First, note that even in that data set there's an unmistakable upward trend. Second, cherry-picking 1998 as one's starting point is.....well....cherry picking, something one doesn't do in science.
Finally (and most importantly), there's a bit more to this than just the graph you copied. See, satellites don't directly measure temperatures like thermometers do. Instead they infer temperatures from proxy data. So when it comes time to analyze the data and infer temperatures, different groups use different methods, which produces different results. Not only that, but not all the satellites that collect this data do so via the same instruments (plus, instruments do decline over time and satellites drift and experience orbital decay). Put all those factors together and you end up with high degrees of discrepancies between final data sets and analyses.
Attempts have been made to correct for all these factors, and some agreement between observed and inferred data has been made, but discrepancies remain. Specifically, the data set for the graph you posted, i.e., the UAH data, is the set that shows the the lowest temps for the lower troposphere, whereas the GISS and Hadley Center set both agree with each other and show warmer temps than the UAH data. Also, the UAH data is the least consistent with direct measurements taken via weather balloons. So basically, focusing on the UAH data set to the exclusion of others is once again, cherry picking.