One run只算一次
A single run sounds very sure of itself. Is it?
只算一次的預報,聽起來很有把握。真的是這樣嗎?
Earth and Weather · Unit 2 · Lesson 10 · 地球與天氣 單元二 第十課
The forecast says there is a 70% chance of rain tomorrow. Where does that number come from, and what does it really mean? In a 3D model, send a band of rain toward a small town and run the forecast once, then ten times. Watch the ten answers drift apart as you look further into the future. Then learn the exact words that forecasters in Taiwan use for the sky.
氣象預報說明天的降雨機率是 70%。這個數字是怎麼來的?它到底是什麼意思?在 3D 模型裡,讓一條雨帶移向一個小鎮,先把預報算一次,再算十次。看得越遠,十次的答案就分得越開。最後,學會台灣的預報員描述天空時用的那幾個詞。
To forecast the weather, people first measure what the air is doing right now, all over the world. Then computers use the rules of nature to work out what the air will do next, step by step. But nobody can measure the air perfectly, and small mistakes grow with time. So a forecast for tomorrow can be quite sure, a forecast for next week is less sure, and an honest forecast tells you how sure it is.
要預報天氣,人們先量出全世界的空氣「現在」是什麼狀況,再讓電腦依照自然的規則,一步一步算出空氣接下來會怎麼變。可是沒有人能把空氣量得完全準確,而小小的誤差會隨著時間越變越大。所以明天的預報可以相當有把握,下星期的預報就沒那麼有把握;一份誠實的預報,會告訴你它有多少把握。
3D Model · 3D 模型
Drag to turn the model, and scroll or pinch to zoom. A band of rain is moving toward a town on a small island. Move the hours and watch when it arrives. The six boxes are the forecast for the town, one box for every 12 hours. Now choose Ten runs. Each run starts with a speed that is only a little different. Do the ten bands stay together? Try a band that is half a day away, and then one that is two days away.
拖曳可以旋轉模型,滾輪或雙指可以縮放。一條雨帶正移向小島上的一個小鎮。撥動時間,看它什麼時候到。那六個格子就是小鎮的預報,每 12 小時一格。接著選「算十次」:每一次計算,雨帶的速度都只差一點點。十條雨帶會一直走在一起嗎?先試試半天後就到的雨帶,再試試兩天後才到的。
Drag to turn · 拖曳旋轉 Scroll or pinch to zoom · 滾輪/雙指縮放
This 3D model needs WebGL, which this browser does not support. The reading and the cards below still explain everything.
這個瀏覽器不支援 WebGL,無法顯示 3D 模型;下方的課文與卡片一樣能看懂。
A teaching model, and every number in it is an example. The island, the town, the speed of the band, and the small differences between the ten runs were all chosen to show the idea. Real forecast centers do run their computer models many times with slightly different starting points, but the models are far more complicated than a band moving in a straight line, and forecasters also use their own judgment before they announce a chance of rain.
這是教學用的模型,裡面每一個數字都是示例。小島、小鎮、雨帶的速度,還有十次計算之間的小差別,都是為了說明原理而設定的。真的預報中心確實會用稍微不同的起始資料,把電腦模式跑很多次;不過真的模式比「一條直線前進的雨帶」複雜得多,而且預報員在發布降雨機率之前,還會加上自己的研判。
Model drawn by My Culture Connect as a teaching diagram. The ideas follow the Central Weather Administration's weather knowledge pages and the other sources listed at the bottom of this page. · 模型由人師教育協會自繪示意;原理依中央氣象署「氣象常識」與本頁最下方列出的其他出處。
Reading · 英文閱讀
Every evening, the weather report tells you about a day that has not happened yet. It says how hot tomorrow will be, and whether you will need an umbrella. How can anyone know the future? Nobody can know it for certain. But the air follows rules, and that makes a good guess possible.
A forecast begins with measuring. All day and all night, weather stations on the ground measure the temperature, the air pressure, the wind, and the rain. Balloons carry small instruments high into the sky, and hundreds of them are sent up around the world every day. Radar sees where rain is falling, and satellites look down at the clouds from space. Together they give a picture of the air right now.
Next comes the computing. The air follows the rules of nature: warm air rises, wind blows from high pressure to low pressure, and water vapor turns into cloud. Scientists have written these rules as mathematics. A very powerful computer starts with the picture of the air right now and works out what it will be like a few minutes later, and then a few minutes after that, again and again, until it reaches tomorrow and the days after.
But there is a problem. Nobody can measure the air perfectly. There are not enough instruments, and each one is a little bit wrong. These small mistakes do not stay small. They grow as the computer works its way into the future. So a forecast for tomorrow is usually good, and a forecast for next week is much less sure. Scientists think that about two weeks is the limit.
Forecasters have a clever way to deal with this. They run the computer many times, and each time they change the starting picture just a little. If almost every run brings rain, rain is very likely. If only a few runs bring rain, it is less likely. In the end, the forecaster studies all the information and gives the chance of rain as a number.
Now you can read that number. In Taiwan, a 70% chance of rain for Taipei means there are seven chances in ten that rain will fall in Taipei during that period. It does not mean that it will rain for 70% of the day or on 70% of the city. And if it stays dry on a 70% day, the forecast was not wrong. Three times in ten, that is exactly what should happen.
Questions · 閱讀理解
What are the two main steps in making a weather forecast?
Why is a forecast for next week less sure than a forecast for tomorrow?
What does a 70% chance of rain mean?
Words & Phrases · 生字及片語
Quick Check · 小測驗
1. What is the first step in making a forecast?
2. Which of these looks down at the clouds from space?
3. Why does a forecast become less sure the further ahead it looks?
4. The chance of rain is 70%. What does that mean?
5. The forecast said 30%, and it rained. Was the forecast wrong?
Read the Forecast · 讀懂氣象報告
Forecasters in Taiwan do not choose these words by feeling. They imagine the whole sky cut into ten equal parts and count how many parts are covered by cloud. Move the slider and see which word they use.
台灣的預報員不是憑感覺選這幾個詞的。他們把整個天空想成十等分,再數一數有幾分被雲遮住。撥動滑桿,看看他們會用哪一個詞。
This follows the Central Weather Administration. Thin cloud and thick cloud both count. When a forecast joins two words, such as sunny at times cloudy (晴時多雲), the first word is the one expected for at least half of the period. The English words are our translations.分法依中央氣象署。不管是薄而透光的雲,還是濃厚的雲,都要算進去。預報把兩個詞連在一起的時候,例如「晴時多雲」,前面那個詞是預計會占一半以上時間的天空狀況。英文是我們的翻譯。
Cloud cover · 雲量
Try It in the Model · 在模型裡試試
Compare the highest number in the six boxes for the near band and for the far band. That difference is why a forecast for tomorrow is surer than a forecast for the day after.
比一比:近的雨帶和遠的雨帶,六個格子裡最高的數字各是多少?這個差別,就是明天的預報比後天的預報更有把握的原因。
A single run sounds very sure of itself. Is it?
只算一次的預報,聽起來很有把握。真的是這樣嗎?
The ten bands are still close together. All ten agree: 100%.
十條雨帶還靠得很近。十次全都同意:100%。
Now the bands have spread out. Some arrive early, and some arrive late.
這一次雨帶散開了。有的早到,有的晚到。
The chance of rain is shared out over three periods. No period reaches 100%.
降雨的機率分散在三個時段裡,沒有任何一個時段是 100%。
Go Further · 延伸閱讀
Forecasts by the Numbers · 預報的數字
From the first idea to the forecast on your phone.
從最早的想法,到你手機裡的預報。
The Central Weather Administration gives the chance of rain for each area four times a day: at 5 and 11 in the morning, and at 5 and 11 in the evening.中央氣象署一天發布 4 次各地的降雨機率:5 時、11 時、17 時和 23 時。
The daily forecasts cover Taiwan, Kinmen, and Matsu in 22 areas, each with its weather, its high and low temperatures, and its chance of rain.每天的天氣預報把臺灣及金馬分成 22 區,各有天氣、最高最低氣溫和降雨機率。
The English scientist Lewis Fry Richardson published the idea of forecasting the weather by calculation, long before there were computers fast enough to do it.英國科學家理查森發表了「用計算來預報天氣」的構想;那時候還沒有夠快的電腦可以做這件事。
An early computer called ENIAC made the first computer weather forecast. It needed almost 24 hours to work out a forecast for the next 24 hours.早期的電腦 ENIAC 做出了第一份電腦天氣預報;它花了將近 24 小時,才算出未來 24 小時的預報。
Around the world, hundreds of balloons carry instruments into the sky every day to measure the air high above the ground.全世界每天有好幾百顆氣球把儀器帶上天,去量高空的空氣。
Scientists think this is about as far ahead as the weather can be forecast accurately, even with perfect measurements and a perfect computer.科學家認為,就算有完美的測量和完美的電腦,天氣能準確預報的極限也大約只有這麼遠。
Forecast Stories · 預報的故事
How forecasting grew up, and one word from the winter forecast.
天氣預報是怎麼長大的,還有冬天預報裡的一個詞。
1922
Lewis Fry Richardson believed that the weather could be worked out with mathematics. But there were no computers. So he imagined a huge round hall, like a theater, filled with people. Each person would do the sums for one small part of the world and pass the answer to the next. It was only a dream, but the method he described is the one that computers use today.
理查森相信,天氣是可以用數學算出來的。可是那時候沒有電腦。於是他想像了一個像劇院一樣的巨大圓形大廳,裡面坐滿了人;每個人負責計算世界上的一小塊地方,再把答案傳給旁邊的人。這只是一個夢想,但他描述的方法,就是今天電腦在用的方法。
1950
In 1950, an early computer named ENIAC made the first weather forecast by computer. The forecast was for the next 24 hours, and the machine needed almost 24 hours to finish it. By the time the answer came out, the weather had already happened! But it worked, and Richardson, who was still alive, called it an enormous scientific advance.
1950 年,一部叫做 ENIAC 的早期電腦,做出了第一份用電腦算出來的天氣預報。它預報的是未來 24 小時,而機器本身就花了將近 24 小時才算完。答案出來的時候,那天的天氣都已經發生過了!不過這個方法是行得通的。當時還在世的理查森說,這是科學上的巨大進步。
Chaos · 混沌
The scientist Edward Lorenz found that a tiny change at the start of a weather calculation could lead to a completely different result later. To explain it, he used a picture that became famous: a butterfly flapping its wings could, weeks later, change where and when a storm forms. He did not mean that butterflies cause storms. He meant that we can never measure the air well enough to be sure about the far future.
科學家羅倫茲發現,天氣的計算裡,起點只要有一點點不同,後來的結果就可能完全不一樣。為了說明這件事,他用了一個後來很有名的比喻:一隻蝴蝶拍動翅膀,幾個星期之後,可能改變一場風暴形成的時間和地點。他的意思不是蝴蝶會造成風暴,而是我們永遠沒辦法把空氣量得夠準,準到能確定很遠以後的事。
Taiwan · 台灣
Winter forecasts in Taiwan use three phrases, and each one has a rule. If the lowest temperature in Taipei is expected to fall to 10°C or below, it is a cold wave (寒流). If it is expected to fall to 12°C or below, the forecast says a strong continental cold air mass. Between 12°C and 14°C, it says a continental cold air mass. Above that, it says only that the northeast monsoon is growing stronger.
台灣冬天的預報有三種說法,每一種都有標準。預測台北的日最低溫會降到 10°C 或以下,就叫「寒流」;預測會降到 12°C 或以下,說的是「強烈大陸冷氣團」;在 12°C 到 14°C 之間,是「大陸冷氣團」;如果不會降到 14°C,就只說「東北季風增強」。
Myth vs. Fact · 常見迷思
✗ Myth · 迷思A 70% chance of rain means it will rain for 70% of the day.降雨機率 70%,就是一天裡有 70% 的時間在下雨。
✓ Fact · 事實The number has nothing to do with how long it rains or how much of the area gets wet. It means seven chances in ten that rain will fall there in that period.這個數字和下多久、下多大的範圍都沒有關係。它的意思是:那個時段裡,那個地方有七成的機會出現降雨。
✗ Myth · 迷思The forecast said 20%, and it rained, so the forecast was wrong.預報說 20%,結果下雨了,所以預報不準。
✓ Fact · 事實A 20% chance means that it should rain about two times in ten. This was one of those times.20% 的意思是十次裡面大約有兩次會下雨。這一次剛好就是其中一次。
✗ Myth · 迷思With a fast enough computer, we could forecast the weather for any day next year.只要電腦夠快,明年任何一天的天氣都可以預報。
✓ Fact · 事實The limit is not the speed of the computer. Small mistakes in measuring the air grow with time, and about two weeks is as far as an accurate forecast can go.極限不在電腦的速度。量空氣時的小誤差會隨時間變大,準確的預報大約只能做到兩個星期。
✗ Myth · 迷思Forecasters decide by looking out of the window.預報員是看看窗外的天空來決定的。
✓ Fact · 事實They use measurements from stations, balloons, radar, and satellites, and the results of computer models, and then they add their own judgment.他們用的是氣象站、氣球、雷達和衛星的觀測資料,加上電腦模式算出來的結果,最後再加上自己的研判。
Remember It · 記憶口訣
Stations, balloons, radar, satellites.
氣象站、氣球、雷達、衛星。
Step by step, with the rules of nature.
照著自然的規則,一步一步算。
So the far future is less sure.
所以越遠越沒把握。
70% means seven chances in ten.
70% 就是十次裡有七次。
Classroom Activity 1 · 課堂活動一
You need · 準備材料
Steps · 步驟
You cannot judge a forecast by one day. A good forecaster is one whose 70% days are rainy about seven times in ten. Ten days is only a start, so keep going if you can.
只看一天,沒辦法判斷預報準不準。好的預報,是「說 70% 的日子裡,十次大約有七次真的下雨」。十天只是開始,可以的話就繼續記下去。
Classroom Activity 2 · 課堂活動二
You need · 準備材料
Steps · 步驟
Two people looking at the same sky often choose numbers that differ by one or two. That is a small example of why no measurement of the air is perfect.
兩個人看同一片天空,常常會差個一兩分。這就是一個小小的例子:對空氣的測量,沒有一種是完美的。
Sources · 資料出處
Facts and numbers on this page were checked against these sources (October 2026). · 本頁的事實與數字依下列資料查證(2026 年 10 月)。