%. The statistical relationship (Fig. 3) further illustrates the feasibility and reliability of the HJ-IRS forestfiredetectionalgorithm.
Fire Information for Resource Management System (FIRMS) integrates remote sensing and GIS technologies to deliver global MODIS hotspot/active fire locations to natural resource managers and other stakeholders around the World. FIRMS was developed by the University of Maryland with funds from NASA. FIRMS is currently be- ing transitioned to an operational system at the United Nations Food and Agriculture Organization (UN FAO). FIRMS is primarily aimed at supporting natural resource managers, researchers, planners and policy makers by helping them understand when and where fires occur and delivering the fire information in near real-time and in easy-to-use formats (Christopher, Louis et al., 2006). Each hotspot/active fire location represents the centre of a 1km pixel (approximately) flagged as containing one or more actively burning hotspots/fires within that pixel. The hotspots/fires are detected usingdata from the MODIS (or Moderate Resolution Imaging Spectroradiometer) instrument, on board NASA’s Aqua and Terra satellites, using a specific firedetectionalgorithm that makes use of the thermal band detection characteristics of the sensor. Shapefiles, text files and kml files were downloaded from FIRMS which would contain the fire information of 24 past hours, 48 hours and last 7 days. Daily near- real time in various resolutions: 4 km, 2 km, 1 km, 500 m, and 250 m was also available (Zheng & Wan, 2007).
Classification is the task of learning a target function f that maps each attribute set x to one of the predefined class label y . The target function is also known as a classification model that can also be used to predict the class label of unknown records . The model classification used is a tree model and a rule-based model. A tree model has a flowchart-like tree structure, where each internal node (non-leaf node) indicates the test on an attribute, each branch represents the results of the test, and each leaf node is a class label . A rule-based model is a set of if-then conditions that have been derived from a tree model into more simple conditions. In this study, the tree and rule-based models implemented C5.0 algorithm. The C5.0 algorithm is generated based on decision trees . It can derive a set of if-then rules, shows the easier and more interpretable rules to understand. The C5.0 decision tree algorithm can be seen in Fig. 1 . The C5.0 algorithm was the development of C4.5 algorithm as an improved version and it has several advantages . The C5.0 trees and rule sets are smaller than the C4.5 counterparts. The trees are constructed using recursive manner (divide and conquer) from a set of training cases .
Handling uncertainty due to data aggregation and missing information requires space-time synthesis in rigorous formalism. Information granulation is at the heart of rough set theory. Rough set theory offers an attribute reduction algorithm and the dependency metric for feature selection . Meteorological data and images are parameters that change over space and time with relatively high frequency. The change of meteorological data could be recognized in hour scale, and the change of image data, taking into account only information connected to forest fires, in minute scale. Also for the forestfire prediction system, meteorological data history (archive values) is quite important. In order to monitor meteorological parameters and collect images in real time, the sensory network has to be established .The most critical issue in a forest ﬁre detection system is immediate response in order to minimize the scale of the disaster. This requires constant surveillance of the forest area. Current medium and large-scale ﬁre surveillance systems do not accomplish timely detection due to low resolution and long period of scan. Therefore, there is a need for a scalable solution that can provide real-time ﬁre detection with high accuracy. We believe that wireless sensor networks can potentially provide such solution. Recent advances in sensor networks support our belief that they make a promising framework for building near real time forest ﬁre detection systems. Currently, sensing modules can sense a variety of phenomena including temperature, relative humidity, and smoke which are all helpful for ﬁre detection systems .
Neural networks are composed of simple elements operating in parallel. We can train a neural network to perform a particular function by adjusting the values of the connections (weights) between elements. Commonly neural networks are adjusted, or trained, so that a particular input leads to a specific target output. Here, the network is adjusted, based on a comparison of the output and the target, until the network output matches the target value. Typically many such input/target pairs are essentially needed to train the network. The procedure which is used to perform the learning process is called a learning algorithm. The training phase is nothing, but the network is provided with enough examples called training sets. Each training set consists of a list of input values and the corresponding output. Training data sets are used by the network to learn the mapping from the input data to the output. The method of learning is quite different for each network as disclosed below. Kohonen Learning Network Classifiers
Forest fires are an important component of the savanna, tundra and boreal forest ecosystems. The increasing rate of the occurrence of fires however has increased the concern over their impacts on climate change and fragile ecosystems. This requires efficient and effective methods in forestfiredetection for near real-time monitoring so as to minimize these impacts. Remote sensing has been widely used in active forestfiredetection; however there are some limitations in contextual algorithms which are used in forestfiredetection. These contextual algorithms are greatly affected by clouds and different land cover types such as land and water with inherent temperatures included in the N x N matrix and this brings errors. As a step towards minimizing these problems an automated multi-temporal threshold algorithm was developed in this study using MSG satellite and ground firedata from Portugal. The algorithm is based on temperature anomalies detected in IR3.9 channel and the difference between IR3.9 and IR10.8 channels as well as the solar zenith angles for day, night and twilight conditions. Thresholds were set to determine actual fires and possible fires depending on how far the temperature of a particular point or pixel deviates from the normal background temperature which is estimated using the images directly prior to the actual image. The accuracy of the algorithm was compared with that of the MSG FIR-G product. The McNemar’s test was used for significance test of the difference between the multi-temporal threshold algorithm and the MSG FIR-G product which uses a contextual algorithm. This study shows that the multi-temporal threshold algorithm has higher firedetection rate (50%) as compared to MSG FIR-G (3.7%) when ground data from Portugal was used for validation. There is a significant difference between these methods (McNemar’s test ( 2
The aforementioned comparison implies that accurately mapping forest is very difficult [ 52 ], and there is still great uncertainty in the spatial extent, distribution, and subcategories of forest in Northeast China. Since 2008, the Landsat data freely released by the United States Geological Survey (USGS) has been the major data source for forest mapping with moderate resolution, but HJ-1 images also showed a reliable alternative or supplementary in forest mapping and subcategories identifying in this study. Similar free and open data policies would enable greater use of these data for public good and foster greater transparency of the development, implementation, and reactions to policy initiatives that affect the world’s forests [ 64 ]. In the future, how to integrate spectral data from Landsat-like images and structural data from radar observations (e.g., PALSAR and Sentinel-1) to improve the accuracy of forest maps should be considered a priority [ 65 ]. On the other hand, to reduce uncertainties in estimating the forest cover in large areas, fine-scale estimates are needed because the rapid land-cover changes related to forest can be observed at a small-scale. UAVs (unmanned aerial vehicles), LiDAR (light detection and ranging), and hyperspectral digital image data sources can provide accurate estimations of many key forest parameters. Results based on the integration of multi-source data have been shown to be superior to results obtained using a single data source [ 66 ], and they could provide important validation data for large scale estimation, but the costs of those data sources are higher.
ABSTRACT: This paper is about developing a php algorithm that sense real time data by a distributed WSN from the area under consideration and uses that information to define the exact location of fire & provides the shortest, safest path for affected people. This algorithm also sends information immediately to the nearest firefighter center & help them to reach the spot for rescue operation using shortest possible path. The area considered to test & validate the applicability of algorithm is Elshagra GAS depots in south Khartoum Sudan. This algorithm takes into account all the fired spots & predicts level of support needed based on intensity of fire. The algorithm is demonstrated on a scattered wireless sensor network test platform and in simulation. The results of simulations are presented in subsequent sections to demonstrate the applicability of algorithm.
coverage. Lloret et al  suggested deploying a mesh network of sensors provided with internet protocol (IP) cameras. Here the sensors detect the fire at the beginning and send an alarm signal to the sink. The sink then sends a message to switch the cameras ‘on’ in the same area of the detected fire to provide real images of the fire at any time. Hartung et al  used WSN for wood firedetection as a hybrid with web cameras. The main target of their studies was to investigate the fire behaviour in forests. They used WSN to provide data for weather status and web cameras to provide the images of the fire.Son  proposes a project for firedetection in South Korea using Cameras surveillance hybrid with WSN. They propose a clustered topology for the network. Each cluster has a head node to do some calculations, for example; fire risk level by measuring temperature, humidity and some other parameters. In addition there is routing and data aggregation tasks included in their algorithm. In this method there is an increase in the power consumption rate in each head nodes, besides they do not consider the power balancing issue, which may result in some sensors deactivating before others thus leading to coverage gaps. Hafeeda et al  presents a very smart system. They base their network action on fire weather index (FWI). This index includes the probability of fire ignition and fire spread rate as well. FWI provides the moisture content in relation to weather observation where the fuel code describes the soil content of forest ground. See Figure 1.
The main focus is the early smoke detection of wildfires in order to reduce the damaged area to a minimum. Large and high-intensity forest fires are widely uncontrollable and cause very high risk. To reduce false alarms, especially in hardly accessible terrain such as a forestfire in the mountains, a remote controlled UAV can fly to the place where a fire is assumed to confirm that the origin of the smoke is most likely a fire. Fig. 10 and Fig. 11 show gas sensor and temperature data measured during flight tests with the AirRobot drone as well as the alarm signal S gas,T . The detectionalgorithm and the alarm threshold were adapted to the situation.
The moderate resolution imaging spectroradiometer (MODIS) instruments launched in December 1999 and May 2002 on board the Terra and Aqua satellites respectively have provided opportunities for improvedfiredetection based on more thermal bands compared to inheritance satellite sensors such as national oceanic and atmospheric administration (NOAA) advanced very high resolution radiometer (AVHRR) and visible atmospheric sounder (VAS) on board geostationary orbiting environmental satellite (GOES) (Kaufman et al., 1998). Description of specification of MODIS instrument and its opportunity in detecting surface temperature anomaly was given by NASA scientists (Giglio et al., 2003). Many researchers strive to use MODIS capability to make better identification on fire pixels to give the more accurate and near real time information of fire occurancy (Zhanga et al., 2008) either active fire or smoldering. MODIS algorithms have been derived to identify fires (Justice et al., 2002; Wang et al., 2007) and there have been some improvements.
This node consists of microcontroller (ATMEGA 16), RF Modern, fire sensor, Accelerometer and DC supply. The controllers will be placed individually on the trees in a given required area which is highly prone deforestation, poachers or forest fires. Fire sensors will identify and send signal if any trace of forestfire is close to proximity. Accelerometer will determine any vibration on angular displacement of tree coordinate, which can be due to cutting of tree by poachers or landslide (Fig. 1).
It is observed that the relative performance of algorithm and main conclusion do not change. If p approaches to 0, it becomes easier to find guilty agents and algorithm performance converges. On the other hand, if p approaches 1, the relative differences among algorithms grow since more evidence is need to find an agent guilty. The algorithm presented implements a variety of data distribution strategies that can improve the distributor’s chances of identifying a leaker. It is shown that distributing objects judiciously can make a significant difference in identifying guilty agents, especially in cases where there is large overlap in the data that agents must receive.
Decision trees helps in the prediction of cause effects in health dataset, meteorological data set and pollution data set. They provide the target classes by divide and conquer method .Here an effort is made to identify different pollutants responsible for air pollution with J48 Machine learning algorithm.
Another alternative technology for detecting forest fires is the use of satellites and satellite images. Usually, satellites provide a complete image of the earth every 1–2 days. This long scan period, however, is not acceptable for detecting forest fires quickly. Additionally, the smallest fire size that can be detected by such a system is around 0.1 hectare, which also prevents firedetection just at the time when the fire starts, and fire localization error is about 1 km, which is not very accurate. Two main satellites launched for forestfiredetection purposes, the advanced very high resolution radiometer (AVHRR) , launched in 1998, and the moderate resolution imaging Spector radiometer (MODIS), launched in 1999 these satellites are used to detect the forestfire. Unfortunately, these satellites can provide images of the regions of the earth every two days and that is a long time for fire scanning; besides the quality of satellite images can be affected by weather conditions . Any existing satellite-based observations for forest fires suffer from severe limitations resulting in a failure in speedy and effective control for forest areas. Some of the limitations in an approach based on direct observation of forest fires from geostationary (GEO) or Low Earth Orbit (LEO) satellite are as follows: it might be impossible to provide a full satellite coverage or even intermittent coverage.
Forest Fires are one of the most important and prevalent type of disasters and they can create a great deal of Environmental Impacts due to which their early detection is very vital. The main need for choosing this particular application for the detection of forest fires is to overcome the demerits present in the existing technologies of MODIS and Basic Wireless Sensor Network -based ForestFireDetection Systems and an advanced system is developed for the detection of forest fires. The two main modules present in the project are the Monitoring Area Module and the Forest Area Module.The outcome of the above implementations reveal that various sensors used in addition to the temperature sensor improves security level for areas located near the forests. It also shows that the Optimized Solar Energy Harvester increases the efficiency to about 85 % and the use of PC based Web Server reduces the bulkiness and cost of the entire system.
transfer agent. Customers can route outgoing messages from their mail servers through the mail transfer agent for additional mail-filtering abilities; questionable e-mail can be held until approved by an administrator. Admin also can create policies to encrypt e-mail carrying sensitive information. This functionality is provided via Code Green's partnership with the Voltage Security Network, which offers e-mail encryption as a service. After connecting the device to network, A selected sources of data that the appliance should protect. It has built-in functionality to fingerprint both structured and unstructured data such as that in CIFS. Setup for CIFS was simply a matter of providing the server and share name, along with appropriate access credentials. The device then scans the share at user-defined intervals. CIFS scanning was trouble-free and didn't cause performance issues on our Windows file server.
ABSTRACT: Fire is one of the major disaster elements in the globe, especially it is caused in the remote or forest surroundings. It is trying to perform crisis scheduling for battling forest fires subject to constrained safeguard assets (that is, vehicles with fire motors), since dousing each fire point should consider different elements, for example, the genuine fire spreading speed, remove from fire motor warehouse to fire focuses, putting out fires speed of fire motors, and the quantity of dispatched vehicles. This system explores a bi-target protect vehicle scheduling issue for multi- point forest fires, which plans to ideally dispatch a predetermined number of fire motors to douse fires. The targets are to limit the aggregate fire stifling time and the quantity of dispatched fire motors. we propose a computerized mechanical framework where it consequently faculties an event of fire. The robot takes controlling segment independent from anyone else in light of the client predefined summon. By utilizing camera it will persistently observing If any fire mishap happen or not. Regardless of whether any fire mishap happens it will consequently enact a water splashing instrument. The framework gives a robotized fire checking and leeway framework.
In Finland VTT and FMI operate a forestfiredetection system using the directly retrieved data from VIS/IR radiometers on board European and US operational weather satellites and som additional scientific satellites. The forestfiredetection is operational since 1994 and provides good results during the forestfire seasons for fires large enough to be detectable by the senors. The amount of erroneously detected fires is in the range of 12 to 16 %, based on the feedback that is sent back to VTT/FMI by the affected authorities. The system is well established and the area observed by the system covers, beside Finland, Sweden, Norway, parts of Russia and most countries around the East Baltic Sea. The email messaging system, which distributes information on detected fires to the regional dispatching centers, sends messages to authorities in the neighboring countries. Among these countries MSB in Sweden receives emails on detected hot spots on Swedish ground or close to the borders. Currently there is no automated further distribution for such messages available.