Use evidence to predict short term weather changes probabilistically - Middle School Earth and Space Science

SECTION 1

Historical Context & Motivation

Have you ever checked the weather app on a phone and seen "70% chance of rain"? That number is a probability (a way of saying how likely something is to happen). People have been trying to predict the weather for thousands of years. Ancient farmers watched clouds, winds, and animal behavior for clues about what was coming next.

For most of history, weather prediction was based on patterns people noticed over time. A red sky at sunset might mean fair weather the next day. A ring around the moon might mean rain. These were helpful hints, but they were not always accurate. Scientists needed better tools and methods.

1643 • Invention of the Barometer

Evangelista Torricelli invented the barometer (a tool that measures air pressure). Falling air pressure often means a storm is approaching.

1844 • Telegraph Changes Everything

The telegraph let weather stations share data across long distances almost instantly. For the first time, scientists could track storms as they moved.

1950 • Computer Weather Models

Scientists used the first electronic computers to run weather models (mathematical simulations of the atmosphere). Forecasts became more accurate.

1960 • Weather Satellites Launch

TIROS-1, the first weather satellite, sent images of clouds from space. Scientists could now see entire storm systems forming over the oceans.

2000s • Probabilistic Forecasting

Modern supercomputers run many simulations at once. They give forecasts as probabilities, like "80% chance of thunderstorms." This helps people make better decisions.

Today, weather forecasting is a blend of data collection, computer modeling, and probability. The big question we will explore is: How do scientists use evidence to predict short-term weather changes, and why do they express those predictions as probabilities?

SECTION 2

Core Principles of Weather Prediction

Predicting weather is all about gathering evidence and looking for patterns. Scientists measure many variables in the atmosphere. They compare today's measurements to what happened in the past when conditions were similar. Let's look at the key ideas.

Weather Variables

Scientists measure temperature, air pressure, humidity (moisture in the air), wind speed, and wind direction. These are the raw evidence for every forecast.

Cause and Effect in the Atmosphere

Weather changes happen because energy from the Sun heats Earth's surface unevenly. Warm air rises, cool air sinks, and air masses (large bodies of air with similar temperature and humidity) push against each other. These interactions cause wind, clouds, and precipitation.

Probability Means "How Likely"

A probability is a number from 0% to 100% that tells how likely an event is. A 0% chance means it almost certainly will not happen. A 100% chance means it almost certainly will. Weather forecasts use probability because the atmosphere is complex and always changing.

Patterns from Data

Scientists look at patterns in weather data over time. For example, if a cold front is approaching and the last 8 out of 10 times a similar front passed, it rained, the forecast might say "80% chance of rain." Past patterns guide future predictions.

Short-Term vs. Long-Term

Short-term forecasts (1–3 days) are much more accurate than long-term ones. The farther into the future you look, the more uncertain the prediction becomes. This is because small changes in the atmosphere can grow into big differences over time.

✦

KEY TAKEAWAY

Think of weather prediction like guessing what your friend will order at a restaurant. If they ordered pizza the last 9 out of 10 times, you would predict pizza with about 90% confidence. You are using past evidence and patterns to make a probabilistic prediction. Weather forecasters do the same thing, but they use measurements of air pressure, temperature, and humidity instead of pizza orders.

SECTION 3

How Weather Data Becomes a Forecast

The diagram below shows the process scientists use to turn raw weather data into a probability-based forecast. Follow the arrows from left to right to see how evidence flows through the system.

This diagram shows the four-step process of weather forecasting. Scientists collect data, find patterns, run computer models, and then produce a probabilistic forecast. The bottom box explains what a percentage like "70% chance of rain" actually means.

Notice how the process starts with real measurements. You cannot make a good prediction without solid evidence first. Each step builds on the one before it. The final forecast is expressed as a probability because the atmosphere is a complex system with many interacting parts. Small changes in one place can lead to big differences later.

SECTION 4

How Probability Works in Weather Forecasting

You do not need advanced math to understand weather probability. The basic idea is simple: scientists count how often certain outcomes happen under certain conditions. Let's look at the math behind it.

BASIC PROBABILITY

Probability = (Number of times the event happened ÷ Total number of similar situations) × 100%

For example, if a cold front with these exact conditions has brought rain 7 out of 10 times in the past, the probability of rain = (7 ÷ 10) × 100% = 70%.

Modern forecasters use ensemble forecasting (running many computer simulations at the same time with slightly different starting conditions). Each simulation is called a model run. If 18 out of 20 model runs predict rain, the probability is:

ENSEMBLE PROBABILITY

Probability of rain = (18 ÷ 20) × 100% = 90%

The more model runs that agree on the same outcome, the higher the probability. When models disagree, the probability drops closer to 50%, meaning the forecast is uncertain.

ℹ

Crosscutting Concept: Cause and Effect

Weather events have causes. A drop in air pressure causes air to rise, which causes water vapor to cool and form clouds. Identifying these cause-and-effect chains helps scientists predict the effects (like precipitation) from the causes (like dropping air pressure). However, because many causes interact at once, the outcome is probabilistic rather than certain.

This is why weather forecasts are not just "yes it will rain" or "no it won't." Instead, scientists use numbers like 30%, 60%, or 90% to communicate how confident they are. A higher percentage means stronger evidence points to that outcome. A lower percentage means the evidence is mixed or unclear.

SECTION 5

Weather Evidence: Tools and Data Sources

Scientists use many tools to gather weather evidence. Each tool measures a different variable (a quantity that can change). The more variables you measure, the better your prediction can be. Let's look at the main tools and what they tell us.

Six key tools used to gather weather evidence. Ground-based instruments (top row) measure temperature, air pressure, humidity, and wind speed. Radar and satellites (bottom row) detect precipitation patterns and large-scale storm systems.

Each of these tools provides a different piece of evidence. Scientists combine all of these measurements to build a complete picture of what the atmosphere is doing right now. They then compare this picture to past situations to estimate what will happen next. This is the Science and Engineering Practice of analyzing and interpreting data.

Tool	What It Measures	Why It Matters for Forecasting
Thermometer	Temperature of the air	Temperature differences drive wind and determine rain vs. snow.
Barometer	Air pressure (in millibars)	Falling pressure often signals approaching storms.
Hygrometer	Humidity (% of moisture)	High humidity means more moisture available for rain.
Anemometer	Wind speed and direction	Tells which air mass is moving toward your area.
Radar	Precipitation location and intensity	Shows where rain or snow is falling right now and where it is heading.
Satellite	Cloud patterns and storm systems	Provides a bird's-eye view of weather moving across entire regions.

Summary of weather tools and their roles in forecasting

SECTION 6

Worked Example: Predicting Tomorrow's Weather

Let's walk through how a student could use weather data to make a probabilistic prediction. Imagine you are a junior meteorologist looking at today's data for your city.

Will It Rain Tomorrow?

1
Step 1 — Gather Today's Evidence

You record today's conditions: temperature is 68°F, air pressure is falling (from 1018 mb to 1005 mb), humidity is rising (from 60% to 85%), and wind is blowing from the south at 15 mph. You also see thick gray clouds building on the radar to the southwest.
2
Step 2 — Identify Patterns

You compare today's conditions to a table of past weather events. You find 20 past days with similar conditions: falling pressure, rising humidity, south wind, and approaching clouds.
3
Step 3 — Count Outcomes

Out of those 20 similar past days, rain occurred the next day on 16 days. On 4 of those days, the rain missed or broke apart before arriving.

16 rainy days out of 20 similar situations
4
Step 4 — Calculate Probability

Probability = (16 ÷ 20) × 100% = 80%. This means there is an 80% chance of rain tomorrow.

80% probability of rain
5
Step 5 — Communicate the Forecast

Your forecast: "There is an 80% chance of rain tomorrow based on falling pressure, rising humidity, and south winds. Conditions are similar to 20 past events, 16 of which brought rain." Notice you explain both the probability and the evidence behind it.

Forecast supported by data and expressed as a probability

✦

KEY TAKEAWAY

Making a weather prediction is like being a detective. You gather clues (data), compare them to past cases (patterns), and then give your best estimate of what will happen. You never say "I am 100% sure" unless all the evidence points one way, because the atmosphere can always surprise you.

SECTION 7

Strengths and Limitations of Weather Forecasting

Weather forecasting has gotten much better over the decades. A 5-day forecast today is as accurate as a 1-day forecast was in 1980! But there are still important limitations. Let's compare what forecasting does well and where it struggles.

Strengths	Limitations
Short-term forecasts (1–3 days) are very accurate — often above 80% correct.	Forecasts beyond 7–10 days are much less reliable because small errors grow over time.
Radar shows real-time precipitation, so we can track storms as they move.	Radar cannot see what is happening hundreds of miles away or over oceans without satellites.
Ensemble models give a range of possible outcomes, helping us understand uncertainty.	Even the best models are approximations. The real atmosphere has more detail than any computer can capture.
Satellites provide global coverage, including remote areas over oceans.	Local effects like sea breezes or mountain updrafts can be hard for large-scale models to predict.
Probabilistic forecasts help people plan for the most likely and worst-case scenarios.	A 30% chance of rain does not mean no rain — people sometimes misunderstand what probabilities mean.

Strengths and limitations of modern weather forecasting

🔄

STABILITY AND CHANGE

The atmosphere is a system that shows both stability (large patterns like fronts and pressure systems move in predictable ways) and change (small disturbances can grow and shift a storm's path). This balance between stability and change is why short-term forecasts work well but long-term ones are uncertain. This is the crosscutting concept of Stability and Change.

SECTION 8

Short-Term Weather vs. Long-Term Climate

It is important to understand the difference between weather and climate. Weather describes what the atmosphere is doing right now or in the next few days. Climate describes the average weather patterns in a region over 30 years or more. Predicting each one requires different approaches.

Feature	Short-Term Weather Prediction	Long-Term Climate Projection
Time Scale	Hours to about 10 days	Decades to centuries
What It Predicts	Specific conditions: rain, temperature, wind	Average patterns: trends in temperature, rainfall, sea level
Key Data	Current observations from radar, satellites, stations	Historical records, ice cores, ocean temperatures, greenhouse gas levels
Accuracy	Very accurate for 1–3 days; decreases rapidly after 7 days	Shows reliable trends, but cannot predict the weather on a specific future day
Uses Probability?	Yes — "60% chance of rain tomorrow"	Yes — "very likely that average temperature will rise by 1–2°C by 2050"

Comparison of weather forecasting and climate projection

In high school and beyond, you will explore climate models that project how Earth's systems will change over decades. These models build on the same principles you are learning now: gather evidence, find patterns, use math, and express results with probability. The skills you develop now in analyzing weather data apply directly to understanding our changing planet.

ℹ

NGSS Connection

This lesson connects to MS-ESS2-5: Collect data to provide evidence for how the motions and complex interactions of air masses result in changes in weather conditions. It also connects to MS-ESS3-2 and the crosscutting concept of Cause and Effect, where scientists use probability to describe the likelihood of weather outcomes.

SECTION 9

Practice Problems

Test your understanding with these five problems. They start simple and get more challenging. Read each scenario carefully and think about the evidence before choosing your answer.

Question 1: CONCEPTUAL

A weather forecast says there is a 40% chance of rain today. What does this mean? A) It will rain for 40% of the day. B) 40% of the city will get rain. C) Out of many days with conditions like today, about 40 out of 100 would have rain. D) The meteorologist is 40% sure it will rain at your house.

Show answer

C) Out of many days with conditions like today, about 40 out of 100 would have rain. A probability of 40% means that when similar conditions occurred in the past or in model simulations, rain happened about 40% of the time. Option A is wrong because the percentage does not describe how long rain lasts. Option B is wrong because probability is not about the area affected. Option D is wrong because probability is not about personal certainty — it is based on data from many situations.

Question 2: BASIC CALCULATION

A student finds 25 past days that had similar conditions to today. On 15 of those days, it snowed the next day. What is the probability of snow tomorrow? A) 15% B) 25% C) 40% D) 60%

Show answer

D) 60% Probability = (15 ÷ 25) × 100% = 60%. Option A (15%) simply uses the number 15 as a percentage, ignoring the total. Option B (25%) uses the total number of days, not the fraction. Option C (40%) would be the probability that it does NOT snow (10 ÷ 25 = 40%). The correct answer uses the formula: favorable outcomes divided by total outcomes, times 100%.

Question 3: INTERMEDIATE

A meteorologist runs 30 ensemble computer model simulations for tomorrow. Of those, 24 predict thunderstorms, 4 predict light rain only, and 2 predict dry weather. Which forecast statement is best supported by this data? A) "It will definitely thunderstorm tomorrow." B) "There is an 80% chance of thunderstorms and a 20% chance of no storms." C) "There is about a 13% chance of light rain, an 80% chance of thunderstorms, and a 7% chance of dry weather." D) "Computer models cannot be trusted, so we should just wait and see."

Show answer

C) "There is about a 13% chance of light rain, an 80% chance of thunderstorms, and a 7% chance of dry weather." Thunderstorms: 24 ÷ 30 = 80%. Light rain: 4 ÷ 30 ≈ 13%. Dry: 2 ÷ 30 ≈ 7%. Option C gives the most complete and accurate picture. Option A is wrong because saying "definitely" ignores the 20% of models that predicted other outcomes. Option B groups light rain with dry weather incorrectly. Option D dismisses useful evidence — while no model is perfect, ensemble results are valuable evidence.

Question 4: APPLIED

Your school's outdoor field day is planned for tomorrow. The barometer has dropped 12 millibars in the last 6 hours, humidity has jumped to 92%, and winds shifted from the west to the south. Radar shows a line of storms 150 miles to the southwest, moving northeast at 30 mph. Based on this evidence, what should the principal be told? A) "The weather will be fine. The storms are far away." B) "There is a high probability of storms arriving within the next 5 hours. We should have an indoor backup plan." C) "It is impossible to predict what will happen, so we should just hope for the best." D) "The barometer reading does not matter for predicting rain."

Show answer

B) "There is a high probability of storms arriving within the next 5 hours. We should have an indoor backup plan." Multiple pieces of evidence point to incoming storms: rapidly falling air pressure, very high humidity, a wind shift (common before fronts), and radar showing storms moving toward the area. At 30 mph, a storm 150 miles away could arrive in about 5 hours (150 ÷ 30 = 5). Option A ignores the evidence. Option C is wrong because we have strong evidence to make a prediction. Option D is wrong because a falling barometer is one of the best indicators of approaching stormy weather.

Question 5: CRITICAL THINKING

Two cities, Springfield and Laketown, have identical temperature and air pressure readings today. However, Springfield is in a flat plains region and Laketown is next to a large lake at the base of a mountain. A forecaster gives Springfield a 50% chance of afternoon rain but gives Laketown a 75% chance. Why might the probabilities differ even though the temperature and pressure are the same? A) The forecaster made a mistake — identical conditions should give identical probabilities. B) Laketown's lake and mountain create local effects (lake breezes and updrafts) that increase the chance of rain beyond what temperature and pressure alone predict. C) Probability is just a random guess, so different numbers are expected. D) Air pressure does not matter for rain, only temperature does.

Show answer

B) Laketown's lake and mountain create local effects (lake breezes and updrafts) that increase the chance of rain beyond what temperature and pressure alone predict. Geography matters! A large lake adds moisture to the air, and mountains force air upward, cooling it and making clouds and rain more likely. These local factors are additional evidence that changes the probability. Option A is wrong because identical values of just two variables do not mean everything else is identical — geography, humidity, and wind patterns also matter. Option C is wrong because probability is based on evidence, not random guessing. Option D is wrong because air pressure is a very important weather variable. This shows the crosscutting concept of Systems and System Models — you need to consider the whole system, not just a few parts.

SUMMARY

Lesson Summary

Summary

Scientists predict short-term weather by collecting evidence from tools like thermometers, barometers, hygrometers, anemometers, radar, and satellites. They identify patterns in the data by comparing current conditions to past events. Ensemble computer models run many simulations to show a range of possible outcomes. Forecasts are expressed as probabilities — like "70% chance of rain" — because the atmosphere is a complex system and no prediction can be perfectly certain.

The key crosscutting concepts in this lesson are Cause and Effect (weather changes have identifiable causes like pressure drops and air mass interactions), Patterns (past weather data reveals repeating patterns that guide predictions), and Stability and Change (the atmosphere shows predictable large-scale patterns but can change in surprising ways). The science and engineering practice is analyzing and interpreting data to construct evidence-based predictions. Short-term forecasts (1–3 days) are far more accurate than long-term ones, and expressing uncertainty through probability helps everyone make better decisions.