等高線判讀

三件事就夠了：一，同一條線上高度相同；二，線越密坡越陡、越疏坡越緩；三，線彎成 V 字時尖端指向上游（河谷），凸出來朝低處時是稜線。這個工具讓你拖一條切線，把這三件事同時畫成剖面驗證一次。

因為河谷是凹進山體的。同一條等高線要繞進谷裡再繞出來，而谷的上游端地勢較高，所以那條線必須往上游方向凹進去才能維持同一高度。在科羅拉多大峽谷標本上，每一條支流都看得到。

等高距是相鄰兩條等高線的高度差。地形完全沒變，等高距變小就切得更密、線更多、細節更多；變大則線更少、只留大輪廓。工具上的 ×½ / ×1 / ×2 就是讓你在同一座山上直接看見這件事。

剖面的水平比例尺通常是公里、垂直是公尺，若照實畫，山會被壓成一條幾乎水平的線。所以縱向會被拉長，那個倍率就是垂直誇大。工具把它直接寫在讀數上：看到 ×4，代表畫面上的高度被拉長了四倍，所以看起來比真實陡很多 —— 真正的坡度請以旁邊那個度數為準，不要把度數乘以四。

兩者都是一圈包一圈的封閉曲線，形狀幾乎一樣。分辨方式有兩個：看標高數字往內是變大還是變小，或看示坡線（垂直於等高線的短線，一律由高處指向低處）。本工具標出計曲線的高程數字，但沒有繪製示坡線；火口湖與富士山兩個標本擺在一起比較最清楚。

是。每一格高程都來自公開的高程資料（以 NASA SRTM 為主，美國境內為 USGS 3DEP），經由 AWS 開放資料的 Tilezen / Mapzen terrain tiles 取得後烘焙成靜態檔隨頁面提供，沒有任何一格是手畫或程式生成的。等高線是在你的瀏覽器裡用 marching squares 從同一份高程網格算出來的 —— 所以等高線圖和剖面圖必然是同一塊地。

SRTM 原始取樣是 30 公尺一格，尖銳的山頂會被平均掉一點；本工具再依每個標本的框景重採樣到約 15～115 公尺一格，所以會比原始資料更平滑。例如玉山在這裡讀到約 3911 公尺，公布高程是 3952 公尺。這是開放高程資料的取樣極限，工具照實顯示，不做美化。

會，所以工具提供「線稿模式」。彩色的暈渲和分層設色很好看，但它會替你把地形讀完 —— 你可能一直在看陰影而不是看線，自己卻不知道。按圖例下方的「線稿模式（像考卷）」，暈渲與色帶會全部關掉，只剩黑線白紙，跟考卷上的圖一樣。等高線本身完全沒有改變，剖面也一模一樣：如果你在線稿模式下還讀得出來，那才是真的會了。

Three things carry most of it. Every point on one line is at the same height; tightly packed lines mean a steep slope and widely spaced lines a gentle one; and where a line bends into a V the point aims upstream, while a bulge toward lower ground is a ridge. This tool lets you draw a cutline and watch all three appear at once in the cross-section.

Because a valley is cut into the hillside. To stay at one height, the contour has to run up into the valley and back out, and the upstream end is the higher one — so the bend must point that way. Every tributary on the Grand Canyon specimen shows it.

It is the height difference between neighbouring lines. The ground does not change: a smaller interval slices more often, giving more lines and more detail; a larger one leaves only the broad shape. The ×½ / ×1 / ×2 control lets you see exactly that on one hill.

A profile is usually kilometres across and metres tall. Drawn true to scale, a mountain flattens into almost nothing — so the vertical axis is stretched, and that factor is the vertical exaggeration. This tool prints it: at ×4 the HEIGHT on screen is four times over-long, which is why the slope looks so severe. The real gradient is the figure in degrees beside it — do not multiply that by four.

Both are nested closed loops and look nearly the same. Two things separate them: whether the labelled heights rise or fall inward, and hachures — short ticks drawn across a contour, always pointing downhill. This tool labels its index contours but does not draw hachures; compare Crater Lake against Mt. Fuji and the difference is immediate.

Yes. Every cell comes from open elevation data (NASA SRTM, with USGS 3DEP inside the USA), fetched as Tilezen / Mapzen terrain tiles via AWS Open Data and baked into static files served with the page. Nothing is hand-drawn or procedurally generated, and the contours are computed in your browser by marching squares from the very same grid the profile samples — so the map and the cross-section are necessarily the same ground.

SRTM samples every 30 metres and a sharp summit gets averaged down; each specimen is then resampled to roughly 15–115 m per cell to frame its landform, so it is smoother still. Yushan reads about 3911 m here against a published 3952 m. That is the sampling limit of open elevation data; the tool shows it honestly rather than rounding up.

It can, which is why there is a line-art mode. Hillshading and an elevation ramp look good, but they read the terrain for you — you may be judging relief from the shadows without realising you are not reading the contours at all. The “Line-art mode (exam style)” switch under the legend turns off the shading and the colour ramp, leaving black lines on paper exactly like a printed exam figure. The contours themselves do not change and neither does the cross-section: if you can still read it that way, you have actually learned it.