Activity 2 Implementation Protocol_v5.0

Goal and Objectives

This activity aims to evaluate the effects of land cover changes on hydrological processes, including water flow and quality, using process-based hydrological models The assessment will focus on nested sites within the Mae Chaem, Ping, Chao Phraya, and Greater Mekong regions in mainland Southeast Asia, as illustrated in Figure 1.

The comparison of physical modeling exercises utilizing real datasets aims to achieve several key objectives Researchers seek to gain insights into specific questions that will enhance understanding in this field.

1) What is the quantitative impact on the range of ‘watershed function’ indicators of the historical land use change between ‘natural vegetation’ and ‘current land use pattern’?

2) What degradation/recovery can be expected for a number of ‘plausible’ scenarios?

Higher resolution models can reveal the significance of landscape patterns in relation to land use cover data and their impact on ecosystem functions, raising the question of whether spatial planning plays a crucial role Preliminary findings indicate that the segregation of forests and agriculture is more beneficial for biodiversity conservation, while integration promotes better watershed function This study will examine the trade-offs between these approaches across various population densities.

This initiative aims to influence the policy agenda regarding biodiversity-related land use changes and their effects on hydrological processes, including sedimentation and landslides It will develop guidelines based on watershed scale, land cover, and topography The accompanying policy brief series will provide essential recommendations for policymakers The available policy options range from broad zonation prohibiting agriculture above a certain altitude to more nuanced land use zoning and specific interventions within categories, such as promoting shade coffee with a litter layer over sun coffee with bare soil, as demonstrated in Sumberjaya, Sumatra, Indonesia.

The analysis of systems and models will enhance our understanding of how land use changes, engineering practices, and the spatial distribution of human activities contribute to the risk of damage A comprehensive evaluation of land use will be conducted to assess its overall impact.

Deforestation and land use zoning significantly impact spatial landscape organization and require targeted management practices within key land use classes Implementing strategies such as enhancing litter layers can improve water infiltration, while sediment filters can effectively reduce overland soil flows These practices contribute to integrated solutions for sustainable land management.

And will focus the more detailed models on a comparison of two contrasting sites:

Figure 1 Nesting of study watersheds in mainland Southeast Asia

Global biodiversity conservation focuses on preserving unique ecosystems and identifying 'hot spots' where local conservation efforts can significantly impact global biodiversity In contrast, understanding watershed functions relies on analyzing the cumulative effects of rainfall on the water balance.

The 'meso' scale of landscape organization serves as a crucial intersection for two types of environmental service functions: biodiversity and watershed management Initial findings indicate that biodiversity thrives with spatial segregation, whereas watershed functions benefit from an integrated approach However, there is significant potential to balance these two functions by focusing on riparian zone vegetation, which is beneficial for both biodiversity and watershed perspectives.

National Conser- vation Strategies Segregate

Human activities significantly affect watershed functions through a combination of factors, including deforestation, engineering projects that alter drainage systems, and the settlement patterns of communities Attributing these impacts to just one of these causes oversimplifies the complex interactions at play.

Transport capacity less than required => flooding

Soil weight exceeds ancho- ring =>landslide starts

Mass wasting reaches stream & modifies channel behaviour

Soil weight exceeds ancho- ring =>landslide starts

Mass wasting reaches stream & modifies channel behaviour

Damage (people, agriculture, infrastructure

Damage (people, agriculture, infrastructure Engineering Engineering

This section sets out the implementation plan for these nested hydrological modeling exercises, as well as a description of the models and scenarios being employed.

Roadmap for Activity 2

Table 1 outlines the roadmap for Activity 2, detailing key aspects such as the model's scope, dataframe structure, and spatial resolution It also covers climatology, temporal resolution, the specific model utilized, and the functions being modeled Additionally, the table addresses land cover scenarios, the processes involved—including parameterization, validation, and sensitivity tests—as well as the reporting and analysis of model runs, including overlays.

The models and scenarios are described in more detail below.

Scope of model, dataframe, spatial resolution

Process, including paramete rization validation sensitivity tests

Reporting and analysis of model runs

WBM (see Activity 1aiv.2) total yield, high flows, low flows

Preindustrial, Contemporary, loss of high biodiverse areas.

See details in Activity 1aiv.2

WBM total yield, high flows, low flows

VIC total yield, high flows, low flows

Transition matrix between classified landcover at 1 km resolution, aggregated to 10 km Landcover time series based on LAI extracted from MODIS

Calibrated at 25Km a.Total yield for major urban centers (Laos, Mekong, Cambodia); and coastal zones (Mekong only). b.Seasonal variability of total flow. c.Duration of storm events.

12/99; aggregated to 4 hour time step

DHSVM total yield, high flows, low flows

Current, historical; irrigation scheme model for Mae Chaem developed based on hydrological models

Initial calibration at 150m; see DHSVM roadmap a.Total yield for major urban centers: town of Mae Chaem. b.Seasonal variability of total flow. c.Duration of storm events

GenRiver and SpatRain total yield, high flows, low flows, buffering

Scenarios from Chiang Mai University sub- contract

Calibrated to local rainfall and riverflow data.

See GenRiver and SpatRain roadmap

1998 GenRiver and SpatRain total yield, high flows, low flows, buffering preindustrial: forest; historical:

1980-200; loss of high BD areas: remove ridge of forest’ extensification: all shade coffe; intensification: all sun coffe

Compared with WaNulCL AS; validated based on frequency distributio n of flows.

See GenRiver and SpatRain roadmap

FALLOW examines the dynamics of soil fertility and land productivity at the plot level, assesses the household economy, and informs strategic decisions regarding land use and labor allocation It also focuses on the implementation of these land use strategies and labor distribution, while analyzing land use and cover changes through ecological succession and growth processes.

(6) human population dynamics; (7) market; (8) fire escape

(9) food security; (10) watershed functions: water balance (water yield, baseflow), sediment loss, soil physical quality dynamics; (11) biodiversity;

Shifting cultivation Contemporary: monoculture plantations (coffee & clove).

Loss of high BD areas: Shifting cultivation at overpopulated conditions.

Intensification: monoculture plantations (coffee and clove).

Sensitivity analysis of net sediment loss from applying 5

‘rules’ for allocating 25% forest cover across the landscape; see Fallow roadmap

Hydrological Models

In Activity 2, various hydrological models are simultaneously applied across a nested set of sub-catchments to examine their complementarity and consistency This project focuses on models that differ in scale and the representation of physical processes, specifically analyzing how changes in soil properties over time impact infiltration We will evaluate the underlying assumptions of the hydrological models used at diverse spatial and temporal scales.

Table 2 summarizes the differences among the hydrological models utilized in Activity 2 and the synoptic Water Balance Model (WBM) employed in Activity 1 Additionally, the table classifies the hydrological models for the BNPP project based on five levels of complexity.

Although the results from these models are not definitive, they represent the most advanced analysis available, utilizing expert modeling techniques and comprehensive data These simulations play a crucial role in understanding the significant impacts of climate and terrain on hydrological functions, allowing for the isolation of the lesser influences of land cover, which is often the target of various interventions Furthermore, these models are vital for accurately predicting both the volume and timing of river flow.

1 will normally include rainfall, energy balance (potential evapotranspiration), soil storage capacity and landscape (routing times in the stream network) properties,

Land cover properties, obtained through remote sensing or other methods, play a crucial role in understanding rainfall interception and actual evapotranspiration These properties are influenced by soil water storage and respond dynamically to variations in the area fractions of different land cover types.

Land cover significantly impacts infiltration capacity and necessitates rainfall intensity data to accurately predict surface runoff in sloped areas Additionally, it is essential to consider how changes in land use affect the peak flow to base flow ratio.

Some studies may specifically address overland flows, the incorporation of soil particles into these flows, and the deposition of soil particles in filtration zones, highlighting the importance of the landscape's spatial organization.

A limited number of studies recognize the alteration in soil properties influencing infiltration as a dynamic process, emphasizing the importance of the timeline of land use changes rather than merely focusing on the final distribution of various land cover types.

All models will be used for a comparison of ‘natural vegetation’ (baseline) versus

This article focuses on the existing land use patterns in relation to the current climate It aims to create location-specific scenarios that predict potential land use changes and assess their impact on hydrological functions Additional details on scenario development can be found in activity c.

Models, when properly executed, provide a clear representation of the implications stemming from various assumptions While no model can be deemed entirely correct or incorrect, the adequacy of these assumptions in accurately depicting observable phenomena is crucial Given that different modelers may interpret the same assumptions differently or adopt varying assumptions, it is important to compare different model implementations, even when they address similar hypotheses In Activity 2, we will examine several models originally designed for distinct circumstances, temporal frameworks, and spatial contexts To effectively present the results of these models, we must first clarify their underlying assumptions, as well as their similarities and differences.

See Appendix 6 for four steps for meso (& micro) scale models to clarify land use change effects on watershed functions.

Model WBM VIC DHSVM GenRiver Fallow WaNuLCAS

Time step Month Day 4 hours Day (+ rainfall intensity)

Questions at the level of model structure and assumptions

The common classification of watershed management into 'upland' and 'lowland' categories often overlooks the critical role of the 'transmission zone.' However, activities and changes occurring in the riverbed significantly influence the overall functionality and health of rivers.

Interactions between ‘lowland’ and ‘upland’ are driven by current ‘perceptions’ of the

‘hydro logic’, not by hard facts and measurable results The ‘emotional basis’ level needs to be balanced by better understanding of patterns, processes, threats and opportunities for better governance.

Roughly, three intervention points can be identified in the chain of events by which

‘land use change’ (as the more sophisticated version of ‘deforestation’), leads to human exposure to risks:

• effects of land use pattern on the way rainfall is converted to streamflow

• hydrological properties of the stream network that relate inputs to the streams to water levels and qualities downstream (‘engineering solutions’ can modulate this step)

• location of human stakeholders relative to the streambed (not living on the wrong place on the wrong time makes a big difference)

• Where the ‘knee jerk’ reaction in many watershed function cases is to look for the forests and land use change as explanations, we need to find a balance between

Downstream stakeholders: irrigation, hydro-electric, domestic/indus- trial water use, houses on floodplains/river bed, fisheries, coral reefs

Riverflow (total water yield, Qmax/Qmin, quality)

Compared to forest: total water yield Qmax/Qmin Baseflow ~ Quality

Upland land use Upland land use Upland land use

Transmission zone land use (esp riparian)

Compared to natural river flow:

Total water yield (extractions) Qmax/Qmin (loss of buffer) Baseflow (extractions) Quality (pollution, loss of filter)

Infiltration, Channels Filter functions Infiltration, Channels Filter functions

‘land use based’, ‘engineering’ and ‘locational/zoning’ solutions to current problems.

Table 3 Measurability of land use impacts by basin size (Kiersch and Tognetti, 2002) x = Measurable impact, - = No measurable impact.

The lack of ‘hard’ data for most watershed functions beyond 10 km 2 may be largely due to inadequacies in study design, but may also reflect the importance of other

‘drivers’ – a prime candidate is the ‘spatial variability in rainfall’ mechanism that the GenRiver + SpatRain model illuminates.

Land use impact Impact Basin site (km 2 )

Land use change scenarios

The project aims to develop four contrasting socio-economic development scenarios that will detail assumptions about land-use changes, particularly focusing on their impact on hydrology This approach allows for an exploration of how different land-use patterns can affect hydrology across various scales By generating plausible future landscapes, these scenarios provide a framework for interpreting the implications of modeled outcomes, with their primary value lying in the comparative analysis rather than the probability of each individual scenario.

1 Story line describing in broad terms how development proceeds at the regional scale;

2 A set of explicit rules (a qualitative model) for evolving landscapes under each of the scenarios

The project aims to create land-use and land-cover scenario maps for the Ping River Basin and Mae Chaem in the Mekong River Basin, covering ten-year intervals from 2000 to 2050 These maps will be developed primarily under a separate sub-project grant from SEA-START RC at Chulalongkorn University, focusing on climate change and water-related sector changes as part of the AIACC initiative Consequently, the funds requested in this proposal will be allocated mainly for work at the other two scales.

The appendices of the Scenarios report will include descriptions of methodologies and relevant statistical relationships This report aims to serve as a foundation for individual research publications or as a component of publications that utilize the scenarios to evaluate hydrological impacts.

We will also explore land use patterns derived for specific rule sets, such as forests on steepest slopes, which depends strongly on resolution of the digital elevation model.

The scenarios will provide rule sets for:

 Fraction forest and other land cover types.

 Water withdrawals (for Urban + Irrigation use) to be represented as multipliers on current use.

 Spatial organization (at least: riparian, midslope, ridges and tops).

Table 4 Main steps in Preparation of the Scenarios narratives and landscape maps.

Analysis of Historical Transitions to Derive Key Relationships.

Analysis of highly aggregated trends between land-use and other development statistics for the Ping River Basin (Chiang Mai Province/Lamphun as focus).

Spatial analysis of historical transitions for much more limited set of variables – using disaggregation algorithms and multi-scale regressions (e.g As in CLUE framework);

Detailed analysis of LU changes in Mae Chaem to develop rules for different kinds of upland areas.

Preparation of set of key constraint layers (elevation, river-network, main cities, infrastructure) to use in “evolving” the landscape Scenario Set.

Liaison with users on primary axes to explore “scenario space” based on preliminary set.

Finalize the framing scenarios with storylines for land-use evolution models.

Develop conversion rules and variants for each scenario (i.e Land-use change models).

Evolve landscapes under each scenario with snapshots.

Derive various landscape metrics at different scales to describe landscape structures - revisions or variants of scenarios.

Write research paper describing scenarios and expected consequences for future landcover.

Document datasets and models for sharing in subsequent work with hydrological models.

The proposed scenario highlights key axes of variation in both landscape and social organization, emphasizing the significance of trade levels and the spatial extent of institutional arrangements as vital examples of social connectivity.

The proposed framework for developing socio-economic and landscape evolution scenarios features inner boxes representing landscape scenarios and outer boxes depicting social development contexts, organized along axes of overall tree abundance and spatial distribution "More patchy" indicates that forest patches are larger and more distinct from other land uses, while "Less patchy" refers to smaller, less distinct tree patches Additionally, a second set of axes illustrating variations in socio-economic development can be superimposed on this framework to highlight the differences among the four scenarios concerning social organization.

3 APPLICATION OF THE MODELS IN BNPP ACTIVITIES

This section of the Protocol reviews data requirements and structure of each of the models concerned

UNIVERSITY OF WASHINGTON PROJECT ACTIVITIES

Terms of Reference for U of W

This section describes specific sub-activities by University of Washington (UW) The

UW, as one of the partners working on Activity 2, is tasked with 5 sub-activities:

Activity 3.1 Assemble the dataframe required for the application of the VIC hydrology model to the Mekong river drainage basin (and Mae Chaem/Ping, and conceivably to Chao Phraya).

Activity 3.2 Assemble the dataframe required for the model application in Northern

Thailand, focusing on VIC and exploring the advantages and feasibility of DHSVM.

Activity 3.3 Provide model simulations on both dataframes (using the respective model) over a standard climatology and either a "standard" rainfall, or a "typical actual rainfall record" (i.e. last 10 years).

Activity 3.4 Produce modeled output with landcover alterations of forested landscapes at various degrees and in different upland and floodplain configurations so as to simulate a range of landuse change scenarios.

Activity 3.5 Report the results of all model simulations in terms of:

• total yield by time at locations upstream from major urban centers and at the coastal zone,

• seasonal variability of total flow related to seasonality of the simulated rainfall data,

• duration of storm events effects on stage height at location upstream from major urban centers.

The specific UW deliverables include:

This study examines the impact of sub-basin size on modeled hydrograph structures by analyzing various results across different basin scales The objective is to uncover the relationship between basin size and the resulting variations in hydrograph characteristics.

• Collaborative work on manuscripts and policy briefs corresponding to Activity 3.2

The tasks will focus on the broader implications of the BNPP program, examining far-field effects in relation to local impacts such as land use changes, alterations in precipitation patterns, and associated hazards, while ultimately addressing the implications for biodiversity and watershed functions.

The UW partners include: ICRAF/Bogor and Chiang Mai, and Chulalongkorn

University; in consultation with the World Bank and ICRAF/ASB.

Strategy and Implementation

Assemble the dataframe required for the application of the VIC hydrology

The primary focus of this study is the Mekong River drainage basin, which includes the Mae Chaem and Ping rivers, and potentially extends to the Chao Phraya River.

The GTOPO30 digital elevation model (DEM) will be utilized in the DHSVM modeling environment for the primary analysis of the Mae Chaem, while the Ping will be incorporated as part of the broader Chao Phraya study, despite the unavailability of higher resolution data for a more detailed examination.

The project aims to develop multiple dataframes to achieve our objectives, focusing on updating our Mekong coverage with the most recent and high-resolution data available A key challenge in this process is the limitations in spatial, temporal, and content information of the core datasets For regional-scale analysis, basic datasets like land cover are sourced from satellites with a spatial resolution of 1 km, which are typically static and compiled from various satellite data streams It is important to note that performing time series analysis using a static land cover dataset imposes the constraint that actual land use remains unchanged.

Long-term climatology datasets, often generated monthly, require interpolation for models operating on daily time steps This process involves making a series of decisions, each carrying significant implications Therefore, our approach is to systematically evaluate the datasets and their associated trade-offs.

We initially developed 1 km resolution data for the Mekong but conducted our first model runs at a resolution of 25 km Currently, we are preparing to execute the VIC model at a 10 km resolution, as 1 km was deemed too fine for the theoretical capabilities of the VIC framework The data sources and methodologies utilized to create the VIC dataframe are detailed in Appendix 3 - Table 1.

A separate dataframe for the Mae Chaem region is currently being developed by our Thai colleagues, while the dataframe intended for the Chao Phraya basin is only partially completed, with most efforts focused on the Mekong and Mae Chaem, making its timely completion uncertain According to Appendix 3 - Table 2, the VIC/Mekong dataframe is nearing completion, with the remaining tasks primarily involving the development of a time series land cover dataset and reaching consensus on scenario datasets It is important to note that the scenarios will primarily be used to analyze land use change consequences rather than focusing on the "most likely" scenarios The outputs from this project will be adequate to drive further analysis and decision-making.

VIC for the analyses of Activities 3.2.3, 3.2.4, and 3.2.5.

Assemble the dataframe required for the model application in Northern Thailand, focusing on VIC and exploring the advantages and feasibility of DHSVM

Thailand, focusing on VIC and exploring the advantages and feasibility of DHSVM.

The initial decision for the Mae Chaem was whether to utilize VIC or DHSVM DHSVM requires quite a complex dataframe to operate.

The available data for Mae Chaem is only slightly adequate for implementing DHSVM; however, we believe that the improved capabilities of this model justify its use in addressing the relevant issues.

(considerable) effort of launching DHSVM a Scope: The scope is the Mae Chaem basin. b Data Inputs and Methods:

While much of the data for the Mekong can be assembled from multiple “international” sources, the higher resolution data for the Mae

Chaem can only be sourced locally, presenting challenges related to data sets in Thai and various projections, formats, and intended applications The collaboration between David Thomas and the local ICRAF team was crucial for this effort Currently, the DHSVM/Mae Chaem dataframe is nearly complete, as detailed in Appendix 3.

Table 4) d Linkages, Milestones, Outputs: The product of this work will be the dataframe sufficient to drive DHSVM.

Provide model simulations on both dataframes (using the respective model)

over a standard climatology and either a "standard" rainfall, or a "typical actual rainfall record" (i.e last 10 years).

The work under this activity will provide the basic model development and refinement, with which to evaluate (1) basic dynamics of the respective basins, and

(2) the “best” combination of model setup and dataframe to analyze the landuse scenarios, and focus on near- versus far-field effects a Scope:

The scope is the Mekong Basin (per Activity

3.1) and the Mae Chaem basin (per Activity

3.2) Only as time permits, will we include the

Chao Phraya. b Data Inputs: The respective dataframes

(Activity 3.1, Activity 3.2) provide the data inputs. c Methods: We are using 2 separate models.

Mekong For the Mekong basin, we are utilizing a model specifically designed for analysis of water (and energy) in a regional- scale, large river, the Variable Infiltration

The VIC (Variable Infiltration Capacity) model, developed by Liang et al in 1994, is a physically based model that effectively parameterizes small-scale processes for application in large river basins It operates at spatial resolutions ranging from 1/8 degree to coarser scales, such as the 2-degree global application discussed by Nijssen et al in 2002 Notable applications of the VIC model include significant continental river basins, such as the Columbia, as demonstrated by Nijssen et al in 1997.

Arkansas-Red (Abdulla et al., 1996), and the Upper Mississippi (Cherkauer and Lettenmaier, 1999), among other rivers VIC has also been applied to the entire area of China (Su and Xie, 2003).

A detailed description of the VIC model can be found in Liang et al (1994, 1996 and

1998) Briefly, the model has parameterizations to represent the vertical exchange of moisture and energy between the vegetation canopy and the atmosphere, similar in

The model stands out from other Soil-Vegetation-Atmosphere Transfer Schemes (SVATS) by effectively incorporating the spatial variability of soil, topography, and vegetation, which significantly influences runoff generation It primarily operates under the saturation excess mechanism, a valid approach commonly used in humid environments.

The VIC model incorporates a nonlinear deep soil drainage parameterization to simulate "slow" or baseflow runoff response It is integrated with a streamflow routing scheme that effectively transports runoff from each grid cell through a designated channel network Notably, this routing model does not consider channel losses, extractions, diversions, or reservoir operations, which are instead addressed in the water management model, as detailed by Lohmann et al (1996; 1998).

Mae Chaem Our original commitment was to also use VIC for the Mae Chem work.

We were concerned that the physics represented in the Variable Infiltration Capacity (VIC) model may not accurately reflect the steep topography and finer-scale issues of the Mae Chaem basin VIC is designed as a larger, regional-scale model, making its application to smaller-scale basins and sub-basins, like Mae Chaem, questionable To address these challenges effectively, we committed to using our high-resolution hydrologic model, the Distributed model.

Hydrology Soil Vegetation Model (DHSVM,

DHSVM, unlike VIC, is designed for small to moderate drainage areas (typically under 1000 km²) and utilizes digital topographic data to accurately depict water movement across surfaces and through subsurfaces While both models employ the saturation excess mechanism for runoff generation, DHSVM uniquely incorporates topographic influences, such as perched water tables, and provides detailed representations of vegetation properties, including root depth, and soil characteristics on a pixel-by-pixel level With a grid resolution of 30-150 m, DHSVM offers significantly higher detail than VIC, but its application is limited to smaller catchments due to computational demands and data constraints Initial experiments have shown that while both models exhibit similar macroscale performance in capturing seasonal runoff fluctuations, they diverge in predicted runoff and surface fluxes at shorter time scales Our current focus has been on developing dataframes and establishing the models, a task complicated by various technical challenges that necessitated an upgrade to “Generation 2” of DHSVM.

Table 5 Target VIC/Mekong and DHSVM/Mae Chaem model runs Results of model simulations will include

The total yield over time at various locations upstream from major urban centers along the Mekong River includes significant areas such as Chiang Saen in Thailand, Luang Prabang, Vientiane, Paksane, Thakhek, Savannakhet, Pakse in Laos, and Stung Treng, Kratie, and Phnom Penh in Cambodia Additionally, the yield is also assessed in the coastal zone of the Mekong, with a specific mention of the town of Mae Chaem.

(2) seasonal variability of total flow related to seasonality of the simulated rainfall data,

(3) duration of storm events, effects on stage height at location upstream from major urban centers.

Purpose Res Veg Soil Status

25 km IGBP’94 FAO Running, results being compiled Effect of template scaling 10 km IGBP’94 FAO 09/01

Effect of veg change 10 km MODIS’02 FAO 09/10

Combined resolution 1 km MODIS’02 MRC 09/15

Time Series 10 km Δ LAI TS MRC/FAO 09/30

Purpose Res Veg Soil Status

150 m LDD LDD Initial runs starting

Scenario Analyses 150 m Scenarios LDD In development e Output-analyses: Our charge is to produce model simulations using a “standard climatology and either a ‘standard’ rainfall, or a ‘typical actual rainfall record’ (i.e last 10 years).”

The standard climatology for the VIC is based on a 20-year climate record and corresponding discharge data from the Mekong region To assess the overall model and data trade-offs and identify the most reliable product, we will conduct a series of model runs that systematically explore the assumptions involved in model scaling and data sets These results will help us determine the optimal combination of parameters for the scenarios outlined in Activity 3.2.4.

The climatological data for Mae Chaem is significantly more limited compared to that of the Mekong, particularly regarding elevation effects Consequently, we will utilize the two-year dataset created under Activity 3.2.2 for our analysis.

Produce modeled output with landcover alterations of forested landscapes

various degrees and in different upland and floodplain configuration so as to simulate a range of landuse change scenarios.

The focus of the work here is to take the basic model analyses (climatology, scale, soils) from Activity 3.2.3, and run with different vegetation covers.

The project focuses on the Mekong Basin and the Mae Chaem Basin, with the possibility of including the Chao Phraya Basin as time allows The primary data inputs for this initiative are outlined in Activity 3.2.3 The methodology involves substituting various land cover scenarios to analyze the impact on these regions.

Activities 3.2.1 and 3.2.2) into the basic modeling framework (Activity 3.2.3). d Status: The timetable for Activity 3.2.4 is summarized in Table 5 We are starting to develop the scenarios themselves. e Output-analyses: The output will be model runs under the different landcover scenarios, which will (presumably) give us insight into how landcover change would effect water flow, both near-field and far-field.

Report the results of all model simulations in terms of

(1) total yield by time at locations upstream from major urban centers and at the coastal zone,

(3) duration of storm events effects on stage height at location upstream from major urban centers.

This activity constitutes the primary “paper and report” writing part of the work. a Scope: The scope is the Mekong and Mae Chaem basins (and conceivably the Chao

The analyses of the Mekong River are limited to the area up to Phnom Penh, as the river subsequently divides into various channels before entering the Mekong Delta, which introduces complexities not covered in this study While efforts will be made to address sediment issues, establishing a comprehensive connection between sediment mobilization and hydrology models may exceed the current scope of this research, particularly beyond basic equations like the Universal Soil Loss Equation Data inputs for this report will include the outputs generated from the modeling process.

Activities 3.2.3 and 3.2.4 (which in turn are predicated on the outputs of Activities 3.2.1 and 3.2.2). c Methods and Output Analyses: The work here is “paper-writing.” We envision the following papers (See Papers 8 to 10). d Status: We have a substantial draft of the time series paper, but only at the coarse resolution.

References

• Abdulla, F., D.P Lettenmaier, E.F Wood, and J.A Smith (1996): Application of a macroscale hydrologic model to estimate the water balance of the Arkansas-Red river basin J Geophysical Research 101 (D3): 744-749.

• Cherkauer, K.A., and D.P Lettenmaier (1999): Hydrologic effects of frozen soils in the Mississippi river basin J Geophysical Research 104 (D16): 19599-19610.

• Kuraji, K and K Punyatrong Altitudinal Increase in Rainfall in the Mae Chaem Watershed, Thailand 2001 Journal of the Meteorological Society of Japan

• Liang, X., D.P Lettenmaier, E.F Wood, and S.J Burges (1994): A simple hydrologically based model of land surface water and energy fluxes for General Circulation Models J Geophysical Research 99: 14415-14428.

• Nijssen, B., D.P Lettenmaier, X Liang, S.W Wetzel, and E.F Wood (1997): Streamflow simulation for continental-scale river basins Water Resources

• Nijssen, B., R Schnur, and D.P Lettenmaier (2001a): Global retrospective estimation of soil moisture using the Variable Infiltration Capacity land surface model, 1980-1993 J Climate 14: 1790-1808.

• Nijssen, B., G.M O’Donnell, D.P Lettenmaier, D Lohmann, and E.F Wood (2001b): Predicting the discharge of global rivers J Climate 14: 3307-3323.

• Su, F.G., and Z.H Xie (2003): The model for assessing effects of climate change on runoff of China Progress in Natural Science.

• James A Tindall, J.A., J Kunkel, and D.E Anderson Unsaturated Zone

Hydrology for Scientists and Engineers Prentice Hall New Jersey, 1999.

• Wigmosta, M.S., L Vail, and D P Lettenmaier, 1994: A distributed hydrology- vegetation model for complex terrain, Wat Resour Res., 30, 1665-1679.

Data requirements and data availability for VIC, DHSVM and WBM

Soils x,y,soilcode grid 5-min yes (local better)

LAI x,y,LAI grid 15-min yes

Climate x,y,maxT,minT,precip,wind asci or grid daily need better

Discharge x,y,discharge asciss daily need better

Landcover x,y,vegcode grid 30 m yes yes

Soil depth x,y,depth grid 30 m yes

Soils x,y,sand,silt,clay grid 30 m yes

Climate x,y,T,wind, precip,humidity, radiation ascii or grid sub-daily recent 14 stations Game_T data set,

Discharge x,y, discharge ascii 30 m since 1954, but with gaps Channel routing width, depth, friction

Stream network stream locations gis line file 30 m

Road network Road locations gis line file 30 m

Culvert locations culvert locations _ gis point file 30 m not an issue not an issue

10 & 50 km WWF, IGBP (IFPRI is working)

Soils x,y, texture grid 10 & 50 km yes (FAO/IGPB)

LAI x,y,LAI grid 10 & 50 km yes

Climate x, y, maxT, minT, precip, wind, vapor pressure grid monthly yes

River network x, y, flow direction grid

10 & 50 km, monthly yesDischarge x, y, discharge gis point file monthly yes

Variable Infiltration Capacity Model (VIC) applied to the Mekong Basin

Lead

Collaborator

Scope, dataframe, spatial resolution

The primary focus is the Mekong basin itself, as derived explicitly from the GTOPO30 digital elevation model (DEM). land use cover

Topographic data were taken from 30-arcsecond GTOPO30 Digital Elevation Model (DEM), obtained from the U.S Geological Survey, from their web site http://edcdaac.usgs.gov/gtopo30/gtopo30.html stream network

(1) Complete transference to VIC scheme

(2) Extract LAI, derive other parameters

Network (10 km) Aggregate 1 km to

Two sources of soil type data are used, to obtain the dominant soil type in each grid cell: (see Appendix-Table 1 for detailed information)

- Soils#1: 5-minute FAO/UNESCO digital soil map of the world (FAO, 1995)

- Soils#2: 1:50k high resolution data from the MRC streamflow data See Appendix 3 – Table 1 dams NA

Climatology

The analysis incorporates various variables and sources, which may be real or simulated, as detailed in Appendix 3 – Table 1 It focuses on spatiotemporal resolution with a daily time step and spatial scales of 1 km, 10 km, and 25 km Additionally, time series data is provided in Appendix 3 – Table 1, offering further insights into the study's parameters.

Machinery

The Variable Infiltration Capacity (VIC) model, developed by Liang et al in 1994, is a physically based hydrological model that parameterizes small-scale processes for effective use in large river basins It operates at various spatial resolutions, typically ranging from 1/8 degree latitude by longitude to coarser scales, such as the 2-degree global application highlighted by Nijssen et al This flexibility allows for the integration of diverse surface forcings, including precipitation and temperature data.

Since its inception in 2002, the Variable Infiltration Capacity (VIC) model has been utilized in significant continental river basins, including the Columbia, Arkansas-Red, and Upper Mississippi rivers Notably, it has also been applied to the entire region of China, showcasing its versatility and effectiveness in hydrological modeling.

Soils Remaining work (if any) Target Date FAO (8 km,

FAO (1 km) Disaggregate 8 km to 1 km

Gridded Surface Climatology Remaining work (if any) Target

A detailed description of the VIC model can be found in Liang et al (1994, 1996 and

The model, developed in 1998, features parameterizations that illustrate the vertical exchange of moisture and energy between the vegetation canopy and the atmosphere, akin to other Soil-Vegetation-Atmosphere Transfer Schemes (SVATS) Its primary distinction lies in its ability to account for spatial variability in soil, topography, and vegetation, and how these factors influence runoff generation, predominantly through the saturation excess mechanism, which is a valid assumption in humid environments.

The VIC model incorporates a nonlinear deep soil drainage parameterization to simulate slow or baseflow runoff responses It is linked to a streamflow routing scheme that effectively transports runoff from each grid cell through a designated channel network However, this routing model does not consider channel losses, extractions, diversions, or reservoir operations, which are instead addressed in the water management model For a comprehensive understanding of the routing model, refer to the detailed descriptions provided by Lohmann et al (1996; 1998).

Functions modeled

Total yield, high flows, low flows.

Land cover scenarios

The VIC's standard climatology is based on a comprehensive 20-year climate record and corresponding discharge data To evaluate the trade-offs between the model and data, we will perform a series of model runs that systematically explore the assumptions involved in model scaling and data sets This process will allow us to identify the most reliable combination of parameters for conducting the scenarios outlined in Activity 3.2.4.

The main focus beyond basic aggregation tasks is to develop the time series land cover dataset and finalize the scenario datasets It is important to emphasize that the scenarios are intended to analyze the impacts of land use changes rather than to predict the "most likely" scenarios.

Our objective is to create a transition matrix that represents the changes in classified landcover types by designating "change" to specific cells This approach will analyze change patterns, transitioning from a random spatial distribution within each landcover category while maintaining consistent quantities.

The analysis focuses on the proportion of total changes observed in a specific type of transition, comparing it to transitions in adjacent types It emphasizes the significance of identifying changes exclusively in upland undisturbed areas before examining the lowland regions.

Resolve process: transition matrices on

Validation

25 Km IGBP' 94 FAO Running, results being compiled

Reporting and analysis of model runs (Reporting of direct hydrological flows)

Results of model simulations will include: (see Appendix-Table 5)

The total yield over time at various locations upstream from significant urban centers along the Mekong River includes key sites such as Chiang Saen in Thailand and several cities in Laos, namely Luang Prabang, Vientiane, Paksane, Thakhek, Savannakhet, and Pakse, as well as Stung Treng.

(Cambodia), Kratie (Cambodia), and Phnom Penh (Cambodia)); and at the coastal zone (Mekong only),

Milestones

0, 1 st , 2 nd and final manuscript.

Notes, questions, comments

References

VIC references See: http://www.hydro.washington.edu/Lettenmaier/Models/VIC/#references.

Distributed Hydrology Soil Vegetation Model (DHSVM) applied to the Mae

Lead

Collaborator

Scope, dataframe, spatial resolution

Domain Mae Chaem basin land use cover

Base land cover map from the Land Development Division (LDD), Ministry of Agriculture, Thailand Landuse 1:50000; 1989; Originally 30 m and aggregated to final 150 m

DEM ICRAF DEM (topo-map derived) 30 meter, UTM

Originally 30 m and aggregated to final 150 m stream network Derived from DEM soils

Soil data very sparse, and restricted to lowlands

- Soil mapping unit Land Development Division (LDD), Ministry of Agriculture (LDD), with 62-group soil description, slope, moisture, permeability)

- Map of soil site location from individual projects (ICRAF) streamflow data Mean daily discharge in m 3 /s at gage P.14 (Ob Luang,

Chiang Mai), Nam Mae Mu and Nam Mae Suk dams NA

Climatology

In this study, we will utilize the Distributed Hydrology Soil Vegetation Model (DHSVM) as our primary tool Initially, we planned to incorporate the Variable Infiltration Capacity (VIC) model for the Mae Chaem project; however, we determined that VIC's representation of physics might not adequately capture the steep topography and intricate details of the Mae Chaem basin.

The VIC model is designed for larger regional applications, raising concerns about its effectiveness in smaller basins like Mae Chaem To address these challenges, we have committed to employing our high-resolution hydrologic model, the Distributed model, for a more accurate analysis at this scale.

Hydrology Soil Vegetation Model (DHSVM, Wigmosta et al, 1994) Unlike VIC, DHSVM is intended for application to small to moderate (typically less than about

1000 km 2 ) drainage areas, over which digital topographic data allows explicit representation of the mechanisms by which water travels over the surface and through the subsurface

Similar to the VIC model, this approach illustrates runoff generation through the saturation excess mechanism However, it distinguishes itself by explicitly accounting for topographic influences, such as the development of perched water tables, on runoff generation Additionally, it incorporates factors like incident solar radiation and net radiation, along with various related variables and sources.

Air temperature( o C) Hourly data from

Kogma 3/98 – 12/99 and correct for height- dependent with the temperature lapse rate Wind speed (m/s) Hourly data from

Kogma 3/98 – 12/99 Assume 2 m/s in the missing data

Relative humidity (%) Hourly data from

Hourly data from Kogma, 3/98 – 12/99 Incoming longwave radiation, (W/m 2 )

Hourly data from Kogma 3/98 – 12/99 Precipitation (m/timestep) 1 Hourly data 6/98 –

2 Originally daily data and disaggregated into hourly 3/98 – 5/98 and 12/98 – 12/99 sources (real or simulated?) Real. spatiotemporal resolution, original and interpolated

40 explicitly represents the vegetation and its properties (like root depth), as well as soil properties, on a pixel-by-pixel basis

The model grid resolution typically ranges from 30 to 150 meters, significantly higher than that of VIC Due to substantial computational demands and data constraints, DHSVM is limited to smaller catchments Limited experiments have been conducted to assess DHSVM's sensitivity to factors such as vegetation and vegetation changes (Van Shaar et al., 2002) While both models exhibit similar macroscale performance in capturing seasonal runoff fluctuations, they demonstrate notable differences in predicted runoff and surface fluxes, particularly over shorter time scales.

Total yield, high flows, low flows

Scenario Analyses 150 m Scenarios LDD In progress

3.5.9 Reporting and analysis of model runs (reporting of direct hydrological flows)

Results of model simulations will include: (see Appendix 3 - Table 5)

The total yield over time at locations upstream from significant urban centers includes key areas such as Chiang Saen in Thailand, and several cities in Laos, namely Luang Prabang, Vientiane, Paksane, Thakhek, Savannakhet, and Pakse, as well as Stung Treng.

Scenarios Under development (using similar protocols of transition matrices as Table 1)

See status/target of activities above

See section 3.3 Data availability and requirements.

DHSVM references See: http://www.hydro.washington.edu/Lettenmaier/Models/DHSVM/publications.htm

Functions modeled

Total yield, high flows, low flows.

Scenario Analyses 150 m Scenarios LDD In progress

Validation

Reporting and analysis of model runs (reporting of direct hydrological flows)

Results of model simulations will include: (see Appendix 3 - Table 5)

The total yield over time at various locations upstream from significant urban centers along the Mekong River includes key sites such as Chiang Saen in Thailand, as well as several locations in Laos, namely Luang Prabang, Vientiane, Paksane, Thakhek, Savannakhet, and Pakse, along with Stung Treng.

Milestones

Scenarios Under development (using similar protocols of transition matrices as Table 1)

Date expected

See status/target of activities above.

See section 3.3 Data availability and requirements.

References

DHSVM references See: http://www.hydro.washington.edu/Lettenmaier/Models/DHSVM/publications.htm

ICRAF SOUTHEAST ASIA PROJECT ACTIVITIES

GenRiver and SpatRain applications to Mae Chaem and Sumber Jaya

Lead

Collaborator

Scope, dataframe, spatial resolution (complete metadata: sources, definitions, dates, resolution, etc)

The Mae Chaem and Sumberjaya watersheds in Southeast Asia will be compared The two ASB benchmark areas included, with annual rainfall of about 1.5 and 2.5 m year-

The total water yield in subhumid and humid tropics, after accounting for an evapotranspiration estimate of 1.3 m year-1, is approximately 0.2 m year-1 in Mae Chaem and 1.2 m year-1 in Sumber Jaya (Way Besai) In Mae Chaem, the water balance is significantly influenced by the difference between actual and potential evapotranspiration Conversely, in Sumber Jaya, understanding the changes in soil structure that affect the partitioning of total water yield into quick and slow flows is crucial Despite similar river discharge, the watersheds exhibit distinct differences in population density and land use history, particularly concerning deforestation Comprehensive historical rainfall and river flow records are available for both regions, alongside extensive studies on land use changes related to the Alternatives to Slash and Burn programme in Thailand and Indonesia.

Figure 7 Rationale for Comparison of Mae Chaem and Sumberjaya watersheds.

Domain Mae Chaem see Thomas et al.2003

Sumberjaya see Verbist et al.2003 land use cover

Land use change (1964 - 2002) see Thomas et al 2003 Land use change (1973, 1983,

DEM 10 m resolution, see Thomas et al

5 m resolution, see Verbist et al 2003

Stream network derived from DEM, precision tests ongoing derived from DEM, precision tests ongoing soils using pedotransfer, see Suprayogo et al 2003 using pedotransfer, see

Suprayogo et al 2003 streamflow data Daily debit data from 1970 –

2000 Daily debit data Simpang Sari from 1975 – 1998

Dams –none within catchment none within catchment

Key climate variables include rainfall, rainfall intensity, spatial correlation of rainfall, and potential evaporation A 30-year rainfall record is available for only one station, while two other stations show similar average rainfall However, intra-station correlation of daily rainfall is low, as noted by Manik et al The station-level data served as input for SpatRain, which generated spatially explicit daily rainfall patterns that align with station-level records in terms of point-level exceedance probabilities, though they differ significantly across locations.

River flow patterns can vary significantly, categorized as 'patchy', 'intermediate', or 'homogeneous', depending on various factors Key variables influencing these patterns include rainfall and fundamental parameters for calculating Penman potential evapotranspiration, whether sourced from real or simulated data, as noted by Manik et al (2003) The analysis also considers spatiotemporal resolution, referencing both original and interpolated data, as discussed by Suyamto et al (2003), to evaluate time series trends.

Climatological and hydrological Data Set for Mae Chaem Daily debit data from 1970 – 2000

Climatological and hydrological Data Set for Sumberjaya

1 Daily rainfall for three stations from 1975 – 1998 (Sumberjaya, Fajar Bulan and Air Hitam)

2 Daily rainfall for PLTA base camp station from 1996 – 1999 (half year data for 1996 & 1999)

3 Daily relative humidity for PLTA base camp station from 1996 –

4 Daily wind speed for PLTA base camp station from 1996 – 1999 (half year data for 1996 & 1999)

5 Daily evaporation for PLTA base camp station from 1996 – 1999 (half year data for 1996 & 1999)

6 Daily rainfall from simple rain gauge installed for erosion plot measurement in Bodong area (May – Dec 2001)

7 Rainfall intensity (1997-1999) from Unila climate station in Sumberjaya

Many existing models focus on a single scale, such as individual plots or entire catchments, while others employ a grid-cell method that allows for interactions between cells, resulting in emergent behavior at the catchment level Additionally, a third category of models directly addresses cross-scale issues by specifying how properties vary with different temporal and spatial scales.

The GenRiver and SpatRain models were developed to investigate how the spatial variability of rainfall affects river flow uniformity, particularly in forested areas To achieve this, we require a representation of rainfall that balances random and fully correlated patterns Additionally, it is crucial to integrate this with a model that incorporates the concept of forests acting as a sponge, allowing for a comparison of the significance of both rainfall variability and forest influence on river flow.

GenRiver were developed for such a purpose.

The GenRiver model was made for data-scarce situations and is therefore based on

"First principles serve as a reliable foundation for various applications, recognizing that empirical models may offer higher precision within specific tested ranges This approach also strives to establish connections across different spatial scales."

The GenRiver model, originally designed to analyze river flow in the Way Besai watershed of Sumberjaya, Lampung, Indonesia, utilizes default input parameters specific to this region To adapt the GenRiver application for different watersheds, it is essential to gather comprehensive data on climate, landform, soils, geology, vegetation, land cover, and actual river discharge.

SpatRain utilizes macros to analyze semivariance based on the increasing distance between observation points, effectively characterizing rainfall patterns accumulated over various timeframes, including daily, weekly, monthly, and yearly intervals.

For comprehensive information, refer to the GenRiver background description available at the World Agroforestry Centre's website or download the zip file from the ASB CGIAR site.

Total yield, high flows, low flows to be determined

Land cover scenarios (Sumber Jaya)

Historical 1980 - 2000 land use change loss of high biodiverse areas Remove ridge top forest extensification All shade coffee intensification All sun coffee

Process

Comparison with WaNuLCAS and series of run-off plots for adjusting parameters describing change in soil structure with age of the garden; see ModSim GenRiver paper.

Validation was mainly based on frequency distribution of flows, see ModSim

GenRiver paper details on inputs and processing (e.g run by bootstrapping or off real data) - see GenRiver manual.

Sensitivity in small watersheds is addressed through single-parameter sensitivity analyses outlined in the manual We have tackled issues related to same-day transfers in small catchments and resolved various challenges within the SpatRain program, enabling us to effectively handle a wide range of rainfall input data Additionally, the GenRiver manual provides further insights into analyzing how watershed functions respond to land use changes, scale, and spatial patterns of rainfall (ICRAF-SEA).

The Genriver/SpatRain analysis illustrates the effects of land use changes on river flow buffering in relation to rainfall patterns, specifically calibrated for the Sumberjaya region in Lampung.

The preliminary results from Indonesia indicate various comparisons that can be drawn regarding the impacts of land use change, evidenced by the differences in height among the three bars within each group Additionally, the analysis highlights the effects of spatial scale, reflected in the height variations between the two groups of three bars Furthermore, it underscores the interaction between scale and land use change, which can be observed through the contrasts between the two comparisons.

All Forest Mixed LU All Grass All Forest Mixed LU All Grass

Large area or Patchy rainfal l Small area or Homogenous rainfall effect of land use change effect of land use change

By September 2003, we will have made graphs like this for the following situations:

- Mae Chaem (Thailand), calibrated to local rainfall and riverflow data, with the existing land use mix in between the two extremes

- Various land use patterns for Mae Chaem derived from the scenario study

Reporting and analysis of model runs (reporting of direct hydrological flows)

Grid cell or stream flow see GenRiver manual

Excedance probability graphs – for what, at what locations? see GenRiver manual

Matryushka diagrams? – of what, for what sub-basins?

Comparing ‘buffering capacity’ across space (plot, subcatchment, catchment) and time (day, week, month) for different rainfall patterns that are all consistent with the existing station records

Milestones

Date expected

0 in July; 1 st draft September 2003; 2 nd draft 8 October 2003; final 1 December 2003.

References

• Braak, C., 1929 The Climate of the Netherlands Indies Koninklijk

Magnetisch en Meteorologisch Observatorium te Batavia, Verhandelingen No.

• 8.Calder, I.R., 2002 Forests and hydrological services: reconciling public and science perceptions Land Use and Water Resources Research 2, 2.1-2.12 (www.luwrr.com)

• Coster, C 1938 Naschrift: herbebossching op Java (Postscript: reforestation on Java) - Tectona 32: 602-605.

• De Haan, J H., 1936 Overwegingen in verband met boschreserveering

(Considerations concerning forest reservation) Het Bosch 4: 1-28.

• Gordon, N D., T A McMahon, et al (1992) Stream Hydrology: An

Introduction for Ecologists Chichester, New York, Brisbane, Toronto,

• Grove, R.H., 1995 Green Imperialism: Colonial Expansion, Tropical Island Edens and the Origins of Environmentalism, 1600-1860 Cambridge

University Press, Cambridge (UK), 540 pp

• Heringa, P.K (1939) De Boschspons Theory? (The Forest Sponge Theory?)

Joshi et al (2003) emphasize the importance of integrating local ecological knowledge with modeling techniques to effectively understand soil and water movement Their work highlights the necessity of scaling up from plot-level observations to broader landscape contexts This approach is crucial for enhancing the management of belowground interactions in tropical agroecosystems, as discussed in the compilation edited by van Noordwijk, Cadisch, and Ong.

• Kaimuddin, 2000 Dampak perubahan iklim dan tataguna lahan terhadap keseimbangan air wilayah Sulawesi Selatan PhD thesis, Program

Pascasarjana, Institut Pertanian Bogor Kartasubrata (1981) Pre-war concepts concerning land use in Java in particular related to forest conservation

Presented at symposium on forest land use planning, Gajah Mada university, Jogyajarta Reprinted in: Kartasubrata, J (ed.) 2003 Social Forestry and Agroforestry in Asia, Book 2.

• Faculty of Forestry, Bogor Asgicultural Unviversity, Bogor, Indonesia Pp 3 –

• 11.Kiersch, B and Tognetti, S., 2002 Land-water linkages in rural watersheds Land Use and Water Resources Research 2, 1.1-1.6(www.luwrr.com)

• Ranieri S., Stirzaker, R., Suprayogo, D., Purwanto, E., de Willigen, P and van Noordwijk, M 2003.

• Managing movements of water, solutes and soil: from plot to landscape scale In: M van Noordwijk, G Cadisch and C.K Ong (Eds.) Belowground

Interactions in Tropical Agrocecosystems CAB International, Wallingford,

• Roessel, B.W.P (1939) Herbebossching op Java (Reforestation on Java) –

Van Noordwijk et al (1998) explore the complexities of erosion and sedimentation as multiscale, fractal processes, highlighting their significance for modeling, experimental design, and real-world applications Their work, featured in "Soil Erosion at Multiple Scales," edited by Penning de Vries, Agus, and Kerr, emphasizes the need for a deeper understanding of these processes to improve soil conservation strategies and environmental management.

Assessing Causes and Impacts CAB International, Wallingford pp 223-253

• Van Noordwijk, M, Farida, A., Suyamto, D., Lusiana, B and Khasanah, N.,

2003 Spatial variability of rainfall governs river flow and reduces effects of land use change at landscape scale: GenRiver and

• SpatRain simulations MODSIM proceedings, Townsville (Australia) July 2003.

• Wulandari, R (2002) Deteksi perubahan penutupan lahan pada areal sempadan sungai di Sumberjaya, Lampung Barat Jurusan Konservasi

Sumberdaya hutan, Fakultas Kehutanan Bogor, Institut Pertanian Bogor and ICRAF-SEA, Bogor, Indonesia: 58.

3.7 FALLOW model (F-orests, A-groforests, L-ow-value, L-andscapes, O-r W- astelands) applied to Sumberjaya and Parameterization in Mae Chaem

Collaborator

Contacts with PCRaster Developer – Utrecht University, the Netherlands for providing the main engine for the model implementation.

University of Washington (Jeffrey Richey) will provide MSEA maps for Land Use distribution at meso-scale (8-50 km applicability) for 1970, 1980, 2000.

Scope, dataframe, spatial resolution(complete metadata: sources, definitions, dates, resolution, etc)

The rural landscape at the edge of the forest in Sumberjaya, Indonesia, is particularly vulnerable to changes in land use and land cover Ongoing research and parameterization efforts are also being conducted in Mae Chaem, Thailand, to further understand these dynamics.

The article discusses various land types, including settlements, agricultural lands categorized by crop type and land management practices (wetland or upland), forests classified by their successional stages, agroforests differentiated by their productivity stages, and monoculture plantations also distinguished by productivity levels The resolution for this analysis is set at 100x100 m², utilizing GIS products for model initialization.

DEM Resolution: 100x100 m 2 Generated from topographical maps stream network Resolution: 100x100 m 2 Generated from topographical maps soils

Qualitative data of soil fertility and soil physical properties (mean and standard deviation) to generate spatial distribution of soil properties randomly as initialization streamflow data

GenRiver outputs: annual constant baseflow (mm) and annual baseflow fraction Maximum groundwater storage (mm) Resolution: plot scale average (1 plot 100x100 m 2 ) dams Not yet considered

Core modules: (1) weather variability affecting crop productivity Watershed function toolbox: (2) annual rainfall data: amount (mm), (3) coefficient of variance,

(4) minimum daily rainfall (mm) and (5) maximum daily rainfall (mm) sources (real or simulated?)

(1) simulated (sensitivity analyses result); (2)-(5) from historical records spatiotemporal resolution, original and interpolated

Temporal scale: annual; spatial scale: plot (100x100 m 2 ), generated from empirical statistic time series Applicable (see GenRiver climatological time series), but not yet applied

FALLOW is a spatially explicit model that focuses on landscape dynamics, highlighting farmers as the primary human agents influencing land use and cover changes It offers toolboxes for evaluating the effects of these changes on food security, watershed functions, biodiversity, and carbon stocks Operating on a yearly time scale and a landscape spatial scale with a plot resolution of 100x100 m², FALLOW incorporates a dynamic single-agent human dimension.

Conceptualization phase: STELLA 5.1.1 – Research Edition

User interface: Microsoft Excel spreadsheet: available in http://www.worldagroforestrycentre.org/sea/Products/AFModels/FALLOW/Fallow.ht m

Before August 2003: Microsoft Excel spreadsheet: available in http://www.worldagroforestrycentre.org/sea/Products/AFModels/FALLOW/Fallow.ht m

From August 2003: stand-alone package developed using Microsoft Visual Basic 6.0

– Professional Edition (not yet available in the website, but can be freely ordered from d.suyamto@cgiar.org )

The core modules focus on several key areas: first, the dynamics of soil fertility and land productivity at the plot level; second, the influence of market mechanisms on household economies; third, strategic decision-making regarding land use and labor allocation, facilitated by learning; fourth, the practical implementation of land use and labor allocation, including site selection; and finally, the changes in land use and cover driven by ecological succession and growth.

Additional modules: (6) human population dynamics; (7) market; (8) fire escape Consequences toolboxes: (9) food security; (10) watershed functions: water balance

(water yield, baseflow), sediment loss, soil physical quality dynamics; (11)

52 biodiversity: plot level; landscape level (through scaling rules); (12) carbon stocks: aboveground c-stocks and belowground c-stocks.

The transition from preindustrial shifting cultivation to contemporary practices has led to the widespread adoption of monoculture plantations, particularly for coffee and clove, resulting in the loss of high-biodiversity areas In overpopulated conditions, shifting cultivation has intensified, prompting the extensification of coffee agroforestry as a response This shift highlights the challenges of balancing agricultural intensification with the preservation of biodiversity in modern farming practices.

Process

Parameterization for: (1) socio-economical data – profitability assessment by

Budidarsono et al (2000) conducted a preliminary study on landscape dynamics in Sumberjaya, while Leimona (2001) contributed to the understanding of the area The spatial data was analyzed by the spatial data analyses unit at ICRAF SEA Additionally, stream flow parameters were derived from GenRiver outputs, and other hydrological parameters were gathered through a literature review, including insights from Verbist et al (2002) and discussions with experts in the field.

(5) biodiversity parameters: discussion with experts and outputs from other projects;

(6) soil fertility dynamics: literature review (Trenbath, 1989 and van Noordwijk, 2002); (7) carbon stocks parameters: literature review (e.g van Noordwijk et al, 2002); (8) products prices are generated based on possible range of historical records;

(9) any others are discussion results with experts.

A literature review, including the work of Milliman et al (1999) as referenced in Suyamto et al (2003), was conducted to validate the range of sediment loss associated with various land use change scenarios For detailed insights, please refer to the full study available at http://www.asb.cgiar/BNPP/phase2/sea/suyamto_etal_2003_608-613.pdf.

3.7.4.3 Details on inputs and processing (e.g run by bootstrapping or off real data)

All spatial data were represented as raster data (with 100x100 m 2 resolution) Non- spatial data were compiled from literature reviews/other projects’ outputs/discussion with experts/simulated data.

Sensitivity analyses were conducted focusing on two primary scenarios: first, to evaluate the effectiveness of forest reserve allocation, and second, to assess farmers' responses to coffee price shocks regarding their land use decisions These decisions significantly influence land use and cover changes within the landscape, ultimately impacting watershed functions such as sediment loss For comprehensive results, refer to Suyamto et al (2003) at http://www.asb.cgiar.org/BNPP/phase2/sea/suyamto_etal_2003_608-613.pdf.

Reporting and analysis of model runs ( Reporting of direct hydrological flows)

Reporting of direct hydrological flows – See Suyamto, et al., 2003 http://www.asb.cgiar.org/BNPP/phase2/sea/suyamto_etal_2003_608-613.pdf

Milestones

Date expected

0 in July; 1 st draft on September 2003; 2 nd draft on 8 October 2003; final 1 December 2003.

References

• Budidarsono, S., Kuncoro, S.A., and Tomich, T.P A Profitability Assessment of Robusta Coffee Systems in Sumberjaya Watershed, Lampung, Sumatra, Indonesia. Southeast Asia Policy Research Working Paper, No 16 ICRAF SEA 2000

• Leimona, B., Modelling land use change and its driving factors: a preliminary dynamic landscape-based model of Sumberjaya Watershed, Master Thesis, Bogor Agricultural University, Bogor, 2001

• Milliman, J.D., Farnsworth, K.L., and Albertin, C.S., Flux and fate of fluvial sediments leaving large islands in the East Indies, Journal of Sea Research 41,

The FALLOW model, developed by Suyamto et al (2003), serves as an assessment tool to evaluate the landscape-level impacts of farmers' land use decisions This model is discussed in the proceedings of the MODSIM 2003 conference, which focused on the international modeling of biophysical, social, and economic systems for effective resource management solutions The full paper can be accessed through the provided links for further insights into its applications and methodologies.

• Trenbath, B.R., The use of mathematical models in the development of shifting cultivation In: Proctor, J (Ed.), Mineral Nutrients in Tropical Forest and Savanna Ecosystems Blackwell, Oxford, pp 353-369, 1989

• van Noordwijk, M., Scaling trade-offs between crop productivity, carbon stocks and biodiversity in shifting cultivation landscape mosaics: the FALLOW model,

In their 2002 study published in Science in China, van Noordwijk et al conducted a comprehensive carbon stock assessment in the Sumber-Jaya region of Lampung, Indonesia, focusing on landscapes transformed from forest to coffee cultivation The research utilized allometric equations to evaluate carbon stocks and analyzed the implications of land use changes on carbon dynamics, providing valuable insights into sustainable agricultural practices and environmental impact in tropical regions.

• Verbist, B.J.P., van Noordwijk, M., Tameling, A C., Schmitz, K C L and

Ranieri, S B.L developed a negotiation support tool to evaluate the impacts of land use changes on erosion within a previously forested watershed in Lampung, Sumatra, Indonesia This research was presented in the proceedings of the 1st Biennial Meeting of the International Environmental Modelling and Software Society (IEMSS) in 2002, edited by Rizzoli, A.E and Jakeman, A.J.

UNIVERSITY OF NEW HAMPSHIRE PROJECT ACTIVITIES

Central America meso-model

Amendment to Activity 2 Implementation Protocols based on discussions during BNPP team meeting in Prague, Czech Republic, 11-12 October 2003:

The original thinking regarding development of a meso-scale model for

During the Prague meeting, it was concluded that the Water Balance Model (WBM) developed by the UHN team is unsuitable for analyzing basins smaller than 30,000 km² (12 pixels) Since all Central American basins fall below this minimum size requirement, it was determined that utilizing the WBM for these regions would not be feasible.

“zoom in” on Central America at higher resolution

In October, discussions among Ken Chomitz, Charles Vorosmarty, and Tom Tomich led to the decision for UNH to concentrate on pan-tropical hydrological analysis, integrating findings with UofW and ICRAF SE Asia results, while postponing finer resolution work in Central America The UNH team is committed to developing higher resolution global data sets (8km) and aims to enhance resolution in Central America beyond the BNPP Phase II scope Additionally, a dialogue has been initiated with UNESCO's Office for Science and Technology for Latin America to explore future collaboration.

In Montevideo, efforts are underway to enhance existing hydrometeorological data sets by integrating them with high-resolution topography and river networks This initiative may serve as a foundation for future developments, potentially leading to the creation of a more detailed meso-model for Central America.

4 List of scientific papers to be prepared based on Activity 2

Paper 1 Spatrain: A Simulator of Space/Time Patterns in Rainfall for Predicting Scale Dependence of Variability of Rainfall-related Processes

The spatial variability of rainfall significantly influences soil-level buffering, affecting the consistency of river flow and reducing the responsiveness of watershed functions to land use changes Additionally, the buffering capacity of river flow in tropical watersheds plays a crucial role in maintaining ecological balance and water management.

Paper 4 Bridging scales and knowledge domains of watershed functions

Paper 5 Plausible scenarios for changes in forest cover and spatial distribution in northern Thailand

Paper 6 Dynamic landscape model for predicting impacts of land use change on biodiversity and watershed functions

Paper 8 Mekong Time Series analysis

Paper 9 Scale and location in determining near-field vs far-field effects of landcover change in the Mekong

Paper 10 Landuse change effects in the Mae Chaem

Paper 1 Spatrain: A Simulator of Space/Time Patterns in Rainfall for

Predicting Scale Dependence of Variability of Rainfall-related Processes

This paper will be submitted before 15 December 2003.

Spatrain: A Simulator of Space/Time Patterns in Rainfall for Predicting Scale Dependence of Variability of Rainfall-related Processes

Base The development of a spatially explicit rainfall simulator for use with spatially explicit hydrological models such as GenRiver

Desi Suyamto, Meine van Noordwijk, Betha Lusiana, Ai Farida and Rita Manik; World Agroforestry Centre, ICRAF-Southeast Asia, P.O Box 161, Bogor 16001, Indonesia

Target journal(s) Agriculture and Water Management (or similar)

Variations in river flow diminish with larger areas due to reduced temporal correlation of rainfall events Rainfall patchiness can enhance yield stability across regions The SpatRain model generates time-series rainfall data compatible with existing daily records while allowing for varying degrees of spatial autocorrelation It starts with the spatial characteristics of a single rainstorm pathway, modeling rainfall intensity that decreases with distance from the core area The model calculates daily rainfall amounts for a grid of observation points, accommodating multiple storm events without surpassing long-term maximum station-level rainfall It also offers options to factor in elevation effects on rainfall SpatRain is available for free on our website as an Excel workbook, featuring macros that analyze semivariance based on distances between observation points to characterize rainfall patterns over designated time periods.

Paper 2 Spatial variability of rainfall interacts with soil-level buffering in its

responsive to land use change

This paper will be submitted before 15 December 2003.

Spatial variability of rainfall interacts with soil-level buffering in its impact on regularity of river flow, making watershed functions less responsive to land use change

This study tests the hypothesis that spatial variability in rainfall can substitute for physical buffering in influencing the regularity of river flow It suggests that as the area considered increases, the impact of land use on watershed functions becomes less significant and more predictable.

The study utilizes the GenRiver and SpatRain models to explore various virtual watersheds, aiming to predict how land use changes—both historical and projected—affect total water flow and its regularity This analysis is grounded in specific scenarios that compare past conditions with present and plausible future developments.

Authors Meine van Noordwijk, Farida, Desi Suyamto, Betha Lusiana

Target journal(s) technical/hydrological target?

Empirical evidence indicates that land use change significantly impacts streamflow characteristics, including total quantity, sediment load, and fluctuation amplitude, particularly within spatial scales up to 100 km² However, there is a notable lack of evidence for effects beyond this scale, raising questions about whether this is due to insufficient research or other influencing factors, such as the low spatial correlation of rainfall events and varying routing times Additionally, it is essential to consider whether local buffering changes associated with land use modifications are overshadowed by these larger-scale effects The GenRiver model, a distributed water balance model at the subcatchment level, utilizes the SpatRain rainfall generator to create diverse rainfall patterns This allows for a comparison of the effects of routing time differentiation and rainfall pattern length scale against land use-related changes in interception, infiltration capacity, and storage Parameterization of the Sumber Jaya area in Lampung, Indonesia, reveals that using a highly disaggregated rainfall pattern produces daily river flow patterns closely resembling observed frequency distributions, unlike more homogeneous rainfall inputs.

‘patchy’ rainfall may induce more surface quickflow at field scale, it tends to a more regular pattern of riverflow at landscape scale

At the landscape scale, particularly for 5th order rivers, the primary factors influencing river flow patterns differ significantly from those at the field scale This suggests that the impact of land use changes, such as forest conversion, is likely to diminish as the scale of analysis increases.

GenRiver, Hydrological impacts, Land use change, Rainfall variability, SpatRain, Scale, WaNuLCAS, water-balance

The 6-page ModSim paper by van Noordwijk et al serves as the foundational reference for this study, with the complete paper included as an appendix in the Phase 2 report For further details, please refer to the document available at www.asb.cgiar.org/BNPP/phase2/sea/vannoordwijk_etal_572-577.pdf.

Paper 3 Buffering of riverflow in tropical watersheds

Title Buffering of riverflow in tropical watersheds

Authors Charles Vorosmarthy, Ellen Douglas, Meine van Noordwijk, Farida

Empirical evidence indicates that land use changes significantly affect streamflow characteristics, including total quantity, sediment load, and fluctuation amplitude, particularly within spatial scales of up to 100 km² However, there is limited data on these effects beyond this scale, raising questions about whether this is due to insufficient research or other influencing factors, such as the low spatial correlation of rainfall events and varying routing times Additionally, it remains unclear if local buffering changes associated with land use are overshadowed by these other effects The GenRiver model, a distributed water balance model at the subcatchment level, utilizes the SpatRain rainfall generator to create diverse rainfall patterns for analysis This model allows for the comparison of the impacts of routing time differentiation and rainfall pattern length scale against the effects of land use changes on interception, infiltration capacity, and storage In the Sumber Jaya area of Lampung, Indonesia, model parameterization can produce daily river flow patterns that closely resemble observed frequency distributions when a highly disaggregated rainfall pattern is applied, in contrast to more uniform rainfall inputs.

‘patchy’ rainfall may induce more surface quickflow at field scale, it tends to a more regular pattern of riverflow at landscape scale

At the landscape scale, particularly in 5th order rivers, the factors influencing river flow patterns differ significantly from those at the field scale As the scale of analysis increases, the impact of land use changes, such as forest conversion, is expected to diminish in importance.

GenRiver, Hydrological impacts, Land use change, Rainfall variability, SpatRain, Scale, WaNuLCAS, water-balance

Paper 4 Bridging Scales and Knowledge Domains of Watershed Functions

The full paper will be attached to Technical report and is currently available.

Title Bridging scales and knowledge domains of watershed functions

Base Base 1 analysis of the impacts of land use change on quantity and temporal distribution of river flow for Mae Chaem and Sumberjaya as test sites for the models

Authors Meine van Noordwijk, Laxman Joshi, Desi Suyamto, Ai Farida and

Target journal(s) In discussion with organizers of CPWF Baseline Workshop Nairobi

Bridging the gap between local, scientific, and public/policy ecological knowledge in natural resource management is challenging, particularly regarding how watershed functions change with scale Empirical evidence indicates that area-based scaling is only suitable for total water yield, while traditional plot-level and paired catchment research fails to adequately assess water use efficiency and watershed functions in real landscapes Recognizing and quantifying lateral flows that influence biophysical scaling is crucial for progress This presentation will explore the use of models as tools for scaling across different spatial and temporal contexts, highlighting the need to connect simple empirical models with complex, process-based ones that require extensive data Understanding the intricacies of biophysical scaling is vital for effective institutional mechanisms and multi-stakeholder negotiations.

Paper 5 Plausible scenarios for changes in forest cover and spatial

Title Plausible scenarios for changes in forest cover and spatial distribution in northern Thailand

Authors Louis Lebel, David Thomas

The project aims to develop four contrasting socio-economic development scenarios that will outline assumptions regarding land-use changes, particularly focusing on factors influencing hydrology This exercise seeks to create a logical framework for exploring the hydrological consequences of different land-use patterns across various scales By generating plausible future landscapes, these scenarios also serve as a context for interpreting modeled outcomes The primary value lies in comparing the scenarios rather than assessing the probability of any single scenario Each scenario is designed to provide insights into potential future developments.

1 Story line describing in broad terms how development proceeds at the regional scale;

2 A set of explicit rules (a qualitative model) for evolving landscapes under each of the scenarios

The project aims to develop land-use and land-cover scenario maps for the Ping River Basin and Mae Chaem in the Mekong River Basin, covering intervals from 2000 to 2050 These maps will be created under a separate sub-project grant from SEA-START RC at Chulalongkorn University, which is part of the AIACC initiative focused on climate change and water-related sector impacts The funding requested will primarily support efforts at the other two scales of analysis.

Paper 6 Dynamic landscape model for predicting impacts of land use change

on biodiversity and watershed functions

The Phase 2 report will include a 6-page ModSim paper by Suyamto et al as an appendix, with the new paper scheduled for submission after December 15.

Title Dynamic landscape model for predicting impacts of land use change on biodiversity and watershed functions

Base FALLOW simulations for Sumberjaya and Mae Chaem

Authors Desi A Suyamto, Meine van Noordwijk, Danan Prasetyo Hadi and

FALLOW is a landscape-dynamics model that focuses on key annual processes affecting agricultural productivity, including soil fertility dynamics during crop and fallow phases, food storage and sales at the village level, and farmer decision-making regarding land use based on labor availability and profitability The model has transitioned from a Stella model to a spatially explicit environment using PCRaster, allowing for applications in larger landscapes with real spatial data FALLOW aids in impact assessments and scenario studies, facilitating stakeholder negotiations by visualizing consequences of factors like price changes, population density, and climate variations An application in Sumberjaya, Lampung, Sumatra, predicts the impacts on watershed functions from spatially allocating forest reserves and land use changes in response to coffee price shocks Findings indicate that maintaining riparian forests at 25% cover results in the lowest sediment load in rivers compared to other allocations, while farmer responses to price shocks vary based on learning styles, significantly affecting predicted sediment loss Keywords: landscape-dynamics model, annual time step, spatially explicit, PCRaster, Sumatra.

Paper 7 Training material

The full paper will be attached as appendix to the Phase 2 report

Title Agroforestry and watershed functions of tropical land use mosaics

Base BNPP Phase I report plus ongoing work of Phase 2.

Authors Meine van Noordwijk, Ai Farida, Bruno Verbist, Tom Tomich

Target journal(s) Proceedings of 2 nd Asia EcoHydrology Workshop/Training Course, July

2003, Indonesia See: http://www.asb.cgiar.org/BNPP/phase2/sea/ecohydrologytext_mvn_26jul2003.doc

An ecohydrology approach emphasizes that the quantity, timing, and quality of water flows are influenced by land cover and use across the entire landscape, rather than solely by forest cover in upper watersheds This article explores varying perceptions of the relationship between forests and watershed functions, highlighting the specific functions relevant to different stakeholders Land use changes can significantly affect evapotranspiration, total water yield, and the pathways water takes, impacting the transport of soil particles, nutrients, agrochemicals, and salts, particularly in arid regions Additionally, infiltration rates influence river flow evenness at smaller scales, while the partial spatial independence of rainfall becomes a key factor in flow evenness over larger areas, diminishing the relative impact of land use.

Agroforestry practices in riparian zones significantly enhance water quality and regulate downstream flow, potentially surpassing the benefits of forest cover in upper watersheds.

Paper 8 Mekong Time Series analysis

Title Mekong Time Series analysis

Authors Jeffrey Richey and others

Target journal(s) Global Change Biology (or equivalent)

Subtle variations in the Mekong's flow regime can lead to significant cumulative downstream impacts, known as "far-field effects," which affect biodiversity, water management, and flooding This paper assesses the influences of climate variability and land use changes over the past two decades, while also exploring historical and potential future scenarios.

Paper 9 Scale and location in determining near-field vs far-field effects of

Title Scale and location in determining near-field vs far-field effects of landcover change in the Mekong

Authors Jeffrey Richey and others.

Target journal(s) Global Biogeochemical Cycles or a hydrology journal

This paper would be the primary effort in looking at the

“scenarios” linking scale and location in a drainage basin with effects.

Tiêu đề	Implementation Protocol for Activity 2: Micro/Meso Modeling of Hydrological Effects and Cross-Scale Effects
Tác giả	Meine Van Noordwijk, Jeffrey Richey, Desi Suyamto
Người hướng dẫn	Sandra Velarde, Thomas Tomich, Dagmar Timmer
Trường học	University of Washington
Thể loại	implementation protocol
Năm xuất bản	2003
Thành phố	Washington

Định dạng
Số trang	68
Dung lượng	0,91 MB