AU2021102843A4 - Deep Learning-Based Weather Prediction in Global Position - Google Patents

Abstract

Our invention Deep Learning-Based Weather Prediction in international Position could be a Weather prediction application and web-based application wherever you may be ready to access all reports associated with weather forecasts for outlined locations. the placement detected by your browser setting and native server configuration can mechanically determine the placement and be ready to gift its weather Data/ info like temperature, wind direction, rainfall, humidness etc. The invention is additionally AN explores the utilization of machine learning advanced methodologies to unravel outlined issues in meteorology and notably focuses on the exploration of methodologies to enhance the add-on accuracy of numerical weather prediction unit. The invention could be a classical methodology like variable statistic regression and binary trees, area unit accustomed perform regression on advanced meteoric meta data and therefore the 1st half notably focuses on prognostication wind standing, whose circular nature creates attention-grabbing challenges for traditional machine learning algorithms. The second a part of this thesis explores the analysis of weather information as a generic structured prediction drawback mistreatment deep neural networks and therefore the Neural networks, like convolutional and perennial networks, give a way for capturing the abstraction and temporal structure inherent in weather prediction models. 1 TOTAL NO OF SHEET: 03 NO OF FIG: 03 100 Collection of ------.-- Data Flow Data Set Unit10 104 Filtering Data Unit and rocess16 Classification of a Data Deep Learning Unit Programming Unt0 114 Bayes Method 112 110 Performance Status and Analysis Unit FIG.1: Weather Prediction flow Chart.

Description

TOTAL NO OF SHEET: 03 NO OF FIG: 03

100 Collection of ------.--Data Flow

Data Set Unit10 104

Filtering Data Unit and rocess16

Classification of a Data Deep Learning Unit Programming Unt0

114 Bayes Method

112 110

Performance Status and Analysis Unit

FIG.1: Weather Prediction flow Chart.

Australian Government IP Australia Innovation Patent Australia

Deep Learning-Based Weather Prediction in Global Position.

Name and address of patentees(s): Roorkee College of Engineering Address: Roorkee College of Engineering Plot 312 Bagheri, Roorkee-247667, Uttarakhand, India. Anurag Chandna Roorkee College of Engineering Plot 312 Bagheri, Roorkee-247667, Uttarakhand, India. Complete Specification: Australian Government.

I

FIELD OF THE INVENTION Our invention is related to a Deep Learning-Based Weather Prediction in Global Position. BACKGROUND OF THE INVENTION Weather forecasting is the use of science and technology to predict the atmospheric conditions of a given area and time and also people have been trying to predict the weather informally for thousands of years and since the +19th century. The Weather forecasts are made by collecting information about the current state of the atmosphere in a particular area and then using the weather to predict how the atmosphere will completely change.

User input is still required to select the best predictive model to establish the prediction when it comes to human action and activity that is largely based on changes in barometric stress, current climate and weather (cloud cover) weather forecasting is now relying on computer-based models that look at a number of celestial objects. The Individual input is still required to select the best predictive unit/model to establish the prediction, which includes pattern recognition skills, telephone WI-FI, communication, unit/model performance information, and model bias information. OBJECTIVES OF THE INVENTION 1. The objective of the invention is to a machine Learning-Based Weather Prediction in Global Position is a Weather prediction application and web based application where you will be able to access all reports related to weather forecasts for defined locations. 2. The other objective of the invention is to a location detected by your browser setting and local server configuration will automatically identify the location and be able to present its weather Data/ information such as temperature, wind direction, rainfall, humidity etc. 3. The other objective of the invention is to a explores the use of machine learning advanced methodologies to solve defined problems in meteorology and particularly focuses on the exploration of methodologies to improve the add-on accuracy of numerical weather prediction unit. 4. the opposite objective of the invention is to a classical methodology like variable statistic regression and binary trees, square measure wants to perform regression on advanced meteoric meta-data and also the 1st half significantly focuses on prognostication wind standing, whose circular nature creates attention-grabbing challenges for traditional machine learning algorithms. 5. The other objective of the invention is to the invention is to a explores the analysis of weather data as a generic structured prediction problem using deep neural networks. 6. The other objective of the invention is to a Neural network, such as convolutional and recurrent networks, provide a method for capturing the spatial and temporal structure inherent in weather prediction models.

SUMMARY OF THE INVENTION The System Analysis could be a term want to describe the method of grouping associate degreed analyzing facts concerning the functioning of an existing atmosphere in order that a fancy integrated software is often developed and used if it's found to be possible.

System analysis can be considered as the real-time and perhaps the most complete method of solving computer complex problems. The fully help and understand and compare the performance implications of a sub-program and system analysis also involves the design of the system which is a function that involves the creation of an integrated unique system based on facts revealed during the analysis.

The predicted results are continuous numerical values, the temperature in us, we use the regression method and also find that Unique Random Forest Regression (URFR) is a superior regressors as it incorporates many decision Binary- trees while making a decision and also, we suggest comparing several other state-of the-art ML strategies with the RFR process.

Restoration strategies included Ridge Regression (Ridge), Advanced Support Vector (ASVR), Neural Multi-Layer Perceptron (NMLPR), and Extra-Binary Tree Regression (EBTR).

Data Preprocessing: After receiving raw information/data from 'underground', we make sure that each line in the database contains records of all ten cities for a period of time. We eliminate any feature with empty or invalid data while creating a database and also, we convert separate elements in the database such as wind direction and position, into dummy- indicator variables Tx, Ty, Tzu using a process called 'One Hot Encoding'.

We underwent this modification prior to the division of training and evaluation data. This is because, in both training and testing data, we need the same amount of feature variability. If we make this adjustment after a split, then there is no guarantee that both will have all the values of the categories of adjustable features. BRIEF DESCRIPTION OF THE DIAGRAM FIG.1: Weather Prediction flow Chart. FIG.2: Weather Prediction Deep Learning Process. Fig.3: Deep Learning-Based Weather Prediction

DESCRIPTION OF THE INVENTION Data related to weather

The classification data for equipment division, you must convert text labels to another form. There are two types of encoding.

1. First label, each text label value is replaced by a number.

2. The second is to enter one code "hot", each text label value becomes a column with a binary value (1 or 0).

The Most machine learning platforms have functions that enable coding itself and also as a rule, hot one-line encoding is appealing, since labeling numbers can sometimes confuse the machine learning algorithm and making you think that the codes were ordered. The calculating Euclidean distances, high-value numbers can begin to "dominate" unreasonably, and even the slightest decline may not coincide.

The evaluation results show that these machine learning models can predict weather features accurately enough to compete with traditional models.

The origins and evolution of weather forecasting

Weather forecasting is defined as the application of science and methods/technology to predict the conditions of the atmosphere for a given location and Real-time in the future and also humans have tried to understand the behavior of the atmosphere by studying the various patterns and relationships between phenomena and relating them with future events.

For example: it's been renowned for hundreds of years that an explosive descent within the atmospheric pressure is commonly followed by precipitation events. Mr. L. F. Richardson planned the utilization of basic mechanic's equations to model the movements of the atmosphere. At that point, there was no thanks to alter calculations, thus this author came up with the concept of cacophonic the surface of the planet up into cells and mistreatment individuals to unravel the differential equations that describe the movements of the atmosphere. per his estimates, a military of +64,000 advanced human computers would be needed to get AN updated forecast for the complete planet.

Numerical weather prediction

The previous section, computers have been used since the +1950 to simulate the state and evolution of the atmosphere and also Ever since this time, capacity of the computers has doubled every +18 months following Moore's Law and similarly the resolution of the weather models. Although the physical equations and methodologies for resolving them have remained fundamentally the same as in the first weather models the spatial and temporal resolution, as well as the frequency of the runs of the models, has constantly increased over time.

This means that the accuracy of a +4-day forecast today is as good as it was +10 years ago for a 2-day forecast. The field of NWP verification has been extensively studied in weather forecasting and also the need for methods that can measure the accuracy and quality of the information produced is fundamental to identify weaknesses in the simulation of the atmospheric processes.

The two baselines that are often used to measure the quality of the forecasts are persistence in short range 0-Day to 2-days forecasts and climatology in mid- 2 day to 5-days to long-range +5days forecasts. The assumption that the meteorological conditions are going to stay the same. In this context, an NWP unit needs to be better at forecasting the weather than a simple model which keeps the variables constant in time. The climatology baseline, on the other hand, assumes that the weather is going to behave as the averaged historical records for that particular area.

The sources of weather data

Our attention on NWP as this is currently the only tool available that is able to "forecast" the evolution and future state of the physical parameters describing the atmosphere. The NWP output is not the only source of information when studying the atmosphere and Sensors in many forms provide accurate values of the observed conditions in a region of the atmosphere.

Examples: of sensors used in weather forecasting are ground stations, atmospheric sounding balloons, weather rears or satellites. Each of all types of sensors has different characteristics in terms of the spatial and temporal resolution and the extent that they can cover. Observational data sets are crucial in the process of weather forecasting, as they provide a live stream of information describing the current state of the atmosphere.

Forecasters use these data operationally to validate the-NWP output and to correct for possible errors or local effects. Especially relevant to this thesis work are -METARs, which are weather observation reports generated in most of the airports in the world.

These reports are used to plan air traffic control in airports, and are one of the highest quality observed data sources available. The following code represents a =METAR report for the airport of Don ostia, Spain. The code starts with the International Civil Aviation Organization code of the airport, date of the report, wind conditions, visibility, cloud coverage, temperature and finishes with the pressure conditions.

Weather forecasting: The machine learning approach:

Before the existence of -NWP-models, humans used a simple technique to forecast the weather based on the observed data collected over the years for a specific region, which is known as climatology. The estimations about long-range trends and cumulative values of variables, such as temperature or precipitation, can be inferred by applying basic statistics to observed weather events over long enough periods of time. The +20th century, with the advent of the technology to accurately measure the atmosphere and to communicate these values across geographically distant places, weather maps became available.

These maps initially started representing the position and shape of the low- and high-pressure systems and allowed the development of weather forecasting methodologies, which predicted the movement and effects of these pressure system in the atmosphere. The application of statistical models to weather forecasting was proposed.

The first -NWP-models were developed, Malone made the argument that "statistics must eventually play some role" in the simulation of the atmosphere. The author demonstrated a methodology, using multiple linear regression, to forecast the sea-level pressure field.

The first attempt to use regression methods to predict the evolution of the surface temperature at Indianapolis (USA). This model uses the atmospheric circulation pressure field in the previous 24 hours as input to forecast the next +24 hours.

The invention is also a to change location you will have to select the options provided below to get its details and Its new avatar and feed burner will also allow its users to receive weather reports directly from their mail where they have not been able to access this particular domain even if the server is down.

This part explores the potential for deep convolutional neural networks to solve difficult problems in meteorology, such as modelling precipitation from basic numerical model fields. The research underpinning this thesis serves as an example of how collaboration between the machine learning and meteorology innovation communities is mutually beneficial and leads to advances in both disciplines. Weather forecasting models and observational data represent unique examples of large structured and high-quality data sets, which the machine learning community demands for developing the next generation of scalable algorithms.

The invention may be a foretelling is even nowadays associate activity in the main performed by humans and a pc simulation play a significant role in modelling the state and evolution of the atmosphere, there's a scarcity of methodologies to modify the interpretation of the knowledge generated by these models.

WE CLAIMS 1) Our invention Deep Learning-Based Weather Prediction in world Position could be a Weather prediction application and web-based application wherever you may be able to

access all reports associated with weather forecasts for outlined locations. the placement detected by your browser setting and native server configuration can mechanically

determine the placement and be able to gift its weather Data/ data like temperature, wind

direction, rainfall, wetness etc. The invention is additionally AN explores the utilization of machine learning advanced methodologies to unravel outlined issues in meteorology and

significantly focuses on the exploration of methodologies to boost the add-on accuracy of numerical weather prediction unit. The invention could be a classical methodology like

variable statistic regression and binary trees, are wont to perform regression on advanced earth science meta-data and therefore the initial half significantly focuses on statement

wind standing, whose circular nature creates fascinating challenges for traditional machine learning algorithms. The second a part of this thesis explores the analysis of weather

information as a generic structured prediction downside victimization deep neural

networks and therefore the Neural networks, like convolutional and perennial networks, give a way for capturing the special and temporal structure inherent in weather prediction

models. 2) According to claims# the invention is to a machine Learning-Based Weather Prediction in

world Position could be a Weather prediction application and web-based application wherever you may be able to access all reports associated with weather forecasts for

outlined locations. the placement detected by your browser setting and native server configuration can mechanically determine the placement and be able to gift its weather

Data/ data like temperature, wind direction, rainfall, wetness etc.

3) in step with claiml,2# the invention is to a explores the utilization of machine learning advanced methodologies to unravel outlined issues in meteorology and significantly

focuses on the exploration of methodologies to boost the add-on accuracy of numerical weather prediction unit.

4) in step with claiml,2,3# the invention is to a classical methodology like variable statistic

regression and binary trees, are wont to perform regression on advanced earth science meta-data and therefore the initial half significantly focuses on statement wind standing,

whose circular nature creates fascinating challenges for traditional machine learning algorithms and additionally the invention is to a explores the analysis of weather

information as a generic structured prediction downside victimization deep neural

networks and therefore the Neural networks, like convolutional and perennial networks, give a way for capturing the special and temporal structure inherent in weather prediction

models.

TOTAL NO OF SHEET: 03 NO OF FIG: 03 26 May 2021 2021102843

FIG.1: Weather Prediction flow Chart.

TOTAL NO OF SHEET: 03 NO OF FIG: 03 26 May 2021 2021102843

FIG.2: Weather Prediction Deep Learning Process.

TOTAL NO OF SHEET: 03 NO OF FIG: 03 26 May 2021 2021102843

Fig.3: Deep Learning-Based Weather Prediction