1Background:¶
The BoRs collect agricultural tax in Pakistan based upon agricultural land categorization, land holdership and production, and land status captured in the Record of Rights by the Patwaris . The BoRs also collect land and property transaction taxes and fees including withholding tax, mutation fees, stamp duties, registration fees, and capital value tax, which all relate to the registered property transaction prices and the valuations available with the BoRs. As the overall process is heavily relying on manual methods, the system is hampered by limited human resources, leading to significant delays and inaccuracies in land record change and valuation practices. Land type and value data is not up to date, and this impacts several streams of land-related revenues for the federal, provincial, and local governments. Agricultural tax revenue yields are comparatively low relative to the cost of collection, raising concerns of its long-term viability.
Focusing on Khyber Pakhtunkhwa (KP) province, only half of the territory has official land records, and these records, registers, and maps are not interlinked. Existing records, registers, and maps are often not interlinked, leading to frequent disputes and challenges in validating property rights, leading The Government of Khyber Pakhtunkhwa (GoKP) misses significant revenue flows annually due to lapses in land records and valuation. To tackle this situation, the GoKP plans to modernize and strengthen the revenue records, mapping, and property valuation systems, utilizing geospatial infrastructures and its technologies, directly lead to increased collection of the above taxes/fees collected by BoR.
In terms of geospatial infrastructure, the GoKP possesses good quality large-scale mapping, but its agencies and local governments lack access to up-to-date and granular geospatial data for efficient urban development and resilience. Survey of Pakistan is the country’s national mapping and surveying organization with a mandate to provide official, topographic, and thematic mapping in the country and regulate geospatial information. Yet, it is focused on large scale topographic mapping. Provider and arrangements for detailed mapping are unclear and in practice urban development and resilience rely on ad hoc mapping financed with donor support.
In this context, the World Bank (WB) is carrying out a technical assistance (TA) on land administration modernization with the aim to improve the Government of Khyber Pakhtunkhwa’s own revenue generation. The TA encompasses three distinct components:
Revenue, which entails a comprehensive review of land information, valuation, and taxation systems;
Public Asset Management, focusing on the efficient management of public land and building assets; and
Land and Geospatial, including a thorough examination of urban and rural land administration as well as the geospatial information framework. This assignment falls under the component
2Objectives:¶
The primary objective of this assignment is to develop and pilot Artificial Intelligence and Machine Learning (AI-ML) based geospatial analysis models in selected agricultural areas in Khyber Pakhtunkhwa (KP). The aim is to demonstrate the models’ effectiveness and potential contribution to enhancing the efficiency and accuracy of KP’s agriculture-related taxation practices. Each pilot model will be carefully developed to align with KP’s agricultural taxation framework and leverage the latest applicable technologies. Furthermore, the assignment will propose recommendations to improve precision and outline strategies for scaling the models for broader implementation.
3GeoAI Models Analysis:¶
Based on the feedback from Task 1, the Consultant will undertake activities to develop three (3) AI-ML geospatial analysis models in the selected area. This involves collaborating with Beneficiary’s relevant authorities to identify a representative pilot area with a size of 1-2 mouzas , for demonstration purposes. Three potential models will be as below, subject to change based on the Beneficiary’s needs and available datasets:
Crop type detection model: This model will identify and classify dominant crop types in each parcel in the pilot area with high accuracy utilizing the satellite imagery. This model will provide spatial distribution data of various crops, enabling better estimation of agricultural tax based on crop type. The outcomes will also aid in monitoring compliance with BoR’s crop-related records.
Crop yield projection model: This model is expected estimate the potential yield of identified crops by analyzing historical data, weather patterns, and vegetation indices from satellite imagery. The results will include quantifiable yield predictions that can improve revenue forecasting for agricultural taxation. This model will also assist in identifying areas with significant discrepancies in reported versus projected yields.
Land use change detection model: This model will analyze temporal satellite imagery to identify patterns of land use change within the pilot area, such as shifts between agricultural and non-agricultural uses. The expected results include detailed maps and datasets highlighting areas of change, enabling authorities to identifty unauthorized land conversions accordingly.