Converting 2W + TLE data to CZML felt unrealistic, reaching 600 + MB, and loading the transformed files reported errors. I don’t know if there’s any other way, Refer to the CelesTrak Orbit Visualization, I wonder how they solved this whole data loading problem
help me
Would streaming the czml incrementally help resolve the issue?
The TLE data of CelesTrak were all obtained in the form of TXT. Do you mean that after the completion of TXT acquisition, CZML is gradually converted and then append? The final file is still very large, I really don’t understand, I hope to make it clear, thank you
Does anyone know how to implement this 
Well, looking at the source code of the site, there’s two text files that are loaded; one for debris, and one for satellites. My guess is that they read and parse the text files, and create Cesium entities using the primitives API for speed, and then apply callbackProperties() to their positions to calculate their orbit. I didn’t check if they use CZML, not sure there’s enough initial data (no historic data, only orbital data). That’s what it looked like without digging too far into it.
Is there anything beyond that you’re wondering about?
Cheers,
Alex
Thank you for your reply. I’m trying to use Satellite.js to achieve this. The current problem is to calculate the real-time position of each fragment in real time. 
Hi！Using satellitejs can solve this problem
This is a tle data
VANGUARD 1
1 00005U 58002B 22048.60497048 .00000295 00000-0 35416-3 0 9992
2 00005 34.2513 148.6678 1844076 218.1613 127.3116 10.84862082271524
First, you need to use satellite to calculate the coordinates you need to obtain
What I calculate here is the coordinates in 24 hours
const SampledPosition = new Cesium.SampledPositionProperty(Cesium.ReferenceFrame.INERTIAL)
const second = parseInt(Cesium.JulianDate.secondsDifference(stopTime, startTime) / 300) + 5
const start = new Date(startTime)
const satrec = satellite.twoline2satrec(tle1, tle2)
let time_step = 0
let current_time = null
for (let i = 0; i < second; i++) {
current_time = new Date(start.getTime() + time_step)
const positionAndVelocity = satellite.propagate(satrec, current_time)
const positionEci = positionAndVelocity.position
const position = Cesium.Cartesian3.fromElements(positionEci.x*1000, positionEci.y*1000, positionEci.z*1000)
if (position) {
SampledPosition.addSample(Cesium.JulianDate.fromDate(current_time), position)
time_step += 300 * 1000
}
const interpolationOptions = {
interpolationDegree: 5,
interpolationAlgorithm: Cesium.LagrangePolynomialApproximation
}
SampledPosition.setInterpolationOptions(interpolationOptions)
}
Sampledposition is the coordinate sampling set we need
Next, we need to create an entity
const period = 1440 / tle[2].substring(52, 63)
const minutesperiod = period * 60
const number_of_full_orbits = Math.floor(minutes / period)
const left_over_minutes = minutes % period
const leadTimecomposite = new Cesium.CompositeProperty()
const trailTimecomposite = new Cesium.CompositeProperty()
let clonestart = Cesium.JulianDate.clone(start)
for (let j = 0; j < number_of_full_orbits + 1; j++) {
let endtime = new Cesium.JulianDate()
if (j === 0) {
Cesium.JulianDate.addMinutes(clonestart, left_over_minutes, endtime)
} else {
Cesium.JulianDate.addMinutes(clonestart, period, endtime)
}
const leadTimeTimeInterval = new Cesium.TimeInterval({
start: clonestart,
stop: endtime,
})
const trailTimeTimeInterval = new Cesium.TimeInterval({
start: clonestart,
stop: endtime,
})
leadTimeTimeInterval.data = new Cesium.SampledProperty(Number)
trailTimeTimeInterval.data = new Cesium.SampledProperty(Number)
leadTimeTimeInterval.data.addSamplesPackedArray([0, minutesperiod, minutesperiod, 0], clonestart)
trailTimeTimeInterval.data.addSamplesPackedArray([0, 0, minutesperiod, minutesperiod], clonestart)
leadTimecomposite.intervals.addInterval(leadTimeTimeInterval)
trailTimecomposite.intervals.addInterval(trailTimeTimeInterval)
clonestart = endtime
}
let intervals = new Cesium.TimeIntervalCollection()
intervals.addInterval(new Cesium.TimeInterval({
start: start,
stop: stop,
}))
const showpath = new Cesium.TimeIntervalCollectionProperty();
showpath.intervals.addInterval(new Cesium.TimeInterval({
start: start,
stop: stop,
data: true
}))
const width = new Cesium.CompositeProperty(6)
const resolution = new Cesium.CompositeProperty(120)
viewer.entities.add({
availability: intervals,
path: new Cesium.PathGraphics({
material: Cesium.Color.AQUAMARINE,
leadTime: leadTimecomposite,
trailTime: trailTimecomposite,
width: width,
resolution: resolution,
show: showpath
}),
point:{
pixelSize: 8,
color: new Cesium.Color.fromCssColorString(color)
},
orientation: new Cesium.VelocityOrientationProperty(SampledPosition),
position: SampledPosition,
})
This entity is the satellite and orbit we need
The above code is transformed from tle data to czml data according to Python code
I hope these can help you solve the problem 
3 Likes
Wow, you are great. You solved a problem that had been bothering me for a long time. Thank you very muc！！ 
Hello~ I’m here. I use primitive to draw points, and now i want to draw line, how to set position to the line? I want to draw an ellipse by it , but i don’t know how to describe it. a polyline or a ellipse?
Hello, thank you very much for your guidance, at present I can achieve more than 20,000 satellites to display. But if the timeline is extended to several months or even years with 20,000 satellites running, then your code will be severely timed out, do you have a solution for this? Thank you very much!
Using this calculation method can help you solve the problem.
This project is being continued as a commercial offering at SpaceAware.io. If you are curious about our progress, check out my profile on LinkedIn.
We used the open source Vallado C++ code available at Celestrak.org, and compiled it to WebAssembly.
Our plan is to let subscribers embed our version of Cesium with the built-in propagator on their own website, and when we are able, open source the WebAssembly propagation compiler pipeline and invite collaboration.
Until then, we would appreciate not reverse engineering our source and posting pieces of it online. The results will be inaccurate unless you reverse engineer the entire app, and currently the app / wasm binaries are proprietary and unlicensed for any use, commercial or private.