Page 110 - EER-2-2

Explora: Environment
and Resource Climate suitability of AWD practice

Table 2. Summary of the potential percolation rates based on monthly. The temporal resolution was acceptable since
soil texture irrigation and drainage water management decisions are
considered weekly or fortnightly. An average precipitation
Soil Soil textural Potential percolation
texture ID classes (mm/day) data for 5 years from 2011 to 2020 in the dekad; thus, the
mean daily dekad in mm/day was acquired from the FAO
LB UB BS
1 Clay 1 5 3 WaPOR portal, developed remotely sensed data with open
access licenses for monitoring water productivity.
3 Sandy clay 3 9 6
Similarly, reference PET, ET, and dekad data from FAO-
4 Loam sandy 1 6 4 WaPOR v2 database was used in this study (Figure S3).
32
6 Clay Sandy loam 3 15 9 The potential ET (dekad, mm/day) was defined as the
7 Loam 2 6 4 summation of the canopy transpiration (T) and surface
9 Sandy loam 3 15 9 evaporation (E) based on grass, the reference crop, using
33
Note: LB, UB, and BS represent lower bound, upper bound, and basic the Penman-Monteith FAO method. Finally, the spatial
settings, respectively. resolution of both continuous and categorical raster
datasets was standardized to a 30 m × 30 m grid, using
type of well-watered crop, capillary rise, surface inflow nearest-neighbor resampling for continuous data and a
or drainage, lateral seepage, and potential percolation, all majority filter for categorical data.
units in mm/dekad, respectively.
In the paddy field, water loss through seepage from 3. Results
a rice field into the neighboring field is counterbalanced 3.1. Ecological niche model suitability distribution
by seepage inflow from neighboring fields. Consequently, of irrigated paddy rice
seepage water loss was ignored in the above equation. The Figure 3 shows the sensitivity and reliability of the model
capillary rise contribution of paddy rice crop water need to evaluate the suitability of AWD conditions. Evaluation
can be considered for aerobic conditions of unflooded of the potentially suitable location of flooded paddy fields
soils. Conversely, in waterlogged paddy conditions, the
capillary rise is prevented by percolation into the paddy in Eastern Uganda with the MaxEnt model depicts high-
rice rhizosphere continuum, which is often ignored in the performing metrics of AUC and percentage correctly
hydrological (water balance) model. classified (PCC) >92% and 90% on training data, respectively.
Likewise, the model shows high performance of AUC
In addition, capillary rise during the non-flooded with 92% and PCC >81% using the independent test data,
periods with AWD practice provides excess water for the indicating that the MaxEnt is reliable for suitability analysis
paddy rice in irrigation schemes with shallow groundwater of AWD practice in paddy fields. Other research studies
tables. However, capillary rise water input was excluded have shown the application and consistency of the MaxEnt
31
from the equation due to limited spatial datasets to avoid software for climate suitability of AWD irrigation practice
partiality in assessing the water balance in the paddy field in paddy fields. For example, research on mapping potential
(Figure S2) of the irrigation schemes with a low water locations for AWD practice in Burkina Faso highlighted that
table. The water balance equation was then modified based the MaxEnt model gave over 80% of AUC on the performing
on the above assumption. metrics. Generally, the sensitivity analysis was influenced
15
I + P = Potential ET + Pot pc (II) by the volume fraction of coarse fragments (CRFVOL),
cp
I + P -(Potential ET + Pot ) = 0 (III) organic carbon stock (OCS), exchangeable potassium
(EXK), available water (AW), precipitation of the warmest
pc
cp
AWD technique uses less water for paddy rice quarter (BIO18), and topographical wetness index (TWI).
31
cultivation compared to the flooding method (CF) and However, the suitability investigation of AWD technique in
is suitable for each period with a deficit water balance and paddy fields can be improved by combining two or more
unsuitable otherwise. 22 predictors, such as MaxEnt and Random Forest.
Water balance deficit if P -(Potential ET + Pot ) < 0 3.2. Model evaluation of environmental predictors’
cp
pc
(IV) importance
Water balance excess if P -(Potential ET + Pot ) > 0 (V) Figure 4 indicates the jackknife test of various
cp
pc
The computation of water balance components environmental predictors that influenced the suitability
(Figure S2) was defined from the dekadal (10-day) time of AWD irrigation conditions. The results from the
steps, covering 36 dekadal periods annually and three dekads MaxEnt model based on the jackknife test show that the

Volume 2 Issue 2 (2025) 6 doi: 10.36922/EER025040005