In this study, we attempt to identify dominant controls on watershed vulnerability to climate and land-use change independent of future projections of climate change. We then use classification and regression trees (CART) to identify the areas of climate and land use change that lead to different classes of the indicator. In this way, we identify critical thresholds for climate and land-use change that result in a vulnerable range of a flow indicator across 77 watersheds in the contiguous United States.
Finally, we use CART to estimate critical combinations of climate and land use change. We apply exploratory modeling to a large number of watersheds and quantify critical thresholds of climate and land use change for each. Quantifying the hydrologic response of a watershed to changing climate and land use is essential for the management of water-dependent ecological and economic systems in the region.
Exploratory modeling assesses the response of a watershed to a wide variety of artificially induced changes in climate and land use. Then the relative change in hydrological indicators of interest compared to their historical values is estimated across all climate and land use change scenarios. Finally, using appropriate search algorithms, the thresholds of change in climate and land use leading to vulnerability can be estimated.
Thus, the maximum tolerable change of climate or land use, beyond which the value of a hydrological indicator is classified as vulnerable, can be used as an indicator of the watershed's vulnerability to change.
Generating threshold projection as vulnerability maps for the United
Second, a decision maker is most likely interested in a watershed's vulnerability to transition to a predetermined streamflow regime, and not just its sensitivity to a unit change in climate or land use. We then determine the critical thresholds of climate and land use change for each indicator. We adopt the exploratory modeling framework proposed by [18] with modifications in sampling strategies of the climate and land use space and parameter uncertainty quantification (Figure 2.1a).
We begin by defining the feasible ranges of climate and land use change, shown by the cube on the left of Figure 2.1a. Climate scenarios are a combination of precipitation and temperature change scenarios, while land use change is represented as a parameter in the hydrological model used in the analysis. We apply exploratory modeling to a large number of watersheds and quantify critical thresholds of climate and land use change for each.
It provides a way to estimate the critical values of climate and land use changes that lead to vulnerability. For example, in Figure 2.1 an indicator classifies into four classes (each class represented by a color) and with the help of CART the climate and land use combinations that lead to vulnerability can be determined.
Hydrologic model
We selected these categories based on a review of the literature, wherever information on critical ranges of indicators was available. After this, classification and regression trees (CARTs) identify regions in the input space that lead to vulnerable classes.
Indicators of vulnerability
For example, on the assumption that mean annual streamflow is a proxy for water availability in the watershed, it is assumed to become vulnerable if its values fall below historically observed values. Reductions in mean annual runoff compared to historical values are classified as C1, C2, or C3, with each class spanning a series of consecutive 25% reductions. Successive relative increases are assigned to classes C1, C2 and C3, while decreases are assigned to C4 (Table 2.1).
For FPIFR and drought indicator, we use predefined class definitions from past literature. The drought indicator is classified into five classes C0–C4 based on streamflow drought index (SDI) values.
Threshold identification via classification and regression trees (CART) 9
Sampling of climate and land use scenarios
Precipitation scenarios are generated by changing the precipitation time series within -40% to +60% of its historical value, and temperature change scenarios are generated by adding 0◦C to 12◦C temperature increases to the historical time series. The precipitation and temperature change values are chosen so that they are wide enough to include the maximum change over a long period of time as reported in the fifth assessment report of the International Panel on Climate Change [ 47 ]. We vary the parameter representing land use between 0 and 1 (completely bare land to full vegetation cover).
First, a wide range of possible values for each parameter is established based on literature review. Finally, the behavioral range of parameters is achieved using NSE and percentage Bias criteria.
Identification of parameter ranges and behavioural parameter
We then generate 50000 sets of parameters using a Latin hypercube sampling method assuming a uniform distribution of parameters within a priori ranges. We further constrain these parameter sets by testing their ability to reproduce the magnitude and variability of the historically observed flux, quantified by Nash Sutcliffe efficiency (NSE) and volumetric bias. Parameter sets that produce an NSE greater than 0 and a percent volumetric deviation within 25% are accepted as behavioral.
After identifying the behavioral parameter sets, we select the top 50 sets based on NSE performance (or the number that produces a positive NSE if less than 50 sets produce positive NSE).
Identification of critical thresholds
Note that for watershed USGS gauge ID, we are unable to identify any parameter sets that meet the aforementioned conditions. Thus, we exclude this watershed from further analysis, reducing the total number of watersheds we analyze to 69. The threshold values are calculated from a weighted average of the threshold values on leaf (end) nodes leading to vulnerability classes.
Red color indicates the highest vulnerability class in the indicator (C3) and red dashed line represents the paths leading to C3 for this watershed. We also found that temperature appeared in CART output less frequently than precipitation and thus focus on precipitation changes to represent watershed climate change. This figure is the fraction of deep-rooted vegetation cover in the watershed above, signifying annual runoff transitions to a vulnerable regime.
Remember that the land use parameter varies from 0 to 1 (bare land to full coverage of deep-rooted vegetation). Thus, a land use threshold of 0.55 implies that if more than 55% of the watershed is covered by deep-rooted vegetation, the increased evapotranspiration is likely to reduce the average annual runoff by more than 50%, thereby moving it to the vulnerable class. It is important to emphasize here that this land-use change threshold applies only to the range of precipitation changes (15% to 25% . reductions) that lead to the node containing land-use change as a partitioning parameter.
Threshold mapping
Each CART output is the result of climate and land use change in the watershed in the presence of parametric uncertainties. Thus, we consider six controls: precipitation change, temperature change, land use change represented by a model parameter, followed by parameters related to snow, soil, and direction, respectively. Some watersheds have land-use change in their secondary divisions and some have it in their tertiary divisions, indicating that it is the second most important control for this indicator.
Precipitation is also a dominant control measure for this indicator, but here land use change emerges as a primary control for a significant number of water sheds. From their Circos diagrams, we find that the primary control on these indicators is precipitation change, followed by land use change. FPIFR appears to be more affected by land use change than the drought indicator as indicated by the greater number of watersheds showing land use change as a secondary indicator for FPIFR.
The controls are classified as – long term precipitation, land use parameter, long term temperature, soil parameters and path parameters. In this study, we attempted to identify the dominant controls on the sensitivity of watersheds to climate change and land-use change. The spatial variability of identified critical thresholds across the United States provides an understanding of watersheds that are sensitive to precipitation and land use change.
The vulnerability map can help water resources administration in planning that watershed that will become vulnerable, even minor change in precipitation or land use and that required more precise monitoring. On the other hand, the vulnerability map for land use change thresholds does not show the same pattern. The analysis of thresholds for land use change across the study watersheds shows that most watersheds will become vulnerable if the percentage of deep-rooted vegetation cover in a watershed exceeds 45%.
Finally, we use Circos plots for each indicator to visualize the dominant controls on the indicators' sensitivity to climate and land-use change. In addition to precipitation and land-use change, hydrologic model parameters that describe soils, flow characteristics, and snow formation and melting are important determinants of watershed susceptibility to flooding. Catchment hydrological response to climate and land-use change in tropical Africa: a case study of South-Central Ethiopia.
Impacts of climate change, population growth, land use change, and groundwater availability on water supply and demand in the contiguous United States. A vulnerability-based approach to identify adverse combinations of climate and land-use change for critical thresholds of hydrologic indicators: application to a watershed in Pennsylvania, USA.
