In which case you have to study, kohei
There are several ways to skin that cat, but none of them 100% reliable.
1. Using the radar you have. careful use of radar will give you speed estimates. it is easier if it is a convoy/task force since the radar display will give you a very good idea of the course of the convoy and this helps a lot.
It will take much more than 3 minutes however, but you can shadow the convoy beyond visual range and try to match course and speed with it until you are confident that you have it right. This can take hours, but equally it can take less time, and besides you can use time compression for this method without loss of accuracy.
2.Build one of these babies, the ISWAS
http://www.subsim.com/radioroom/show...ighlight=iswas
once mastered you can find the speed of the target on an intercept course by setting a collision course.
3. Estimate the length of the target, and pointing your bow at the target, measure the time it takes for the target to completely cross your bow. Then convert this measurement of target length/time into knots. The esxact formula you use depends on the units for your estimate of target length. Eg. if you estimate in meters, then meters/secs x1.8 = knots (x2 is good enough in practice). if you estimate in feet, then (feet/secs)/1.7 = knots
4. practice observing bow wake - you can find this helpful, although i do not use this as the primary method, it is useful for seeing if the ship has changed speed. Low speeds are easier to judge with this method, i find.
5. Stadimeter observations, and lots of them, and incorporate the AOB observation too in order to help you with the line of best fit, since range estimates vary so much. if you can rely on your ability to judge AOB by eye to help you with the target course it much reduces the effect of range error.
6. Mysteries of Target Motion Analysis!
and there are more!