Here's a natural bound on the integrand: .

But to be more precise, you can find the extreme points on the function by setting the partial derivatives to zero: and . Then you need to calculate the Hessian matrix to check if the points found are max, min or saddle points. Also compare these points to the boundary values.