The environment is made up of both living and non living things.
COLLECTION AND APPLICATIONS OF WEATHER & CLIMATE DATA
The most important aspect of forecasting is to present forecasts which are relevant to the time they are produced i.e. current and that they can be easily understood by the end-user. End-users include fisherman, navy personnel, armed forces, farmers, orchardists, airline pilots, and the agriculture, viticulture, mining and building industries. In fact a whole range of people need to be able to interpret and understand weather information. As such, information in the form of graphs, formulas, symbols, and so forth is not always the best approach. Instead verbal statements and written paragraphs interpreting data are usually most readily understood.
Weather maps show temperatures by depicting isotherms, where an isotherm is a type of contour line on the map which connects all the points that have the same temperature. Other weather maps might use isotachs which depict all the areas having the same wind speed. Another popular format is the isobar map which demonstrates areas of high and low pressure.
A type of data map used by meteorologists is the station model. This map is a simplified symbolic representation of weather recordings at a particular weather station location and includes data on a range of variables such as temperature, humidity, dew point, air pressure and precipitation.
Information acquired by satellites has become a very important part of data collection for weather forecasting. Information is collected from both polar orbiting satellites and geostationary ones which hover over the same area of the equator. This data is often used to fill in gaps, most notably over the ocean regions.
Weather satellites are able to gather information about clouds and cloud systems as well as garner environmental information such as air pollution, dust storms, auroras, volcanic eruptions and ash clouds, wildfires, ice caps, and ocean currents.
The information gathered by satellites may be in various formats such as infrared (heat) images, visible images, and water vapour images. Visible images depict actual images of the earth’s geographical features and atmospheric conditions. Infrared images may be used to determine surface water temperatures, cloud heights and types, and so on. Infrared images display the radiation emitted from objects and this information is useful for predicting areas of precipitation within cyclones.
Information about land and water temperatures can be used to help agriculturalists, farmers, and fisherman to make decisions about managing crops, animals, fish, and so on. Satellites can also be used to monitor pollution such as oceanic oil spillages, desert sand storms, and smoke from bushfires, and this information is useful in directing management strategies.
In the future it is expected that satellites will be able to record other atmospheric variables such as temperatures at different distances from the earth, humidity, and wind speeds.
This is used to detect precipitation and its movement. From this information the type of precipitation can also be predicted e.g. rain, sleet, or snow. These days the radars are mainly pulse-Doppler types which not only detect precipitation but also its intensity. This is valuable information for forecasting storms and their propensity to cause damage.
As already outlined, radar data is often used along with numerical data in NWP forecasts.
Tropical Rainfall Measuring Mission (TRMM)
The Tropical Rainfall Measuring Mission is joint project between the United States and Japan. It is the first mission to be established which measures both tropical and subtropical rainfall using sensors which detect microwave and visible infrared rays. It also boasts the first rain radar to be launched into space.
Through measuring tropical rainfall and its heat it is anticipated that the project will be able to provide valuable information about climate change through better understanding of how heat energy forces atmospheric circulation. It travels on a continuous course between 35°north and 35° south of the equator.
Given that the most accurate forecast predictions are associated with the El Nino–Southern Oscillation (ENSO) effect many of the less well developed tropical countries could benefit the most from a greater understanding of these weather events. However, given that ENSO events also have much further reaching effects such as cyclones, monsoons, and hurricanes producing worldwide flooding, colder winters in some temperate zone regions, and dryer monsoons in the southern hemisphere – the whole world can benefit from improved knowledge of rainfall systems.
Forecast verification is a means of testing the validity of the forecast. This is done by comparing the forecast to what actually happened. It is important to verify forecasts in order to:
• Check quality and whether forecasts are getting better
• Identify where improvements to techniques can be made
• Compare different forecast systems
It is widely agreed that the best forecasts are those which reflect what happened and this is considered to be the forecast ‘quality’. A number of researchers have broken quality down into further elements of which accuracy and skill (which is a measure of accuracy relative to the forecast type) are most important.
Forecasts are compared to what is often referred to as ‘truth’ data. This is recorded from actual observations provided by satellite cloud data, temperature measurements, rain gauge measurements, and so forth. Of course, even these measurements are prone to some degree of error such as sampling error, measurement errors, or analysis errors. Nevertheless, these observational errors are often ignored. The verification data is still valuable where the error is small or there isn’t much data to use as it still gives a good indication of forecast accuracy.
Naturally, the validity of verification techniques is higher when there is sufficient observational data to compare with and the quality of the data is strong. The results of verification analyses are often displayed within some confidence limit to acknowledge a degree of possible error.
Sometimes the reliability of samples is increased by pooling data. A larger pool of data implies more accurate comparison data, but a more accurate method is to put data into homogeneous groups e.g. on the basis of geographical region or the season.
Methods of Standard Verification
There are a number of different methods which can be used to verify results.
1) Eyeball Method
This is the traditional technique which relies on comparing the actual observations with the forecast data to note agreements and discrepancies. This method is useful for comparing data on just a few forecasts as it can be quite time-consuming. The problem is that it is prone to observer biases and personal interpretation and does not employ any quantitative techniques.
2) Dichotomous Methods
These are based on two premises: yes, an event will happen, or no an event will not happen. They are often used for forecasting snow, rain or storms. A contingency table can be drawn up which shows the various permutations.
From theese details, a weather event which is forecast to occur and which is observed to actually occur is a ‘hit’. An event which is forecast not to occur and does not occur is a correct negative, and so forth. A perfect forecast would only include hits and correct negatives but, of course, this is unlikely to occur.
Once data has been added to a contingency table different types of errors can easily be seen. This data is then subjected to different statistical formulas to produce a range of verification findings isolating different areas of interest. The types of statistical verification data produced include: accuracy, frequency bias, false alarm ratio, success ratio, hit rate, and so on.
3) Multiple Category Methods – this method also uses contingency tables where different forecast categories are compared with observed categories. This method is advantageous in that forecast errors can be more easily recognised. The problem is that fewer statistical analyses can be conducted on the findings since results are difficult to present as a single number. Histograms may be used to represent category frequencies.
4) Continuous Variables Methods – a table is used to record forecast and observed values at regular time intervals e.g. one day. Findings may be plotted on graphs and a range of statistical computations performed on them to determine bias, error, probability, skill, and so forth.
5) Probabilistic Forecast Methods – these aim to give the probability of a weather event occurring. Probabilities are produced with a range of 0 and 100% (represented as 0 and 1). A better forecast would have a probability towards either of the extremes rather than somewhere nearer 0.5, and this is referred to as sharpness. Graphs of reliability are produced and a range of complex statistical formula may be applied to findings.
Other Verification Methods
In addition to standard verification techniques, a number of scientific (diagnostic) methods are used to verify data which are beyond the scope of this course. They are much more complex.