Epilepsy Action aims to improve the quality of life and promote the interests of people living with epilepsy. • We BLOCKINprovide BLOCKINinformation BLOCKINto BLOCKINanyone BLOCKINwith BLOCKINan BLOCKINinterest BLOCKINin BLOCKINepilepsy. • We BLOCKINimprove BLOCKINthe BLOCKINunderstanding BLOCKINof BLOCKINepilepsy BLOCKINin BLOCKINschools BLOCKINand raise educational standards. • We BLOCKINwork BLOCKINto BLOCKINgive BLOCKINpeople BLOCKINwith BLOCKINepilepsy BLOCKINa BLOCKINfair BLOCKINchance BLOCKINof BLOCKINfinding and keeping a job. • We BLOCKINraise BLOCKINstandards BLOCKINof BLOCKINcare BLOCKINthrough BLOCKINcontact BLOCKINwith BLOCKINdoctors, nurses, BLOCKINsocial BLOCKINworkers, BLOCKINgovernment BLOCKINand BLOCKINother BLOCKINorganisations. • We BLOCKINpromote BLOCKINequality BLOCKINof BLOCKINaccess BLOCKINto BLOCKINquality BLOCKINcare. Epilepsy Action has local branches in most parts of the UK. Each branch offers support to local people and raises money to help ensure our work can continue.

How is our work

Cholera is an intestinal disease and is characterized by diarrhea and severe dehydration. While cholera has mainly been eliminated in regions that can provide clean water, adequate hygiene and proper sanitation; it remains a constant threat in many parts of Africa and Asia. Within this paper, we develop an agent-based model that explores the spread of cholera in the Dadaab refugee camp in Kenya. Poor sanitation and housing conditions contribute to frequent incidents of cholera outbreaks within this camp. We model the spread of cholera by explicitly representing the interaction between humans and their environment, and the spread of the epidemic using a Susceptible-Exposed-Infected-Recovered model. Results from the model show that the spread of cholera grows radially from contaminated water sources and seasonal rains can cause the emergence of cholera outbreaks. This modeling effort highlights the potential of agent-based modeling to explore the spread of cholera in a humanitarian context. An agent-based model was developed to explore the spread of cholera.The progress of cholera transmission is represented through a Susceptible-Exposed-Infected-Recovered (SEIR) model.The model integrates geographical data with agents' daily activities within a refugee camp.Results show cholera infections are impacted by agents' movement and source of contamination.The model has the potential for aiding humanitarian response with respect to disease outbreaks.

An agent-based modeling approach applied to the spread of cholera

Modern societies are exposed to a myriad of risks ranging from disease to natural hazards and technological disruptions. Exploring how the awareness of risk spreads and how it triggers a diffusion of coping strategies is prominent in the research agenda of various domains. It requires a deep understanding of how individuals perceive risks and communicate about the effectiveness of protective measures, highlighting learning and social interaction as the core mechanisms driving such processes. Methodological approaches that range from purely physics-based diffusion models to data-driven environmental methods rely on agent-based modeling to accommodate context-dependent learning and social interactions in a diffusion process. Mixing agent-based modeling with data-driven machine learning has become popularity. However, little attention has been paid to the role of intelligent learning in risk appraisal and protective decisions, whether used in an individual or a collective process. The differences between collective learning and individual learning have not been sufficiently explored in diffusion modeling in general and in agent-based models of socio-environmental systems in particular. To address this research gap, we explored the implications of intelligent learning on the gradient from individual to collective learning, using an agent-based model enhanced by machine learning. Our simulation experiments showed that individual intelligent judgement about risks and the selection of coping strategies by groups with majority votes were outperformed by leader-based groups and even individuals deciding alone. Social interactions appeared essential for both individual learning and group learning. The choice of how to represent social learning in an agent-based model could be driven by existing cultural and social norms prevalent in a modeled society.

/pdf/risk-perception-and-behavioral-change-during-epidemics-3603uck0ym.pdf

Risk perception and behavioral change during epidemics: Comparing models of individual and collective learning.

High resolution satellite imagery and modern machine learning methods hold the potential to fill existing data gaps in where crops are grown around the world at a sub-field level. However, high resolution crop type maps have remained challenging to create in developing regions due to a lack of ground truth labels for model development. In this work, we explore the use of crowdsourced data, Sentinel-2 and DigitalGlobe imagery, and convolutional neural networks (CNNs) for crop type mapping in India. Plantix, a free app that uses image recognition to help farmers diagnose crop diseases, logged 9 million geolocated photos from 2017–2019 in India, 2 million of which are in the states of Andhra Pradesh and Telangana in India. Crop type labels based on farmer-submitted images were added by domain experts and deep CNNs. The resulting dataset of crop type at coordinates is high in volume, but also high in noise due to location inaccuracies, submissions from out-of-field, and labeling errors. We employed a number of steps to clean the dataset, which included training a CNN on very high resolution DigitalGlobe imagery to filter for points that are within a crop field. With this cleaned dataset, we extracted Sentinel time series at each point and trained another CNN to predict the crop type at each pixel. When evaluated on the highest quality subset of crowdsourced data, the CNN distinguishes rice, cotton, and “other” crops with 74% accuracy in a 3-way classification and outperforms a random forest trained on harmonic regression features. Furthermore, model performance remains stable when low quality points are introduced into the training set. Our results illustrate the potential of non-traditional, high-volume/high-noise datasets for crop type mapping, some improvements that neural networks can achieve over random forests, and the robustness of such methods against moderate levels of training set noise. Lastly, we caution that obstacles like the lack of good Sentinel-2 cloud mask, imperfect mobile device location accuracy, and preservation of privacy while improving data access will need to be addressed before crowdsourcing can widely and reliably be used to map crops in smallholder systems.

Mapping Crop Types in Southeast India with Smartphone Crowdsourcing and Deep Learning

Multitemporal Earth observation capability plays an increasingly important role in crop monitoring. As the frequency of satellite acquisition of remote sensing images becomes higher, how to fully exploit the implicit phenological laws in dense multitemporal data is of increasing importance. In this article, we propose a CNN-transformer approach to perform the crop classification, in the model, we borrow the transformer architecture from the knowledge of NLP to dig into the pattern of multitemporal sequence. First, after unifying the spatial-spectral scale of each multiband data acquired from different sensors, we obtain the scale-consistent feature and position feature of multitemporal sequence. Second, with adopting multilayer encoder modules derived from the transformer, we mine deep correlation patterns of multitemporal sequence. Finally, the feed-forward layer and softmax layer serve as output layers of the model to predict crop categories. The proposed CNN-transformer approach is illustrated in a crop-rich agricultural region in central California, where 65 multitemporal profiles from multisensor Sentinel-2 A/B and Landsat-8 are obtained in 2018. Through multiband multiresolution fusion, sequence correlation extraction of multitemporal data and category feature extraction, the classification results show that the proposed method has a significant performance improvement compared with other traditional methods.

/pdf/a-cnn-transformer-hybrid-approach-for-crop-classification-34armczkrw.pdf

A CNN-Transformer Hybrid Approach for Crop Classification Using Multitemporal Multisensor Images

It is of paramount importance to have sustainable agriculture since agriculture is the backbone of many nations’ economic development. Majority of agricultural professionals rarely capture the cropping patterns necessary to promote Good Agricultural Practises. Objective of this research is to explore the potential of mapping cropping patterns occurring on different field parcels on small-scale farmlands in Zimbabwe. The first study location under investigation are the International Maize and Wheat Improvement Center (CIMMYT) research station and a few neighboring fields, the second is Middle Sabi Estate. Fourier time series modeling was implemented to determine the trends befalling on the two study sites. Results reveal that Sentinel-1 synthetic aperture radar (SAR) time series allow detection of subtle changes that occur to the crops and fields respectively, hence can be utilized to detect cropping patterns on small-scale farmlands. Discrimination of the main crops (maize and soybean) grown at CIMMYT was possible, and crop rotation was synthesized where sowing starts in November. A single cropping of early and late crops was observed, there were no winter crops planted during the investigation period. At Middle Sabi Estate, single cropping on perennial sugarcane fields and triple cropping of fields growing leafy vegetables, tomatoes and onions were observed. Classification of stacked images was used to derive the crop rotation maps representing what is practised at the farming lands. Random forest classification of the multi-temporal image stacks achieved overall accuracies of 99% and 95% on the respective study sites. In conclusion, Sentinel-1 time series can be implemented effectively to map the cropping patterns and crop rotations occurring on small-scale farming land. We recommend the use of Sentinel-1 SAR multi-temporal data to spatially explicitly map cropping patterns of single-, double- and triple-cropping systems on both small-scale and large-scale farming areas to ensure food security.

Exploring the Potential of Mapping Cropping Patterns on Smallholder Scale Croplands Using Sentinel-1 SAR Data

Decision-level fusion of Sentinel-1 SAR and Landsat 8 OLI texture features for crop discrimination and classification: case of Masvingo, Zimbabwe

Crops mapping unequivocally becomes a daunting task in humid, tropical, or subtropical regions due to unattainability of adequate cloud-free optical imagery. Objective of this study is to evaluate the comparative performance between decision- and pixel-levels data fusion ensemble classified maps using Landsat 8, Landsat 7, and Sentinel-2 data. This research implements parallel and concatenation approach to ensemble classify the images. The multiclassifier system comprises of Maximum Likelihood, Support Vector Machines, and Spectral Information Divergence as base classifiers. Decision-level fusion is achieved by implementing plurality voting method. Pixel-level fusion is achieved by implementing fusion by mosaicking approach, thus appending cloud-free pixels from either Sentinel-2 or Landsat 7. The comparison is based on the assessment of classification accuracy. Overall accuracy results show that decision-level fusion achieved an accuracy of 85.4%, whereas pixel-level fusion classification attained 82.5%, but their respective kappa coefficients of 0.84 and 0.80 but are not significantly different according to Z-test at $\alpha = {\text{0.05}}$ . F1-score values reveal that decision-level performed better on most individual classes than pixel-level. Regression coefficient between planted areas from both approaches is 0.99. However, Support Vector Machines performed the best of the three classifiers. The conclusion is that both decision-level and pixel-level fusion approaches produced comparable classification results. Therefore, either of the procedures can be adopted in areas with inescapable cloud problems for updating crop inventories and acreage estimation at regional scales. Future work can focus on performing more comparison tests on different areas, run tests using different multiclassifier systems, and use different imagery.

/pdf/comparative-performance-evaluation-of-pixel-level-and-33mav2os9a.pdf

Comparative Performance Evaluation of Pixel-Level and Decision-Level Data Fusion of Landsat 8 OLI, Landsat 7 ETM+ and Sentinel-2 MSI for Crop Ensemble Classification

Accurate and spatially explicit cropland maps are crucial for many applications, which include sustainable crop monitoring, food security, and land and agriculture planning and management. Zimbabwe lacks reliable data on cropland extent of the old and new re-allocated areas for inventory purposes. Objectives of this paper are to map cropland utilizing: 1) automatic classification; 2) multi-classifier system (MCS); and 3) normalized difference vegetation index and bare-soil index (NDVI-BSI) thresholding and determine the spatiotemporal cropland changes. Change detection is implemented through a post-classification statistical method. The classified results are compared with SADC and ESA land cover products, GFSAD30AFCE cropland layer, and Google Earth imagery. Results reveal that MCS and NDVI-BSI performed the best and achieved overall accuracies of 80.54% and 79.32% for 2013, and for 2018, they attained accuracies of 87.90% and 88.56%, respectively. Automated classification, MCS, and NDVI-BSI thresholding produced average cropland areas of 3416396, 10346778, and 9788833 Ha, respectively. Visual assessment observations show that NDVI-BSI thresholding outperformed the other two techniques. Comparing further the MCS and NDVI-BSI thresholding approaches’ results of total cropland areas of Zimbabwe’s ten provinces for the years 2013 and 2018, coefficients of determination of 0.8404 and 0.9619, respectively, are achieved. Change detection shows a general increase in the cropland area due to human activities despite the prolonged drought. However, we recommend further exploration of NDVI-BSI threshold values to derive cropland layers since the method is robust and can be automated easily and faster without inputting training data. We also recommend simulation of the changes in cropland areas using cellular automata and/or agent-based modeling.

/pdf/cropland-mapping-and-change-detection-toward-zimbabwean-4et95yn647.pdf

Cropland Mapping and Change Detection: Toward Zimbabwean Cropland Inventory

This paper introduces a spatially explicit agent-based simulation model for micro-scale cholera diffusion. The model simulates both an environmental reservoir of naturally occurring V.cholerae bacteria and hyperinfectious V. cholerae. Objective of the research is to test if runoff from open refuse dumpsites plays a role in cholera diffusion. A number of experiments were conducted with the model for a case study in Kumasi, Ghana, based on an epidemic in 2005. Experiments confirm the importance of the hyperinfectious transmission route, however, they also reveal the importance of a representative spatial distribution of the income classes. Although the contribution of runoff from dumpsites can never be conclusively proven, the experiments show that modelling the epidemic via this mechanism is possible and improves the model results. Relevance of this research is that it shows the possibilities of agent-based modelling combined with pattern reproduction for cholera diffusion studies. The proposed model is simple in its setup but can be extended by adding additional elements such as human movement and change of behaviour of individuals based on disease awareness. Eventually, agent-based models will open opportunities to explore policy related research questions related to interventions to influence the diffusion process.

/pdf/agent-based-modelling-of-cholera-diffusion-nhhpae5uxs.pdf

Juliana Useya

Papers

Exploring the Potential of Mapping Cropping Patterns on Smallholder Scale Croplands Using Sentinel-1 SAR Data

Decision-level fusion of Sentinel-1 SAR and Landsat 8 OLI texture features for crop discrimination and classification: case of Masvingo, Zimbabwe

Comparative Performance Evaluation of Pixel-Level and Decision-Level Data Fusion of Landsat 8 OLI, Landsat 7 ETM+ and Sentinel-2 MSI for Crop Ensemble Classification

Cropland Mapping and Change Detection: Toward Zimbabwean Cropland Inventory

Agent-based modelling of cholera diffusion