Page 113 - EER-2-2

Explora: Environment
and Resource Climate suitability of AWD practice

cation exchange capacity, and available water (WWP) MaxEnt and ENM are suitable tools for estimating
volumetric fraction. All of these influence irrigation and the suitability potential of paddy rice areas suitable for
soil properties for AWD applications. The differences in AWD practice. This is important for irrigation planning
the model’s accuracy were associated with the complexity and improving paddy rice water management and rice
and computation time during model fitting. The statistical productivity for food security in Uganda. However, future
robustness of the response curves is shown in Figure S4. studies should consider using ensemble modeling, which
The blue boundaries represent the 95% confidence intervals combines predictions from multiple machine learning
around the predicted relationships, implying a 95% chance models and is a powerful tool for improving the accuracy
that the true relation lies within these boundaries. The and robustness of the suitability predictions. Predictions
cloglog output identifies the probability of suitability and it using ensemble techniques such as bagging (Bootstrap
is plausible that rice growth will thrive in an area with that Aggregating) or boosting (AdaBoost, XGBoost) are more
34
corresponding environmental factor. accurate and stable.
3.5. Estimated percolation rates
3.4. Potential suitability for irrigatable paddy rice
cultivation by spatial prediction Table 3 shows the estimated percolation rate with
soil texture. Soil properties, including bulk density,
The expected appropriate spatial distribution was mineralogy, organic matter content, salt type and
predicted (Figure 6). The findings indicate that potentially concentration, and soil structure and texture, influenced
suitable locations are distributed close to the center south the percolation rate in paddy fields. The physical action
31
of the Eastern region of Uganda between the latitudes of puddling leads to the formation of the hardpan,
of 33.5°N and 34.4°N from the west and have sandy clay which changes the soil structure, although variation
loam soils. This was attributed to the high prevalence of of percolation rates remains in different soil textural
environmental predictors influencing water and land in classes. AWD suitability increases with an increase in
potential locations. Defining the potential locations for potential percolation, assuming that evapotranspiration
AWD was influenced by the effect of various percolation and precipitation are constant. Lower percolation values
rates defined in reasonable boundaries. One of the are less suitable for AWD, but suitability increases as the
drawbacks during modeling was the absence of accurate percolation rate increases. Table S4 shows the exact ratio
in situ information on percolation rates, which we suggest as a function of depth and total area. It was arrived at
field estimation in future research studies. by multiplying the corresponding ratios of the different

Figure 6. Predicted potentially suitable paddy rice locations for AWD irrigation practice using the MaxEnt model
Abbreviation: AWD: Alternate wetting and drying.

Volume 2 Issue 2 (2025) 9 doi: 10.36922/EER025040005