# Discussion

This chapter provides an analysis of the results obtained in the previous chapter. Section 4.1 
discusses the uncertainties of the hydrological model PCR-GlobWB. Section 4.2 compares the 
results obtained in Chapter 3 to other researches. Lastly, Section 4.3 discusses future research 
directions. 

## Uncertainties of PCR-GlobWB 

This research relies on a hydrological model rather than direct measurements, introducing 
potential for uncertainties in the results. These uncertainties arise from various factors, including  
input data, spatial scales and model structure.  

The input data consists out of a parameter set and a forcing dataset. For example, the ERA5 
datasets may contain errors in observed data or missing data, which could account for some 
extreme outliers. Similar concerns apply to the CMIP6 dataset, as inaccuracies and unavailable 
data can be the origin of uncertainties in the simulations. Additionally, the spatial scales of the 
datasets differ. ERA5 and CMIP6 operate on grids of varying resolution. The coarser the resolution 
of CMIP6, especially after rescaling to align with ERA5, can temper the precision of the results and 
potentially affecting their accuracy. 

The structure of the model can also be a source of uncertainties. The model incorporates 
many hydrological components, and not all of which vary over time. For instance, as mentioned in 
Section 2.2, the irrigation factor in PCR-GlobWB is constant over time. In reality, this factor 
fluctuates. Therefore, it is a static representation and it likely influences the simulation outcomes. 

The results of groundwater recharge simulation, presented Sections 3.1 and 3.2, possibly 
contain an uncertainty. The low groundwater recharge values suggest potential issues with either 
the model, the input data, or the interpretation of the simulated outputs. These values are near 
zero, and indicate an absence of groundwater recharge, which is inconsistent with reality. This 
suggests that the model’s outcome or the quality of the data may be flawed. It is important to 
acknowledge the model that is simulating future climate scenarios is not an exact reflection of 
reality.

## Comparing the Results with Other Researches

To evaluate the results of this research it is compared to a report of the USGS, researching the 
same region with comparable future climate scenarios. The United States Geological Survey 
published a report in 2020, containing research on the trends in recent historical and projected 
climate data from 2020 to 2100 for the Colorado River basin. The projected climate data evaluated 
97 Coupled Model Intercomparison Project phase 5 (CMIP5) projections and all Representative 
Concentration Pathways (RCP) from year 1951 to 2099. In this research, it is expected that the 
lower Colorado River basin will experience a slight decrease in precipitation, and an increase in 
temperature. The projected groundwater infiltration decreases rapidly, compared to the 
historical data.  

These findings align with the results generated by the PCR-GlobWB model, which similarly 
predict a decline in groundwater recharge under more extreme future climate scenarios. The 
alignment of these results reinforces the reliability of the PCR-GlobWB model.

## Future Research

To achieve more reliable and accurate outcomes in hydrological modelling, several factors are in 
need of further investigation. One approach is to apply and compare multiple hydrological models. 
By incorporating a range of models, researchers can identify similarities and differences in the 
outcomes, improving the accuracy of the results. 

Additionally, more focus on important hydrological factors could enhance the reliability. 
For example, analysing how temperature changes with each climate scenario, analysing shifts in 
precipitation patterns and evaluating changes in discharge. Understanding the hydrological cycle 
in more detail can give more insight into the effect on groundwater recharge. 

As mentioned in Section 2.5, this study simplified the groundwater flow by assuming it 
occurs in a single direction. While this simplifies the results, it does not accurately reflect reality. 
The groundwater flow is often a multiple directions and is influenced by several factors. 
Incorporating the realistic groundwater flow would yield more accurate groundwater recharge 
results. 

Lastly, to obtain the future climate simulations, three SSPs (SSP1-2.6, SSP2-4.5 and SSP5-8.5) were used. However, CMIP6 contains many more scenarios. Therefore, in future research, 
evaluating more scenarios can lead to a higher reliability.