It is more complex than you might like to think - in a simple world, the rating would fall to zero at 70C, as no additional heating would be permissible as it would already be at max temp, and at the reference temperature the rating would be 100% of the book value. As heat  goes as I2 R one might expect a parabolic (x= y2 type) curve to link the two, or at least nearly, as a small correction resistance does change a bit with temp.

Actually if you plot it out, that is almost what the tables suggest - in fact it is a little way out - at 50 degrees the max heating to permitted to reach 70C is half that at 30C, so you may expect 70.7 % of the current rating, and when you are 3/4 f the way from refernce temp to maximum already - 60C, then the dissipation should be 1/4 so the current rating halved.
But this only works where the cooling is by conduction through contact with solids, and heat flow is in linear proportional to temp difference. If cooling is by convection however, then the air speed changes as things warm up, so cooling improves disproportionately well for increasing temperature difference. So use the tables for conduction cooling, but take great care with baskets and  so forth, where the aim is to get a decent temperature gradient and to drive  convection of the cooling gas.

Mike

It is more complex than you might like to think - in a simple world, the rating would fall to zero at 70C, as no additional heating would be permissible as it would already be at max temp, and at the reference temperature the rating would be 100% of the book value. As heat  goes as I2 R one might expect a parabolic (x= y2 type) curve to link the two, or at least nearly, as a small correction resistance does change a bit with temp.

Actually if you plot it out, that is almost what the tables suggest - in fact it is a little way out - at 50 degrees the max heating to permitted to reach 70C is half that at 30C, so you may expect 70.7 % of the current rating, and when you are 3/4 f the way from refernce temp to maximum already - 60C, then the dissipation should be 1/4 so the current rating halved.
But this only works where the cooling is by conduction through contact with solids, and heat flow is in linear proportional to temp difference. If cooling is by convection however, then the air speed changes as things warm up, so cooling improves disproportionately well for increasing temperature difference. So use the tables for conduction cooling, but take great care with baskets and  so forth, where the aim is to get a decent temperature gradient and to drive  convection of the cooling gas.

Mike