Obviously you will not be able to use observations to look at climate change (change over a century or longer). The best you could do is to look at climate variability. For example what is the change from year to year or from decade to decade. If you are interested in climate change the best you could do would be to use climate models projections. Then the challenge would be to downscale from the coarse resolution of climate models to the watershed scale. There are some climate products that are already down scaled, however, so I would check if they include your region of interest.
Adding up on Alberto's answer, you could also use dynamically downscaled projections of climate change. There is a number of coordinated projects such as CORDEX where such future scenarios can be obtained from.
Then you would need to make focus on each of your regions of interest and, in case you needed future streamflows within each of them, then you would need for example to adopt a hydrological model and force it with your new precip scenarios.
In addition to what Ramiro wrote, be sure to quantify the uncertainties in the projected climate models if used eventually. That gives a more robust result.
An alternative approach would be to try to identify proxy indicators (e.g. tree rings) of climate variables (e.g. rainfall, temperature, floods, etc.) and use them to look for hints of climate variability/change in the past. For example you can use the proxy data to extend the observational record backwards. Maybe somebody with expertise in this approach and familiar with existing global proxy datasets can share what is available (if any) for Iraq. Historical accounts of floods for example, if there were any, could also be used along with proxy indicators. Of course when using proxy data your uncertainties would be larger so the ideal situation would be to find proxy records that overlap the observational record for validation.