scholarly journals Global Wildfire Outlook Forecast with Neural Networks

2020 ◽  
Vol 12 (14) ◽  
pp. 2246
Author(s):  
Yongjia Song ◽  
Yuhang Wang
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
South America ◽  
Land Cover ◽  
Southern Africa ◽  
Burned Area ◽  
Good Prediction ◽  
Wildfire Occurrence ◽  
Fire Emissions ◽  
Burned Areas

Wildfire occurrence and spread are affected by atmospheric and land-cover conditions, and therefore meteorological and land-cover parameters can be used in area burned prediction. We apply three forecast methods, a generalized linear model, regression trees, and neural networks (Levenberg–Marquardt backpropagation) to produce monthly wildfire predictions 1 year in advance. The models are trained using the Global Fire Emissions Database version 4 with small fires (GFEDv4s). Continuous 1-year monthly fire predictions from 2011 to 2015 are evaluated with GFEDs data for 10 major fire regions around the globe. The predictions by the neural network method are superior. The 1-year moving predictions have good prediction skills over these regions, especially over the tropics and the southern hemisphere. The temporal refined index of agreement (IOA) between predictions and GFEDv4s regional burned areas are 0.82, 0.82, 0.8, 0.75, and 0.56 for northern and southern Africa, South America, equatorial Asia and Australia, respectively. The spatial refined IOA for 5-year averaged monthly burned area range from 0.69 in low-fire months to 0.86 in high-fire months over South America, 0.3–0.93 over northern Africa, 0.69–0.93 over southern Africa, 0.47–0.85 over equatorial Asia, and 0.53–0.8 over Australia. For fire regions in the northern temperate and boreal regions, the temporal and spatial IOA between predictions and GFEDv4s data in fire seasons are 0.7–0.79 and 0.24–0.83, respectively. The predictions in high-fire months are better than low-fire months. This study illustrates the feasibility of global fire activity outlook forecasts using a neural network model and the method can be applied to quickly assess the potential effects of climate change on wildfires.

scholarly journals Estimates of biomass burning emissions in tropical Asia based on satellite-derived data

2010 ◽  
Vol 10 (5) ◽  
pp. 2335-2351 ◽  
Author(s):  
D. Chang ◽  
Y. Song
Keyword(s):  
Biomass Burning ◽  
Atmospheric Chemistry ◽  
Crop Residue ◽  
Fire Season ◽  
Burned Area ◽  
Published Data ◽  
Tropical Asia ◽  
Fire Emissions ◽  
Burned Areas ◽  
In Fire

Abstract. Biomass burning in tropical Asia emits large amounts of trace gases and particulate matter into the atmosphere, which has significant implications for atmospheric chemistry and climatic change. In this study, emissions from open biomass burning over tropical Asia were evaluated during seven fire years from 2000 to 2006 (1 March 2000–31 February 2007). The size of the burned areas was estimated from newly published 1-km L3JRC and 500-m MODIS burned area products (MCD45A1). Available fuel loads and emission factors were assigned to each vegetation type in a GlobCover characterisation map, and fuel moisture content was taken into account when calculating combustion factors. Over the whole period, both burned areas and fire emissions showed clear spatial and seasonal variations. The size of the L3JRC burned areas ranged from 36 031 km2 in fire year 2005 to 52 303 km2 in 2001, and the MCD45A1 burned areas ranged from 54 790 km2 in fire year 2001 to 148 967 km2 in 2004. Comparisons of L3JRC and MCD45A1 burned areas using ground-based measurements and other satellite data were made in several major burning regions, and the results suggest that MCD45A1 generally performed better than L3JRC, although with a certain degree of underestimation in forest areas. The average annual L3JRC-based emissions were 123 (102–152), 12 (9–15), 1.0 (0.7–1.3), 1.9 (1.4–2.6), 0.11 (0.09–0.12), 0.89 (0.63–1.21), 0.043 (0.036–0.053), 0.021 (0.021–0.023), 0.41 (0.34–0.52), 3.4 (2.6–4.3), and 3.6 (2.8–4.7) Tg yr−1 for CO2, CO, CH4, NMHCs, NOx, NH3, SO2, BC, OC, PM2.5, and PM10, respectively, whereas MCD45A1-based emissions were 122 (108–144), 9.3 (7.7–11.7), 0.63 (0.46–0.86), 1.1 (0.8–1.6), 0.11 (0.10–0.13), 0.54 (0.38–0.76), 0.043 (0.038–0.051), 0.033 (0.032–0.037), 0.39 (0.34–0.47), 3.0 (2.6–3.7), and 3.3 (2.8–4.0) Tg yr−1. Forest burning was identified as the major source of the fire emissions due to its high carbon density. Although agricultural burning was the second highest contributor, it is possible that some crop residue combustion was missed by satellite observations. This possibility is supported by comparisons with previously published data, and this result may be due to the small size of the field crop residue burning. Fire emissions were mainly concentrated in Indonesia, India, Myanmar, and Cambodia. Furthermore, the peak in the size of the burned area was generally found in the early fire season, whereas the maximum fire emissions often occurred in the late fire season.

scholarly journals Uncertainty-Based Human-in-the-Loop Deep Learning for Land Cover Segmentation

2020 ◽  
Vol 12 (22) ◽  
pp. 3836
Author(s):  
Carlos García Rodríguez ◽  
Jordi Vitrià ◽  
Oscar Mora
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Deep Learning ◽  
Land Cover ◽  
Satellite Images ◽  
State Of The Art ◽  
Low Cost ◽  
Human Intervention ◽  
Human In The Loop ◽  
Learning Techniques

In recent years, different deep learning techniques were applied to segment aerial and satellite images. Nevertheless, state of the art techniques for land cover segmentation does not provide accurate results to be used in real applications. This is a problem faced by institutions and companies that want to replace time-consuming and exhausting human work with AI technology. In this work, we propose a method that combines deep learning with a human-in-the-loop strategy to achieve expert-level results at a low cost. We use a neural network to segment the images. In parallel, another network is used to measure uncertainty for predicted pixels. Finally, we combine these neural networks with a human-in-the-loop approach to produce correct predictions as if developed by human photointerpreters. Applying this methodology shows that we can increase the accuracy of land cover segmentation tasks while decreasing human intervention.

scholarly journals MULTI-TEMPORAL LAND COVER CLASSIFICATION WITH LONG SHORT-TERM MEMORY NEURAL NETWORKS

2017 ◽  
Vol XLII-1/W1 ◽  
pp. 551-558 ◽  
Author(s):  
M. Rußwurm ◽  
M. Körner
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Land Cover ◽  
Short Term Memory ◽  
Land Cover Classification ◽  
Support Vector ◽  
Short Term ◽  
Term Memory ◽  
Classification Techniques ◽  
Long Short Term Memory

<i>Land cover classification (LCC)</i> is a central and wide field of research in earth observation and has already put forth a variety of classification techniques. Many approaches are based on classification techniques considering observation at certain points in time. However, some land cover classes, such as crops, change their spectral characteristics due to environmental influences and can thus not be monitored effectively with classical mono-temporal approaches. Nevertheless, these temporal observations should be utilized to benefit the classification process. After extensive research has been conducted on modeling temporal dynamics by spectro-temporal profiles using vegetation indices, we propose a deep learning approach to utilize these temporal characteristics for classification tasks. In this work, we show how <i>long short-term memory</i> (LSTM) neural networks can be employed for crop identification purposes with SENTINEL 2A observations from large study areas and label information provided by local authorities. We compare these temporal neural network models, <i>i.e.</i>, LSTM and <i>recurrent neural network</i> (RNN), with a classical non-temporal <i>convolutional neural network</i> (CNN) model and an additional <i>support vector machine</i> (SVM) baseline. With our rather straightforward LSTM variant, we exceeded state-of-the-art classification performance, thus opening promising potential for further research.

scholarly journals SEMI-SUPERVISED SEMANTIC SEGMENTATION NETWORK VIA LEARNING CONSISTENCY FOR REMOTE SENSING LAND-COVER CLASSIFICATION

2020 ◽  
Vol V-2-2020 ◽  
pp. 609-615
Author(s):  
B. Zhang ◽  
Y. Zhang ◽  
Y. Li ◽  
Y. Wan ◽  
F. Wen
Keyword(s):  
Neural Network ◽  
Remote Sensing ◽  
Neural Networks ◽  
Land Cover ◽  
Deep Neural Networks ◽  
Semantic Segmentation ◽  
Land Cover Classification ◽  
Unlabeled Data ◽  
Weighted Sum ◽  
Affine Transform

Abstract. Current popular deep neural networks for semantic segmentation are almost supervised and highly rely on a large amount of labeled data. However, obtaining a large amount of pixel-level labeled data is time-consuming and laborious. In remote sensing area, this problem is more urgent. To alleviate this problem, we propose a novel semantic segmentation neural network (S4Net) based on semi-supervised learning by using unlabeled data. Our model can learn from unlabeled data by consistency regularization, which enforces the consistency of output under different random transforms and perturbations, such as random affine transform. Thus, the network is trained by the weighted sum of a supervised loss from labeled data and a consistency regularization loss from unlabeled data. The experiments we conducted on DeepGlobe land cover classification challenge dataset verified that our network can make use of unlabeled data to obtain precise results of semantic segmentation and achieve competitive performance when compared to other methods.

Investigation of Applying Physics Informed Neural Networks (PINN) and Variants on 2D Aerodynamics Problems

2020 ◽  
Author(s):  
Wee-Beng Tay ◽  
Murali Damodaran ◽  
Zhi-Da Teh ◽  
Rahul Halder
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Network ◽  
Training Data ◽  
Batch Size ◽  
Test Case ◽  
Good Prediction ◽  
Time Varying ◽  
Artificial Neural ◽  
Artificial Neural Network Ann

Abstract Investigation of applying physics informed neural networks on the test case involving flow past Converging-Diverging (CD) Nozzle has been investigated. Both Artificial Neural Network (ANN) and Physics Informed Neural Network (PINN) are used to do the training and prediction. Results show that Artificial Neural Network (ANN) by itself is already able to give relatively good prediction. With the addition of PINN, the error reduces even more, although by only a relatively small amount. This is perhaps due to the already good prediction. The effects of batch size, training iteration and number of epochs on the prediction accuracy have already been tested. It is found that increasing batch size improves the prediction. On the other hand, increasing the training iteration may give poorer prediction due to overfitting. Lastly, in general, increasing epochs reduces the error. More investigations should be done in the future to further reduce the error while at the same time using less training data. More complicated cases with time varying results should also be included. Extrapolation of the results using PINN can also be tested.

Application of Artificial Neural Networks to Predict Cotton Production: A Case Study in Diyarbakır Province, Turkey

2020 ◽  
Author(s):  
Nazire Mikail ◽  
Mehmet Fırat BARAN
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Networks ◽  
Plant Production ◽  
Good Prediction ◽  
Ann Model ◽  
Cotton Production ◽  
Factors Affecting ◽  
Artificial Neural Network Method ◽  
Artificial Neural

Abstract Cultivators are always curious about the factors affecting yield in plant production. Determining these factors can provide information about the yield in the future. The reliability of information is dependent on a good prediction model. According to the operating process, artificial neural networks imitate the neural network in humans. The ability to make predictions for the current situation by combining the information people have gained from different experiences is designed in artificial neural networks. Therefore, in complex problems, it gives better results than artificial neural networks.In this study, we used an artificial neural network method to model the production of cotton. From a comprehensive datum collection spanning 73 farms in Diyarbakır, Turkey, the mean cotton production was 559.19 kg da-1. There are four factors that are selected as pivotal inputs into this model. As a result, the ultimate ANN model is able to forshow cotton production, which is built on elements such as farm states (cotton area and irrigation periodicity), machinery usage and fertilizer consumption.At the end of the study, cotton yield was estimated with 84% accuracy.

scholarly journals About Validation-Comparison of Burned Area Products

2020 ◽  
Vol 12 (23) ◽  
pp. 3972
Author(s):  
Germán M. Valencia ◽  
Jesús A. Anaya ◽  
Éver A. Velásquez ◽  
Rubén Ramo ◽  
Francisco J. Caro-Lopera
Keyword(s):  
South America ◽  
Land Cover ◽  
Northern Hemisphere ◽  
Random Matrix Theory ◽  
Random Matrix ◽  
Matrix Theory ◽  
Permutation Tests ◽  
Comparison Method ◽  
Burned Area ◽  
Tropical Regions

This paper proposes a validation-comparison method for burned area (BA) products. The technique considers: (1) bootstrapping of scenes for validation-comparison and (2) permutation tests for validation. The research focuses on the tropical regions of Northern Hemisphere South America and Northern Hemisphere Africa and studies the accuracy of the BA products: MCD45, MCD64C5.1, MCD64C6, Fire CCI C4.1, and Fire CCI C5.0. The first and second parts consider methods based on random matrix theory for zone differentiation and multiple ancillary variables such as BA, the number of burned fragments, ecosystem type, land cover, and burned biomass. The first method studies the zone effect using bootstrapping of Riemannian, full Procrustes, and partial Procrustes distances. The second method explores the validation by using distance permutation tests under uncertainty. The results refer to Fire CCI 5.0 with the best BA description, followed by MCD64C6, MCD64C5.1, MCD45, and Fire CCI 4.1. It was also found that biomass, total BA, and the number of fragments affect the BA product accuracy.

scholarly journals Global fire emissions and the contribution of deforestation, savanna, forest, agricultural, and peat fires (1997–2009)

2010 ◽  
Vol 10 (23) ◽  
pp. 11707-11735 ◽  
Author(s):  
G. R. van der Werf ◽  
J. T. Randerson ◽  
L. Giglio ◽  
G. J. Collatz ◽  
M. Mu ◽  
...  
Keyword(s):  
South America ◽  
Carbon Emissions ◽  
Agricultural Waste ◽  
Trace Gases ◽  
Plant Productivity ◽  
Tropical Deforestation ◽  
Burned Area ◽  
Fire Emissions ◽  
Active Fire ◽  
Peat Fires

Abstract. New burned area datasets and top-down constraints from atmospheric concentration measurements of pyrogenic gases have decreased the large uncertainty in fire emissions estimates. However, significant gaps remain in our understanding of the contribution of deforestation, savanna, forest, agricultural waste, and peat fires to total global fire emissions. Here we used a revised version of the Carnegie-Ames-Stanford-Approach (CASA) biogeochemical model and improved satellite-derived estimates of area burned, fire activity, and plant productivity to calculate fire emissions for the 1997–2009 period on a 0.5° spatial resolution with a monthly time step. For November 2000 onwards, estimates were based on burned area, active fire detections, and plant productivity from the MODerate resolution Imaging Spectroradiometer (MODIS) sensor. For the partitioning we focused on the MODIS era. We used maps of burned area derived from the Tropical Rainfall Measuring Mission (TRMM) Visible and Infrared Scanner (VIRS) and Along-Track Scanning Radiometer (ATSR) active fire data prior to MODIS (1997–2000) and estimates of plant productivity derived from Advanced Very High Resolution Radiometer (AVHRR) observations during the same period. Average global fire carbon emissions according to this version 3 of the Global Fire Emissions Database (GFED3) were 2.0 Pg C year−1 with significant interannual variability during 1997–2001 (2.8 Pg C year−1 in 1998 and 1.6 Pg C year−1 in 2001). Globally, emissions during 2002–2007 were relatively constant (around 2.1 Pg C year−1) before declining in 2008 (1.7 Pg C year−1) and 2009 (1.5 Pg C year−1) partly due to lower deforestation fire emissions in South America and tropical Asia. On a regional basis, emissions were highly variable during 2002–2007 (e.g., boreal Asia, South America, and Indonesia), but these regional differences canceled out at a global level. During the MODIS era (2001–2009), most carbon emissions were from fires in grasslands and savannas (44%) with smaller contributions from tropical deforestation and degradation fires (20%), woodland fires (mostly confined to the tropics, 16%), forest fires (mostly in the extratropics, 15%), agricultural waste burning (3%), and tropical peat fires (3%). The contribution from agricultural waste fires was likely a lower bound because our approach for measuring burned area could not detect all of these relatively small fires. Total carbon emissions were on average 13% lower than in our previous (GFED2) work. For reduced trace gases such as CO and CH4, deforestation, degradation, and peat fires were more important contributors because of higher emissions of reduced trace gases per unit carbon combusted compared to savanna fires. Carbon emissions from tropical deforestation, degradation, and peatland fires were on average 0.5 Pg C year−1. The carbon emissions from these fires may not be balanced by regrowth following fire. Our results provide the first global assessment of the contribution of different sources to total global fire emissions for the past decade, and supply the community with an improved 13-year fire emissions time series.

scholarly journals Assessing the wildfire activity in protected areas of the Amazon basin

2021 ◽  
Author(s):  
Emmanuel Da Ponte ◽  
Fermin Alcasena ◽  
Tejas Bhagwat ◽  
Zhongyang Hu
Keyword(s):  
Land Use ◽  
Land Cover ◽  
Protected Areas ◽  
Amazon Basin ◽  
Burned Area ◽  
Land Use Land Cover ◽  
Fire Control ◽  
Burned Areas ◽  
Burn Probability ◽  
Fire Ignition

&lt;p&gt;Despite &amp;#160;growing concerns regarding the Amazonian wildfires, the magnitude of the problem is poorly understood. In this study, we assessed the wildfire activity in the &amp;#160;protected natural sites (n= 428) of Bolivia, Brazil, Colombia, Ecuador, French Guyana, Guyana, Peru, Suriname, and Venezuela, encompassing an area of 1.4 million km&lt;sup&gt;2 &lt;/sup&gt;of the Amazon basin. A 250 m resolution spectroradiometer sensor imaging (MODIS) was used to obtain land-use/land-cover (MODIS land use land cover product) changes and derive the wildfire activity data (ignition locations and burned areas (MODIS active fire products)) from 2001 to 2018. First, we characterized the mean fire return interval, wildfire occurrence, and empiric burn probability. Then, we implemented a transmission analysis to assess the burned area from incoming fires. We used transmission analysis to characterize the land use and anthropic activities associated to fire ignition locations across the different countries. On average, 867 km &lt;sup&gt;2&lt;/sup&gt; of natural forests were burned in protected natural sites annually, and about 85 incoming fires per year from neighboring areas accounted for 10.5% (9,128 ha) of the burned area. The most affected countries were Brazil (53%), Bolivia (24%), and Venezuela (16%).Considerable amount of fire ignition points were detected in open savannas (29%) and grasslands (41%) , where the fire is periodically used to clear extensive grazing properties. The incoming fires from savannas were responsible for burning the largest forest areas within protected sites, affecting as much as 9,800 ha in a single fire event. In conclusion, we &amp;#160;discuss the potential implications of the main socioeconomic factors and environmental policies that could explain increasing trends of burned areas. Wildfire risk mitigation strategies include the fire ignition prevention in developed areas, fire use regulation in rural communities, increased fuels management efforts in the buffer areas surrounding natural sites, and the early detection system that may facilitate a rapid and effective fire control response. Our analysis and quantitative outcomes describing the fire activity represent a sound science-based approach for an well defined wildfire management within the protected areas of the Amazonian basin.&lt;/p&gt;

scholarly journals Prediction of Air pollution in Al-Hmadi City using Artificial Neural Network (ANN)

2020 ◽  
Vol 8 (4) ◽  
pp. 1390-1399
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Good Prediction ◽  
Air Velocity ◽  
Air Quality Prediction ◽  
Good Prediction Performance ◽  
Engineering Problems ◽  
Artificial Neural ◽  
Artificial Neural Network Ann ◽  
Carbon Dioxide Co2

Neural networks application (ANNs) has been used at recent years in several technology areas. Applying artificial neural networks (ANN) has too many environmental engineering problems provide remarkable success. The objective of this project is to predict and show the accuracy of using neural network in air quality prediction in Al Ahmadi city. Several parameters such as carbon dioxide (CO2), ammonia, benzene, relative humidity, temperature, air velocity and air direction are used in this study, for period of four years. The correlation coefficient was used to indicate the performance of ANN. The networks provide a good prediction performance with R value 0.93308 for CO2, 0.81805 and 0.923 for Ammonia and Benzene, respectively.

Sign in / Sign up

Export Citation Format

Share Document